School Readiness

January 1, 2005

A PUBLICATION OF THE WOODROW WILSON SCHOOL OF PUBLIC AND INTERNATIONAL AFFAIRS AT
PRINCETON UNIVERSITY AND THE BROOKINGS INSTITUTION
3 Preface
5 Introducing the Issue
15 Assessment Issues in the Testing of Children at School Entry
35 Can Family Socioeconomic Resources Account for Racial and Ethnic
Test Score Gaps?
55 Genetic Differences and School Readiness
71 Neuroscience Perspectives on Disparities in School Readiness and
Cognitive Achievement
91 Low Birth Weight and School Readiness
117 Health Disparities and Gaps in School Readiness
139 The Contribution of Parenting to Ethnic and Racial Gaps
in School Readiness
169 Early Childhood Care and Education: Effects on Ethnic and Racial
Gaps in School Readiness
School Readiness:
Closing Racial
and Ethnic Gaps
The Future
of Children
P R I N C E T O N - B R O O K I N G S
VOLUME 15 NUMBER 1 SPRING 2005
The Future of Children seeks to translate high-level research into information that is useful to
policymakers, practitioners, and the media.
The Future of Children is a publication of the Woodrow Wilson School of Public and International
Affairs at Princeton University and the Brookings Institution.
The views expressed in this publication do not necessarily represent the views of the Woodrow
Wilson School at Princeton University or the Brookings Institution.
ISSN: 1550-1558
ISBN: 0-8157-5559-7
Senior Editorial Staff
Sara McLanahan
Editor-in-Chief
Princeton University
Director of the Center for Research on
Child Wellbeing and Professor of Sociology
and Public Affairs
Ron Haskins
Senior Editor
Brookings Institution
Senior Fellow, Economic Studies Program
Christina Paxson
Senior Editor
Princeton University
Director of the Center for Health and
Wellbeing and Professor of Economics and
Public Affairs
Cecilia Rouse
Senior Editor
Princeton University
Director of the Education Research Section
and Professor of Economics and Public
Affairs
Isabel Sawhill
Senior Editor
Brookings Institution
Vice President and Director, Economic
Studies Program
Journal Staff
Elisabeth Hirschhorn Donahue
Associate Editor
Princeton University
Brenda Szittya
Managing Editor
Brookings Institution
Margy Waller
Outreach Director
Brookings Institution
Lisa Markman
Senior Outreach Coordinator
Princeton University
Julie Clover
Outreach Coordinator
Brookings Institution
Maureen Marchetta
Outreach Coordinator
Princeton University
Anne Hardenbergh
Editorial Assistant
Brookings Institution
Stacey O?Brien
Webmaster
Princeton University
Board of Advisors
Jeanne Brooks-Gunn
Virginia and Leonard Marx Professor of
Child Development and Education,
Teachers College and College of Physicians
and Surgeons, Columbia University
Peter Budetti
Bartlett Foundation Professor and Chairman,
Department of Health Administration and
Policy, University of Oklahoma
Judith Feder
Dean of Public Policy, Georgetown
University
Sandra Feldman
President Emeritus, American Federation of
Teachers
William Galston
Interim Dean, School of Public Policy,
University of Maryland
Jean B. Grossman
Senior Vice President for Research,
Public/Private Ventures, and Lecturer in
Economics and Public Affairs,
Princeton University
Kay S. Hymowitz
Senior Fellow and Contributing Editor,
Manhattan Institute for Policy Research
Charles N. Kahn III
President, Federation of American Hospitals
Rebecca Maynard
University Trustee Professor of Education
and Social Policy, University of Pennsylvania
Marguerite W. Sallee
President and CEO
America?s Promise
Lynn Thoman
Managing Partner, Corporate Perspectives
Heather B. Weiss
Director, Harvard Family Research Project,
Harvard University
Amy Wilkins
Principal Partner, The Education Trust
The Future of Children would like to thank the David and Lucile Packard Foundation, the
Annie E. Casey Foundation, the Doris Duke Charitable Foundation, and the W.K. Kellogg
Foundation for their generous support.
The Future of Children is copyrighted by Princeton University and the Brookings Institution,
all rights reserved. Authorization to photocopy articles for personal use is authorized by The
Future of Children. Reprinting is also allowed, so long as the journal article is properly given attribution.
For volumes 14, no. 1, and before, please use this attribution: ?From The Future of
Children, a publication of the David and Lucile Packard Foundation.? For subsequent journals,
please use this attribution: ?From The Future of Children, a publication of The Woodrow
Wilson School of Public and International Affairs at Princeton University and the Brookings Institution.?
It is the current policy of the journal not to charge for reprinting, but this policy is
subject to change.
To purchase a subscription, access free electronic copies, or sign up for our e-newsletter, go to
our website, www.futureofchildren.org. If you would like additional information about the journal,
please send questions to foc@princeton.edu.
School Readiness: Closing
Racial and Ethnic Gaps
3 Preface
5 Introducing the Issue by Cecilia Rouse, Jeanne Brooks-Gunn, and
Sara McLanahan
15 Assessment Issues in the Testing of Children at School Entry
by Donald A. Rock and A. Jackson Stenner
35 Can Family Socioeconomic Resources Account for Racial and
Ethnic Test Score Gaps? by Greg J. Duncan and Katherine A.
Magnuson
55 Genetic Differences and School Readiness by William T. Dickens
71 Neuroscience Perspectives on Disparities in School Readiness
and Cognitive Achievement by Kimberly G. Noble, Nim Tottenham,
and B. J. Casey
91 Low Birth Weight and School Readiness by Nancy E. Reichman
117 Health Disparities and Gaps in School Readiness by Janet Currie
139 The Contribution of Parenting to Ethnic and Racial Gaps in School
Readiness by Jeanne Brooks-Gunn and Lisa B. Markman
169 Early Childhood Care and Education: Effects on Ethnic and
Racial Gaps in School Readiness by Katherine A. Magnuson and
Jane Waldfogel
VOLUME 15 NUMBER 1 SPRING 2005
The Future
of Children P R I N C E T O N - B R O O K I N G S
www. f u t u r e o f c h i l d r e n . o r g
Welcome to the new
Future of Children. In
addition to the new look,
the journal has a new
home?the Woodrow
Wilson School at Princeton University in partnership
with the Brookings Institution. We
are excited about this new venture and are
honored to have been given the reins of such
a respected journal by the Packard Foundation.
Although the journal has a new publisher, the
overall mission remains the same: translating
the best social science research into information
that is useful to policymakers, practitioners,
and the media. We have developed plans
to ensure that we get our information into the
hands (and heads) of as many people as possible:
shorter publications on our website and
public events and conferences aimed at policymakers
and practitioners. All the information
about our outreach efforts can be found
on our website, www.futureofchildren.org,
and we encourage you to visit and sign up for
the e-newsletter so that we can keep you apprised
of our efforts.
Probably the biggest change is that we have
become primarily a web-based journal. Although
some print copies of the journal will
continue to be produced, for the most part
readers will find us on the web. We made this
decision based on the reality of the high costs
of producing and mailing large numbers of
print volumes for free, coupled with the
growing access people have to the internet.
We do understand, however, that many readers
would like to keep receiving print copies
of the journal, so we now offer moderately
priced subscriptions as well as single-copy
sales.
We hope you like our changes and continue to
read and use the journal. We have received a
great gift in being able to produce The Future
of Children, and we will work hard to ensure
its quality, integrity, and accessibility.
Sara McLanahan
Editor-in-Chief
The Future of Children
VOL. 15 / NO. 1 / SPRING 2005 3
Preface
Introducing the Issue
Cecilia Rouse, Jeanne Brooks-Gunn, and Sara McLanahan
VOL. 15 / NO. 1 / SPRING 2005 5
www.future of children.org
Cecilia Rouse is Director of the Education Research Section and Professor of Economics and Public Affairs at Princeton University. Jeanne
Brooks-Gunn is Virginia and Leonard Marx Professor of Child Development and Education at Teachers College and the College of Physicians
and Surgeons, Columbia University. Sara McLanahan is Director of the Center for Research on Child Wellbeing and Professor of Sociology
and Public Affairs at Princeton University. The authors appreciate the assistance of Steve Barnett, William Galston, Amy Wilkins, Christine
Connelly, and Wendy Wilbur
Although racial and ethnic gaps
in educational achievement
have narrowed over the past
thirty years, test score disparities
among American students
remain significant. In the 2002 National Assessment
of Educational Progress, 16 percent
of black and 22 percent of Hispanic twelfthgrade
students displayed ?solid academic
performance? in reading, as against 42 percent
of their white classmates.1 Similar gaps
exist in mathematics, science, and writing. In
response to such findings, policymakers have
devised high-profile education initiatives to
help schools address these disparities. The
No Child Left Behind Act of 2002, for example,
explicitly aims at closing achievement
gaps. And such policies are important. As
Christopher Jencks and Meredith Phillips,
two highly regarded social scientists, conclude,
?reducing the black-white test score
gap would probably do more to promote
[racial equality] than any other strategy that
commands broad political support.?2
To date, policymakers and practitioners have
focused most attention on the gaps in
achievement among school-aged children.
And yet by many estimates sizable racial and
ethnic gaps already exist by the time children
enter kindergarten. Indeed, according to one
report, about half of the test score gap between
black and white high school students is
evident when children start school.3
Why is so much attention focused on schoolaged
children? One reason is the lack of data
on younger children. Many large and detailed
surveys include only older children, and
school-based administrative data necessarily
exclude preschoolers. A second reason is that
federal, state, and local policy focuses on
public education, which has traditionally
started with kindergarten. Finally, until recently
the lives of preschool children were
largely viewed as falling under the purview of
the family and outside the scope of public
policy.
Nevertheless, research findings and common
sense both suggest that what happens to children
early in life has a profound impact on
their later achievement. The behavioral and
academic skills that children bring with them
to school not only determine how schools
must spend resources but also potentially affect
disparities in outcomes. And some analysts
argue that attending to disparities in the
early years is likely to be cost effective. As
Nobel laureate James Heckman notes, evaluations
of social programs targeted at children
from disadvantaged families suggest that it is
easier to change cognition and behavior in
early childhood than in adolescence.4
This issue of The Future of Children shines
the spotlight on school readiness. In its broadest
sense, school readiness includes the readiness
of elementary school teachers and staff
as well as of children and parents. Yet although
schools must be ready for the children
who arrive at their doors, in this volume we
focus on the skills of the children themselves.
Why Gaps in School Readiness
Matter
Children who enter school not yet ready to
learn, whether because of academic or social
and emotional deficits, continue to have difficulties
later in life. For example, children
who score poorly on tests of cognitive skills
during their preschool years are likely to do
less well in elementary and high school than
their higher-performing preschool peers and
are more likely to become teen parents, engage
in criminal activities, and suffer from
depression. Ultimately, these children attain
less education and are more likely to be unemployed
in adulthood.5
Although most research focuses on academic
skills, such as vocabulary size, complexity of
spoken language, familiarity with the alphabet
and books, basic counting, classification,
and what is called ?general knowledge,?
readiness for school also requires social and
emotional skills. Children must be able to follow
directions, work with a group, engage in
classroom tasks, and exert some impulse control.
In a 1997 report, the National Education
Goals Panel emphasized that preparedness
went beyond academics.6 And a poll of
kindergarten teachers found that they rate
knowledge of letters and numbers as less important
readiness skills than being physically
healthy, able to communicate verbally, curious
and enthusiastic, and able to take turns
and share.7
Like the child whose academic skills are
weak, the child who cannot sit still (even for a
few minutes), who interferes with his neighbors,
who has temper tantrums, or who yells
or hits (more than the average kindergartner)
is likely to have difficulty in school.8 Such
early problems of self-regulation, as they are
sometimes called, are predictive of future
problems. For example, preschool children
who exhibit highly aggressive and disruptive
behaviors are at risk for juvenile delinquency
and school drop-out during adolescence.9
Not surprisingly, children with poor selfregulation
not only spend less time ?on task?
in classrooms, which may lead to academic
difficulties, but also elicit more negative reactions
from their peers and teachers, further
reducing social skills and encouraging disengagement
from school.
At the Schoolhouse Door:
Ready or Not
How many children arrive at school each year
not ready to learn? In a national survey of
more than 3,500 kindergarten teachers in the
late 1990s, 46 percent of teachers indicated
that at least half of the children in their classrooms
were having problems following directions,
some because of poor academic skills
and others because of difficulties working in
a group.10 Problems were more common
among black and Hispanic children than
among whites. Similarly, teachers in schools
Cecilia Rouse, Jeanne Brooks-Gunn, and Sara McLanahan
6 THE FUTURE OF CHILDREN
with a high proportion of minority children
reported substantially more problems than
teachers in schools with a low proportion of
minorities. In short, kindergarten teachers
perceived their black and Hispanic children
as lagging behind white children in both the
academic and the self-regulatory aspects of
school readiness.
We emphasize at the onset that by focusing
on school readiness we do not mean to ?let
schools off the hook.? Much work remains to
be done to understand and improve education
for all children. Rather, by focusing on
essential aspects of children?s lives before
they enter school, we seek to understand how
we might ultimately close the racial and ethnic
gaps in educational outcomes. Only by
having a comprehensive view of all the factors
that underlie academic achievement can
policymakers and practitioners begin to close
those gaps.
In addition, we have chosen to focus on racial
and ethnic differences in school readiness as
opposed to levels of readiness. Although we
agree that the ultimate goal of public policy
should be to improve the readiness of all children,
we believe that in a divided society
such as the United States, attempting to raise
the bar for the most needy students is a worthy
goal, consistent with basic American values.
We also felt that by focusing on the racial
and ethnic gaps in readiness, we would simultaneously
highlight policies that were
likely to raise the bar for all students.
What the Issue Does
The articles in this issue address several
questions. How large are the racial and ethnic
gaps in school readiness? How much of
the gap is due to differences in children?s socioeconomic
background or to genetics?
How much do disadvantages like poor health,
poor parenting, low-quality preschool child
care, and low birth weight contribute to the
gaps? What lessons can we learn from new
research on brain development? What do we
know about what works and what does not
work in closing the gaps?
Contributors to this issue were chosen carefully,
and each is an expert in his or her field.
In our original charge, we encouraged the
authors not to discuss every paper written on
a particular topic but rather to identify the
most important findings and give the reader
their ?best assessment? of the bottom line.
We also asked them to indicate when important
information was missing or ambiguous.
Thus the issue seeks to clarify what we do
and don?t know about disparities in school
readiness.
We also note that many articles in this issue
focus on the black-white test score gap rather
than on gaps for other races and ethnicities.
The lack of emphasis on Hispanics (and to an
even larger degree other races and ethnicities)
is largely due to limits in the available
data. Newer data sets include more students
from a wider range of backgrounds, and we
expect to learn much more about other racial
and ethnic gaps in the future.
I n t r o d u c i n g t h e I s s u e
VOL. 15 / NO. 1 / SPRING 2005 7
By focusing on essential
aspects of children?s lives
before they enter school, we
seek to understand how we
might ultimately close the
racial and ethnic gaps in
educational outcomes.
Finally, the articles focus more on the academic
than on the social and emotional skills
that make up school readiness. As yet, researchers
simply know less about racial and
ethnic gaps in social and emotional skills and
about the conditions (parenting, child care,
child and maternal health) that account for
these gaps. Whenever such information is
available, the authors include it. In the years
ahead we expect researchers to place more
emphasis on the social and emotional aspects
of school readiness.
What We Have Learned
The articles that follow provide the latest information
and findings on a wide range of
questions, and full summaries are provided at
the beginning of each. In this section, we
highlight what we see as the most important
findings.
Testing for School Readiness
A variety of standardized tests show substantial
racial and ethnic disparities at the time
children enter school. Estimates of the gap in
school readiness range from slightly less than
half a standard deviation to slightly more
than 1 standard deviation. According to Don
Rock and Jack Stenner, estimates of the racial
and ethnic gaps in school readiness among
preschool children depend on the type of test
used to measure readiness. Vocabulary tests
typically show gaps of 1 standard deviation or
more. Reading and math achievement tests
show gaps ranging from four-tenths to sixtenths
of a standard deviation. A key question
that has yet to be answered by researchers
and policymakers is what accounts for the
difference in the estimates. The fact that the
smallest gap comes from a recent survey conducted
by the U.S. Department of Education?
the Early Childhood Longitudinal Survey
of Kindergarten children (ECLS-K)?
makes the question even more crucial. Is the
ECLS-K estimate a more accurate measure
(meaning that the gap is overstated in other
tests)? Is it smaller because it comes from
more recent data (meaning that the gap has
narrowed over time)? Or is it smaller because
it measures a different aspect of readiness
from the other tests? The strongest evidence,
although still inconclusive, suggests the difference
lies in how the tests measure school
readiness.
Socioeconomic Background:
Important but Elusive
Greg Duncan and Katherine Magnuson document
that 10 percent of white children, as
against 37 percent of Hispanic and 42 percent
of black children, live in poverty. Further,
the better the socioeconomic status of a
child?s family, the more likely that child is to
be ?ready? for school. Given the close links
between race and ethnicity and family socioeconomic
status, on the one hand, and socioeconomic
status and school readiness, on
the other, it is not surprising that family socioeconomic
status appears to explain a substantial
portion of the racial and ethnic gaps
in readiness.
In some respects, estimates of the role of
family socioeconomic status complicate efforts
to understand the racial and ethnic
gaps. One problem is that family socioeconomic
status is a proxy for many of the underlying
factors that affect school readiness. For
example, parents in families with low socioeconomic
status are less likely to talk to, read
with, and teach young children than are parents
in families with high socioeconomic status.
And both socioeconomic conditions and
parenting behavior are associated with school
readiness.
Another problem is that researchers have not
been able to pinpoint what socioeconomic
Cecilia Rouse, Jeanne Brooks-Gunn, and Sara McLanahan
8 THE FUTURE OF CHILDREN
conditions provide for families vis-?vis children.
If poor parents were to get more
money, would they purchase better child
care, more learning materials for the home,
or increased access to health care? Does increased
income alter parenting, and if so,
why? Do parents with more money lead less
stressful lives, resulting in less depression
and anxiety and perhaps less harsh parenting?
Or do parents with more money have
more time to spend with their children? Similarly,
what does parents? higher education
provide for families?
Even more problematic, if, as is often the
case, researchers use an aggregate measure of
socioeconomic conditions (one that includes
income, parental occupation, and parental education),
it is not possible to know which aspect
or aspects of socioeconomic conditions
are contributing to the improvement in children?s
preparation for school. Because researchers
do not as yet have definitive answers
to these questions, knowing that family
socioeconomic status matters is not the same
as knowing why it matters and hence how this
knowledge can be used to close the gap.
Other Contributors to the
Readiness Gap
Other articles in this issue focus on the individual
factors that contribute to cognitive development
and school readiness and for which
socioeconomic status is likely a proxy: environmental
stress, health, parenting, early child
care experiences, the impact of being born
low birth weight, and genetic endowment.
ENVIRONMENTAL STRESS. Although still in
its infancy, new research on brain development
can potentially shed much light on how
to close the gap in school readiness. Kimberly
Noble, Nim Tottenham, and B. J. Casey explain
that chronic stress or abuse in childhood
can impair development of the hippocampus,
the region of the brain involved in
learning and memory, and reduce a child?s
cognitive ability. Thus the impact of stress on
brain development during childhood may explain
a large portion of the gap in school
readiness. Another finding of neuroscience
research?that children?s brains remain plastic
and capable of growth and development
much longer than previously believed?suggests
that targeted educational interventions
have the promise of improving both brain
function and behavior even among children
in the most distressing life circumstances.
HEALTH. Improving the health of mothers
and infants may also help to close racial and
ethnic gaps in school readiness. Janet Currie?s
back-of-the-envelope calculations suggest
that racial differences in health may account
for about 13 percent of the racial gap in
school readiness, maternal breastfeeding another
6 percent, and maternal depression yet
another 6 percent. She estimates, then, that
child health combined with maternal health
and behavior may account for as much as onefourth
of the racial gap in school readiness.
Nancy Reichman reports that racial and ethnic
disparities in low birth weight only explain
up to 4 percent of the aggregate gap in
school readiness. Although there are substantial
black-white differences in rates of low
birth weight, and although disabilities arising
from very low birth weight can seriously impair
cognitive development, Reichman notes
that the overall effect on the racial and ethnic
gaps is relatively small, because low birth
weight affects only a small share of children.
PARENTING. Jeanne Brooks-Gunn and Lisa
Markman document substantial racial and
ethnic variation in certain parenting behaviors,
such as nurturance, discipline, teaching,
I n t r o d u c i n g t h e I s s u e
VOL. 15 / NO. 1 / SPRING 2005 9
and language use, that are linked to children?s
cognitive, social, and emotional skills. Most
striking are differences in language use.
Black and Hispanic mothers talk less with
their young children than do white mothers
and are less likely to read to them daily. Black
and Hispanic families also have fewer reading
materials in their homes. The authors
conclude that parenting differences can explain
as much as one-half of the racial and
ethnic differences in school readiness.
EARLY CHILDHOOD EDUCATION PROGRAMS.
Katherine Magnuson and Jane
Waldfogel note that children who attend center-
based child care or preschool programs
enter school more ready to learn. And they
find racial and ethnic differences both in the
share of children enrolled in preschool programs
and in the quality of care they receive.
Black children are more likely than white
children to be enrolled in preschool, particularly
in Head Start, the publicly funded program
for children from impoverished families.
Hispanic children are much less likely
than white children to attend preschool;
those who do attend are more likely to attend
Head Start. Black children are more likely to
attend lower-quality preschool programs than
their white peers. According to the authors,
equalizing access to center-based care could
close up to 26 percent of the gap between
Hispanic and white children. Improving the
quality of Head Start programs could close
between 4 and 10 percent of the black-white
gap and between 4 and 8 percent of the Hispanic-
white gap. It is not clear, however, how
much the racial and ethnic gaps in school
readiness would be reduced if all centerbased
programs, not just Head Start, were to
become high quality.
GENETICS. Although cognitive ability is both
highly heritable and important for school
achievement, William Dickens concludes
that genetic endowment does not contribute
significantly to black-white gaps in school
readiness. He notes, though, that studies of
the role of genes and environment in determining
school readiness offer some useful
lessons in designing interventions to narrow
the gaps. For example, he cites the positive
effects of preschool interventions designed to
increase cognitive ability and suggests ways
to counter their often-noted ?fadeout effects??
that is, the decline in cognitive gains
once the program ends. Such interventions,
he says, can induce long-lasting changes by
setting off multiplier processes, whereby improved
ability leads to more stimulating environments
and still further improvements in
ability. The best interventions, he argues,
would saturate a social group (say, all members
of a community or school) and reinforce
initial positive effects with new interventions
in the elementary school years and perhaps
beyond.
Accounting for the Gaps in Readiness:
A Caution
As noted, several authors provide estimates
of how much different factors contribute to
the overall readiness gap. We caution that
tempting as it is to try to do so, one cannot
simply add up these estimates to determine
how much of the overall gap they explain.
The difficulty is that these factors are highly
correlated with one another, and thus when
viewed individually, any one factor is likely to
be picking up the effect of others. For example,
one set of authors argues that approximately
40?50 percent of the racial and ethnic
gap in school readiness may be attributed to
parenting behaviors, while another author attributes
one-fourth of the gap to differences
in child and maternal health and behaviors.
Yet it would be a mistake to conclude that
taken together parenting and child and ma-
Cecilia Rouse, Jeanne Brooks-Gunn, and Sara McLanahan
10 THE FUTURE OF CHILDREN
ternal health and behaviors explain 65?75
percent of the gap. Why? Because part of the
reason why maternal health and behavior
matter is that physically and mentally healthy
mothers may be better parents. In any case,
the effect of child health and maternal health
and behaviors on cognitive development is already
(at least partly) accounted for by parenting.
Adding the two estimates together
would overstate what we know about the gap.
The same could be said for socioeconomic
status and child care.
That said, Roland Fryer and Steven Levitt
have examined what explains racial and ethnic
gaps in school readiness using the most
recent data from the U.S. Department of Education.
They found that family socioeconomic
status, number of books in the home,
low birth weight, and other factors account
for 70?80 percent of the gaps in reading and
math.11 In essence, the message of this issue
is similar: taken together, family socioeconomic
status, parenting, child health, maternal
health and behaviors, and preschool attendance
likely account for most of the racial
and ethnic gaps in school readiness.
Closing the Gaps: What Works
and What Doesn?t
What does this issue tell us about how to close
the racial and ethnic gaps in school readiness?
We?ve learned that some strategies that might
seem obvious turn out to be less promising
than expected. Although child health, for example,
is an important determinant of school
readiness and of the racial and ethnic gaps in
school readiness, increasing poor children?s
access to Medicaid and state child health insurance
is unlikely to narrow these gaps because
poor and near-poor children are already
eligible for public insurance. The problem is
that not all eligible children are enrolled. And
increasing enrollment may not be the answer
either: socioeconomic disparities persist in
Canada and the United Kingdom despite universal
public health insurance.
Similarly, given the importance of socioeconomic
factors, it might appear that the best
way to close the gaps in school readiness
would be to reduce racial and ethnic disparities
in parents? economic resources. Programs
such as the earned income tax credit (which
supplements the earnings of low-income parents),
the minimum wage, and the child tax
credit increase low-income families? economic
well-being. Making the child tax credit
refundable for those who do not earn enough
to pay taxes would do even more to raise the
family incomes of poor and minority children.
To date, however, there is no strong evidence
that increasing parental income positively affects
the school readiness of children.
Helping parents further their education
might also appear to be an effective strategy.
Increasing the schooling of all black and Hispanic
mothers by one or two years, for example,
would significantly narrow the school
readiness gap of their children. But to date
few interventions have been able to produce
such gains in maternal schooling. Although
more intensive programs might enjoy more
success, they may not be cost effective. In
sum, although programs that increase the socioeconomic
status of families are likely to reduce
economic disparities and make a modest
impact on racial gaps, we believe that
approaches that directly address the child
and parental behaviors that contribute to
school readiness will prove more effective.
One such strategy that holds long-term
promise comes from the nascent field of neuroscience.
Researchers are making great
strides in understanding how the brain develops
and what aspects of experience help or
I n t r o d u c i n g t h e I s s u e
VOL. 15 / NO. 1 / SPRING 2005 11
hinder the process. Educational interventions
are already able both to raise children?s
scores in tests of reading and to increase activity
in the brain regions most closely linked
with reading. The areas of the brain that are
most critical for school readiness may thus
prove quite responsive to effective therapeutic
interventions?even making it possible to
tailor particular interventions for individual
children. Although this field is in its infancy,
such tailoring may one day make educational
interventions quite effective in closing racial
and socioeconomic gaps in readiness and
achievement.
For the present, however, we believe that by
far the most promising strategy is to increase
access to high-quality center-based early
childhood education programs for all lowincome
three- and four-year-olds. Such a step
would measurably boost the achievement of
black and Hispanic children and go far toward
narrowing the school readiness gap.
So what should these programs look like?
First and foremost, the education component
of these programs must be of high quality.
This means having small classes with a high
teacher-pupil ratio, teachers with bachelor
degrees and training in early childhood education,
and curriculum that is cognitively
stimulating. Few of the child care centers
and Head Start programs that now serve lowincome
children meet all of these standards.
Second, the new programs should train
teachers to identify children with moderate
to severe behavioral problems and to work
with these children to improve their emotional
and social skills. Although such training
is now being provided by some Head Start
programs and some preschool programs, it is
not available in most center-based child care
programs.
Third, the new programs should include a
parent-training component that reinforces
what teachers are doing in school to enhance
children?s cognitive and emotional development.
Examples of such training would include
encouraging parents to read to their
children on a daily basis and teaching parents
how to deal with behavioral problems. Improving
parental skills would have important
multiplier effects on what teachers were
doing in the classroom.
Fourth, the new programs should provide
staff to identify health problems in children
and to help parents get ongoing health care
for their children. Including an annual home
visit as part of this service would allow staff to
further screen for serious mental health
problems among parents. Although some
Head Start programs and child care centers
in low-income communities do link parents
with health care services for their children,
these programs do not include a home visit,
nor do they address the health needs of
parents.
Finally, the new programs should be well integrated
with the kindergarten programs
that their children will eventually attend so
Cecilia Rouse, Jeanne Brooks-Gunn, and Sara McLanahan
12 THE FUTURE OF CHILDREN
We believe that by far
the most promising strategy
is to increase access to
high-quality center-based
early childhood education
programs for all low-income
three- and four-year-olds.
that the transition from preschool to kindergarten
is successful for children, parents,
and teachers. Again, to have their greatest
impact, high-quality programs must aim at
saturating the classroom and the community
and changing multiple aspects of the child?s
environment.
We know that high-quality early childhood
programs exist. And the best research confirms
that they make great headway in closing
racial and ethnic gaps in school readiness.
The problem is that these programs reach
only a small proportion of low-income children.
Decades ago, this country made a commitment
to do the unthinkable?to put a
man on the moon. Today our aim is both
more and less lofty. We know how to help a
child begin school ready to learn. We know
how to begin to close racial and ethnic gaps
in school readiness. We simply must decide
to do so.
I n t r o d u c i n g t h e I s s u e
VOL. 15 / NO. 1 / SPRING 2005 13
Endnotes
1. Wendy S. Grigg and others, The Nation?s Report Card: Reading 2002, NCES 2003-521 (U.S. Department
of Education, Institute of Education Sciences, National Center for Education Statistics, 2003).
2. Christopher Jencks and Meredith Phillips, ?The Black-White Test Score Gap: An Introduction,? in The
Black-White Test Score Gap, edited by Jencks and Phillips (Brookings, 1998), pp. 3?4.
3. Meredith Phillips, James Crouse, and John Ralph, ?Does the Black-White Test Score Gap Widen after
Children Enter School?? in The Black-White Test Score Gap, edited by Jencks and Phillips (Brookings,
1998).
4. James J. Heckman and Alan B. Krueger, Inequality in America (MIT Press, 2004).
5. Nazli Baydar, Jeanne Brooks-Gunn, and Frank F. Furstenberg, ?Early Warning Signs of Functional Illiteracy:
Predictors in Childhood and Adolescence,? Child Development 64 (1993): 815?29; Jeanne Brooks-
Gunn, ?Do You Believe in Magic? What Can We Expect from Early Childhood Intervention Programs??
Social Policy Report of the Society for Research in Child Development 17, no. 1 (2003) (www.srcd.org/
spr17-1.pdf).
6. National Education Goals Panel, Special Early Childhood Report (Government Printing Office, 1997).
7. Sheila Heaviside and Elizabeth Farris, Public School Kindergarten Teachers? Views of Children?s Readiness
for School, NCES 1993-410 (Department of Education, National Center for Education Statistics, 1993).
8. C. Cybele Raver, ?Emotions Matter: Making the Case for the Role of Young Children?s Emotional Development
for Early School Readiness,? Social Policy Report of the Society for Research in Child Development
16, no. 3 (2002) (www.srcd.org/spr16-3.pdf); Lisa McCabe and others, ?Games Children Play: Observing
Young Children?s Self-Regulation across Laboratory, Home and School Settings,? in Handbook of Infant
and Toddler Mental Health Assessment, edited by Rebecca DelCarmen-Wiggins and Alice Carter (Oxford
University Press, 2004), pp. 491?521.
9. B. R. Hamre and Robert C. Pianta, ?Early Teacher-Child Relationships and the Trajectory of Children?s
School Outcomes through Eighth Grade,? Child Development 72 (2001): 625?88.
10. Sara E. Rimm-Kaufman, Robert C. Pianta, and Martha J. Cox, ?Teachers? Judgments of Problems in the
Transition to Kindergarten,? Early Childhood Research Quarterly 15 (2000): 147?66.
11. Roland G. Fryer and Steven D. Levitt, ?Understanding the Black-White Test Score Gap in the First Two
Years of School,? Review of Economics and Statistics 86, no. 2 (May 2004): 447?64.
Cecilia Rouse, Jeanne Brooks-Gunn, and Sara McLanahan
14 THE FUTURE OF CHILDREN
Assessment Issues in the Testing
of Children at School Entry
Donald A. Rock and A. Jackson Stenner
Summary
The authors introduce readers to the research documenting racial and ethnic gaps in school
readiness. They describe the key tests, including the Peabody Picture Vocabulary Test (PPVT),
the Early Childhood Longitudinal Study (ECLS), and several intelligence tests, and describe
how they have been administered to several important national samples of children.
Next, the authors review the different estimates of the gaps and discuss how to interpret these
differences. In interpreting test results, researchers use the statistical term ?standard deviation?
to compare scores across the tests. On average, the tests find a gap of about 1 standard deviation.
The ECLS-K estimate is the lowest, about half a standard deviation. The PPVT estimate
is the highest, sometimes more than 1 standard deviation. When researchers adjust those gaps
statistically to take into account different outside factors that might affect children?s test scores,
such as family income or home environment, the gap narrows but does not disappear.
Why such different estimates of the gap? The authors consider explanations such as differences
in the samples, racial or ethnic bias in the tests, and whether the tests reflect different aspects
of school ?readiness,? and conclude that none is likely to explain the varying estimates. Another
possible explanation is the Spearman Hypothesis?that all tests are imperfect measures of a
general ability construct, g; the more highly a given test correlates with g, the larger the gap will
be. But the Spearman Hypothesis, too, leaves questions to be investigated.
A gap of 1 standard deviation may not seem large, but the authors show clearly how it results in
striking disparities in the performance of black and white students and why it should be of serious
concern to policymakers.
VOL. 15 / NO. 1 / SPRING 2005 15
www.future of children.org
Donald A. Rock is with the Educational Testing Service. A. Jackson Stenner is chairman and CEO of Metametrics Inc. The authors thank Timothy
Taylor, managing editor of the Journal of Economic Perspectives, for extensive contributions to the improvement of this article.
D o n a l d A . R o c k a n d A . J a c k s o n S t e n n e r
16 THE FUTURE OF CHILDREN
In study after study over the past ten
years, researchers from a variety of
fields using a variety of testing approaches
have consistently found a
gap between the readiness of white
children and the readiness of black and Hispanic
children to enter school. The concept
of ?readiness,? however, has no obvious unit
of measurement. Lacking such a tool, researchers
have used a range of tests to measure
different dimensions of the skills and behaviors?
word comprehension, reading,
math, the ability to sit still?that make a child
?ready? to enter school. If a test is accurate, a
child?s score can be used to predict his future
success or achievement. A student who is
measured as more ?ready? should have
greater success in meeting the demands or
challenges of school.
We begin by introducing the main tests that
researchers have used to measure the readiness
gap for children entering kindergarten.
We then review the range of evidence that
these studies have produced about the size of
the gap. Perhaps not surprisingly, the evidence
on the size of the gap differs somewhat
from one study to the next, and we discuss
how to interpret these differences. The articles
that follow in this volume explore possible
underlying causes of the readiness gap:
family and neighborhood characteristics, genetic
differences, neuroscience and early
brain development, prenatal experiences,
health of young children, and differences in
parenting, child care, and early education.
How Can Readiness Be Assessed
at Kindergarten Entry?
Many experts in the field suggest that it is difficult
if not impossible to assess a child?s academic
performance accurately before age
six.1 Some studies have argued that scores on
preschool or kindergarten readiness tests can
predict no more than 25?36 percent of the
variance in performance in early grades.2
Even if these estimates are correct, predicting
25 to 36 percent of the variance in later
achievement is not to be sneezed at. But we
believe that readiness tests have improved
substantially in the past decade or so and that
the new tests are likely to provide a better
measure of readiness. For example, kindergarten
test scores in the Early Childhood
Longitudinal Study, which we discuss in
more detail later, predict about 60 percent of
the variance in performance at third grade.
Before reviewing the main tests of kindergartners?
readiness to enter school, we will
consider some general characteristics of
these tests and how they work.
Key Characteristics of Readiness Tests
Readiness tests may be given on a group or
individual basis. Group tests can be less expensive
to administer. But for kindergarten
students, individual tests are preferred for
several reasons. Administrators are more
likely to be able to get and hold the attention
and cooperation of a beginning kindergartner
in a one-on-one setting than in a group.3
Small children often enjoy the individual attention
they get from the test administrator,
which helps make the scores more accurate.
In a longitudinal study, one scheduled to
have multiple retestings over several years, a
sizable share of the follow-ups might require
one-on-one retestings because the children
scatter as time passes. Starting with a group
administration and then switching to one-onone
follow-ups could cause variance in the
data that would be difficult to quantify. Individualized
testing gives children the time
they need to finish the assessment and thus
gathers relatively complete information on
each child. It also allows the test to be
adapted to some degree to the abilities of
each child.
Indeed, the best readiness tests are adaptive,
which means that instead of asking every
child identical questions, they give children
harder questions if they do well on the early
questions and easier questions if they do
poorly early on. Operationally, a single test
form is liable to be too hard for 10?20 percent
of the children in the sample and too
easy for another 10?20 percent. In this case,
a ?floor and ceiling? problem will arise: a substantial
share of children will answer all or almost
all of the questions correctly, while another
substantial share will answer all or
almost all incorrectly. Floor and ceiling problems
are the bane of all readiness tests, because
they mean that the distribution of test
scores at the top and bottom of the scale will
barely spread out at all, thus artificially narrowing
the range of student achievement.
Floor and ceiling problems also make it difficult
to measure whether student scores
change over time, because students clustered
at the top or the bottom will often remain in
this pattern when retested. An adaptive test
avoids these problems and allows test scores
to reflect the full range of student achievement.
The main disadvantage of adaptive
testing is cost. It is expensive to develop a
large pool of items to cover the appropriate
span of abilities and to ensure that a common
procedure is followed in deciding when students
will receive harder or easier questions.
A computer-assisted test format is often helpful
in advising the administrator which items
are appropriate for each child. Indeed, adaptive
tests for older, computer-knowledgeable
children can be administered and scored in
real time at a computer terminal.
A useful test must be reliable, which means
that it will produce essentially the same results
on different occasions. Reliability can be
measured in three ways: retesting, equivalent
form, and internal consistency. Retesting, or
giving the same test over again to the same
students, raises obvious questions about how
students react to being given the same test
twice. But retesting that produces dramatically
different results would certainly raise
some flags about reliability. The equivalent
form approach uses two equivalent versions
of a test, which can then be compared with
each other. The internal consistency approach
breaks a single test into parts, which
are then compared with each other. For example,
the results of all even questions might
be compared with those of all odd questions
(the ?split half? test). Or more complex mathematical
formulas might be used to split up
the test in many different ways and then average
those results (to generate a measure
known as ?coefficient alpha?). Whatever the
measure, reliability is assessed along a scale
from 0 to 1, where 1 means that a test has
perfect reliability and gives exactly the same
result each time and 0 means that the results
from the test at one time are completely uncorrelated
with the results the next time. A
reliability score of .90 or above would represent
high reliability; in the .80s, medium reliability;
and in the .60s or .70s, low but ac-
A s s e s s m e n t I s s u e s i n t h e Te s t i n g o f C h i l d r e n a t S c h o o l E n t r y
VOL. 15 / NO. 1 / SPRING 2005 17
Floor and ceiling problems
are the bane of all readiness
tests, because they mean that
the distribution of test scores
at the top and bottom of the
scale will barely spread out at
all, thus artificially
narrowing the range of
student achievement.
ceptable reliability. A reliability score in the
.50s or lower would raise serious questions
about the usefulness of the test.
Some have expressed concern that readiness
tests may not be reliable for very young children
because of their short attention spans.
But individualized test assessment typically
retains the attention of younger children.
And very young children may be less likely
than, say, seniors in high school to respond
randomly or counterproductively to test
questions. Brief descriptions of the major
readiness tests used in this volume follow.
Peabody Picture Vocabulary Test?
Revised
The Peabody Picture Vocabulary Test?Revised
(PPVT-R) is an individually administered
test of hearing (or receptive) vocabulary.
4 Each of two forms of the test contains
five practice items and a set of 175 test items
ordered by difficulty. An easy item might be
?cat?; a difficult one, ?carrion.? All items appear
in the same format: four black-andwhite
illustrations on a single cardboard stock
plate. The examiner says a stimulus word
aloud, and the examinee selects the image
that best illustrates the meaning of the word.
The test is adaptive, establishing a floor
below which the examinee is assumed to
know all word meanings, so that no more
words below the floor are asked, and a ceiling
above which the examinee is assumed to
know no word meanings, so that no more
words above the ceiling are asked. Testing
typically takes between sixteen and thirty
minutes, and the examinee typically responds
to thirty-five to forty-five items.
The PPVT-R is a direct measure of vocabulary
size. The rank order of item difficulties is
highly correlated with the frequency with
which the words are used in spoken and written
discourse.5 The PPVT-R was normed on a
nationally representative sample of 4,200
children and 828 adults.
The PPVT-R is a widely used test, with good
reliability. Reviews of its reliability conducted
by the ERIC Clearinghouse on Assessment
and Evaluation found split-test reliabilities
ranging from the .60s to the .80s and testretest
reliabilities ranging from the .70s to
the .90s.
For studies of kindergarten readiness, it is
useful to test a large sample of children about
whose families substantial background data
are available. Two large samples of kindergarten
children have taken the PPVT-R.
The first is the National Longitudinal Surveys,
a set of U.S. government surveys that
track people over time. The National Longitudinal
Survey of Youth 1979 (NLSY79),
began tracking a nationally representative
sample of 12,686 young men and women
aged fourteen to twenty-two in 1979. They
were interviewed each year through 1994
and have been interviewed every other year
since. The NLSY79 collected some data on
children born to participants in the study, but
in 1986 the survey began collecting much
more intensive data about all children born
to mothers in the NLSY79. The expanded
survey administered the PPVT-R to children
aged three to five (with some differences, according
to the survey year).
A second large data sample of kindergartners
is the Infant Health and Development Program
(IHDP), a study funded by several private
foundations and the U.S. government. It
identified a group of 985 infants born with low
birth weights in eight different cities in 1985
and tracked their development through 2000
using various tests, including the PPVT-R,
D o n a l d A . R o c k a n d A . J a c k s o n S t e n n e r
18 THE FUTURE OF CHILDREN
which was administered when the children
were three and again when they were five.
The PPVT-R finds substantial differences in
black-white readiness for kindergarten. For
example, the vocabulary of black children in
first grade is about half that of white first
graders.6 But two puzzles have arisen about
PPVT-R findings. First, the PPVT-R often
finds a larger black-white readiness gap than
do other readiness tests. Second, studies
using the PPVT-R on different samples of
children have produced estimates of the
black-white readiness gap that vary relatively
widely, given that all involve nationally representative
samples of children of comparable
age using the same vocabulary measure.
These issues will be discussed further below.
Wechsler Preschool and Primary Scale of
Intelligence?Revised
The Wechsler Preschool and Primary Scale
of Intelligence?Revised (WPPSI-R) is an individually
administered test of general intellectual
functioning for children from ages
three to seven years and three months. It
does not require reading or writing. The total
battery contains many subtests: information,
vocabulary, word reasoning, comprehension,
similarities, block design, matrix reasoning,
picture concepts, picture completion, object
assembly, symbol search, coding, receptive
vocabulary, and picture naming. Each subtest
may include questions of several types. In the
vocabulary subtest, for example, the child is
asked to name an object (like a hammer)
when she sees its picture and is asked to define
a word when she hears it spoken. The
test is not adaptive.
The components of the Wechsler test can be
analyzed for individual patterns of learning,
but readiness studies typically use an overall
score based on all test components. Raw
scores are converted into IQ scores with an
average of 100. The IQ scores are scaled according
to age groups, based on a nationally
representative sample of 1,700 children in
the relevant years. Reliability estimates for
scores on the Wechsler test are high, typically
ranging from the high .80s into the mid-.90s,
depending on the kind of reliability that is
reported.
The Wechsler test is often administered to
learning-disabled or gifted children, but because
such children are not randomly selected,
their tests are of little use in researching
the readiness gap. The WPPSI-R was,
however, given to the children in the Infant
Health and Development Program when
they were five years old, thus providing a
broad sample for analysis.
Stanford Binet
The Stanford-Binet Intelligence Scale, fourth
edition (SB-IV), is a measure of ?cognitive
abilities that provides an analysis of pattern,
as well as the overall level of an individual?s
cognitive development,? according to the examiner?s
handbook.7 The SB-IV is individually
administered. It uses results from the vocabulary
test to determine starting items for
fourteen other tests, and thus is somewhat
adaptive. Items in each of the fifteen tests are
ordered as to difficulty. Raw scores are then
converted to standard age scores for four
cognitive areas: verbal reasoning, abstract/visual
reasoning, quantitative reasoning, and
short-term memory. The scores for each of
these cognitive areas plus a composite standard
age score (CSAS) are set to average 100
for each age group.
Reliability scores for the composite Stanford-
Binet score as calculated by the internal consistency
method (that is, dividing the test into
parts and comparing the parts with each
A s s e s s m e n t I s s u e s i n t h e Te s t i n g o f C h i l d r e n a t S c h o o l E n t r y
VOL. 15 / NO. 1 / SPRING 2005 19
other) range from .95 to .99. The reliability of
the four cognitive area scores ranges from .80
to .97. These high correlations between the
four area scores and the composite scores
suggest that the cognitive area profiles are unlikely
to provide reliable diagnostic information
beyond that provided by the total score.
Like the Wechsler Preschool and Primary
Scale of Intelligence, the Stanford-Binet test
was also given as part of the Infant Health and
Development Program (IHDP), in this case
when the children were three years old, thus
providing a substantial sample for analysis.
Woodcock?Johnson Psycho-Educational
Battery?Revised
The Woodcock-Johnson?Revised (WJ-R) is
an extensive battery of cognitive and academic
achievement tests intended for people
as young as two and as old as ninety-five. All
tests are individually and adaptively administered.
Seven abilities are tested and separately
reported: fluid reasoning; comprehension/
knowledge; visual processing; auditory
processing; processing speed; long-term
retrieval; and short-term memory. The standard
battery then reports on four achievement
clusters: broad reading, broad mathematics,
broad written language, and broad
knowledge. Two forms are available for the
achievement tests. Raw scores are converted
into grade and age equivalents.
The test manual reports high reliability. Internal
consistency reliabilities for the cognitive
and achievement clusters are all in the
.90s. The shorter cognitive subtests that contribute
to the seven ability scores have internal
consistency reliabilities in the mid .70s to
low .90s. The reliabilities of the achievement
subtests that contribute to the broad achievement
clusters are all in the high .80s and low
.90s. Although alternate forms are available
D o n a l d A . R o c k a n d A . J a c k s o n S t e n n e r
20 THE FUTURE OF CHILDREN
for the achievement clusters, these reliabilities
are not reported in the manual.
Measures of Behavioral Readiness
The tests discussed so far have focused on academic
achievement?that is, skills involving
words, patterns, and the like. But another important
dimension of readiness for kindergarten
involves behavior, such as the ability to
manage one?s own emotions and to work well
with others.
The Achenbach System of Empirically Based
Assessment offers a range of diagnostic tests
for behavior. The Child Behavior Checklist
(CBCL), once called the Revised Child Behavior
Questionnaire, asks mothers 120 questions
about how frequently they have observed
various behaviors in their children
over the past six months. The checklist was
given to the mothers of the children in the
IHDP dataset when the children were aged
three and five, thus providing a broad basis
for analysis. The Achenbach checklist can be
used to diagnose many behavioral issues, but
it commonly focuses on two broad concerns:
?internalizing? behavior, such as being too
fearful, anxious, unhappy, sad, or depressed;
and ?externalizing? behavior, such as destroying
objects or having temper tantrums.
The Behavioral Problems Index (BPI), derived
from the Achenbach test and other
tests of child behavior, asks mothers twentyeight
questions about the frequency of behaviors
they have observed in their children
over the past three months.8 Results can be
used to produce internalizing and externalizing
scores. The test also produces an overall
composite score, which is expected to average
100. The BPI was given to the women
who entered the NLSY data set in 1979 after
they had become mothers, when their children
were at least four years old.
Yet another approach to assessing a child?s
behavioral readiness is direct observation.
Often a parent and child are asked to play
with some toys or to solve a puzzle together.
The session is videotaped. Coders who have
had extensive training watch the videotapes
and rate behaviors like enthusiasm, persistence,
frustration, and engagement.9
Early Childhood Longitudinal Study?
Kindergarten Battery
Until the late 1990s, the study of school
readiness rested on the few tests already described
(all of which were originally developed
for broader or different purposes than
assessing school readiness) and on the two
main sources of systematic data already mentioned,
the NLSY and the IHDP. Without in
any way disparaging the work done with
these data, researchers felt that addressing a
new source of nationally representative data
with up-to-date instruments for evaluation
might prove extremely helpful. The result
was the Early Childhood Longitudinal Study,
Kindergarten Class of 1998?99 (ECLS-K),
administered by the National Center for Education
Statistics. The new data set began
with a base year fall assessment of 21,260
kindergartners who were then reassessed in
the spring of their kindergarten year and in
the spring of their first and third grade
years.10 Retests are also scheduled for the
spring of fifth grade.
In an effort to move away from onedimensional
cognitive assessments toward
multidimensional approaches, the ECLS-K
evaluates kindergartners along several dimensions
in tests that are individually administered
and adaptive in design.11 The direct
cognitive assessments focus on three areas:
reading, mathematics, and ?general knowledge?
(knowledge of the social and physical
world). In addition, kindergarten teachers assess
both cognitive progress and social or behavioral
skills, and parents assess social competence
and skills. Finally, children receive a
physical assessment, including measures of
fine and gross motor skills. So far, the
parental questions and the tests of fine and
gross motor skills have not proven reliable.
With the former, the main concern is that
parents often have little basis for determining
whether behavior is age appropriate. With
the latter, the main concern is that the scores
may be measuring a child?s ability to comprehend
the instructions as much as his motor
skills. As a result, we will not discuss the parents?
assessments or motor skills tests.
Cognitive tests of kindergarten readiness
tend to concentrate on reading and to a lesser
extent on mathematics because reading and
math abilities are believed to be more modifiable
by preschool programs, parental behavior,
and formal schooling than some other
aspects of readiness. In the ECLS-K the
adaptive tests in reading and mathematics
begin with a first-stage test of fifteen to eighteen
test items covering the full range of difficulty.
A computer calculates a score and then
advises the test administrator which secondstage
form is appropriate for that child. The
direct cognitive assessment takes from fifty to
seventy minutes.12
Because most entering kindergartners cannot
read, the ?reading? test at the kindergarten
level emphasizes the child?s performance on
the sequential learning steps based on the
phonics approach to reading development,
including tasks having to do with familiarity
with print, identifying upper- and lower-case
letters by name, associating letters with
sounds at the beginning of words, associating
sounds with letters at the end of words, and
recognizing common words by sight. As the
ECLS-K moves through later grades, the em-
A s s e s s m e n t I s s u e s i n t h e Te s t i n g o f C h i l d r e n a t S c h o o l E n t r y
VOL. 15 / NO. 1 / SPRING 2005 21
phasis in the item pool shifts toward reading
comprehension skills, such as showing a
more complete understanding of what is
read, connecting knowledge from the text
with the child?s personal knowledge, and
showing some ability to take a critical stance
toward the text.
The ECLS-K mathematics test assesses
knowledge in the following areas (in order of
difficulty): identifying one-digit numerals,
recognizing geometric shapes, and one-toone
counting up to ten objects; reading all
one-digit numerals, counting beyond ten, and
using nonstandard units of length to compare
objects; reading two-digit numbers, recognizing
the next number in a sequence, ordinality
of objects; solving simple addition and subtraction
problems; and solving simple multiplication
and division problems. Again, the
kindergarten test emphasizes the easier skills,
and the tests in later grades shift toward the
more advanced skills.
The direct cognitive measures of reading and
mathematics have reliability in the low .90s?
equal to or better than scores typically found
in cognitive achievement tests given to older
children. Moreover, it was frequently reported
that the children did not want to end
their assessment, largely because they enjoyed
the individual attention from the test
administrator. The test administrators received
considerable training, including practice
sessions, and the materials in the test
were colorful and ?game-like.?
Kindergarten teachers also evaluated their
students along both cognitive and behavioral
dimensions. Good rating scales attempt to
anchor subjective assessments by including
specific descriptions of grade-appropriate
performance or behaviors that are then rated
on a five-point scale, with the highest number
indicating that the child is proficient at
the specified skill. In testing cognitive skills,
the teacher evaluations follow the same general
categories of reading, math, and general
knowledge. The teacher social skills rating
scale (TSRS) rates the kindergarten children
on five socioemotional skills. ?Approaches to
learning? rates a child?s attentiveness, task
persistence, eagerness to learn, learning independence,
flexibility, and organization.
?Self-control? measures the child?s ability to
control behavior by respecting the property
rights of others, controlling temper, accepting
peer ideas for group activities, and responding
appropriately to peer pressure. ?Interpersonal
skills? rates the child?s behavior in
forming and maintaining friendships; getting
along with people who are different; helping
and comforting other children; expressing
feelings, ideas, and opinions in positive ways;
and being sensitive to the feelings of others.
?Externalizing problem behaviors? measures
the likelihood that a child argues, fights, gets
angry, acts impulsively, and disrupts ongoing
activities. ?Internalizing problem behaviors?
measures anxiety, loneliness, low self-esteem,
and sadness.
Although these teacher ratings may seem
subjective, they proved almost as reliable as
D o n a l d A . R o c k a n d A . J a c k s o n S t e n n e r
22 THE FUTURE OF CHILDREN
Good rating scales attempt
to anchor subjective
assessments by including
specific descriptions of gradeappropriate
performance or
behaviors that are then rated
on a five-point scale.
the direct cognitive scores. The teacher?s rating
of the child?s reading development was a
very respectable .87, while the teacher?s rating
of a child?s mathematical development
was .92. Similarly the teacher social ratings
all had reliability close to .90, except for the
measure of self-control, which had an acceptable
reliability of .79.
How well are the direct cognitive ratings correlated
with the teacher evaluations? Such
correlations help evaluate what researchers
call ?construct validity,? the extent to which a
test measures what it is intended to measure.
A measure has construct validity if it correlates
well with other tests that theory suggests
are measuring similar things (?convergent
validity?) and if it correlates relatively
poorly with other tests that theory suggests
are measuring different things (?discriminant
validity?).13 In this case, the difficulty is that
the teacher evaluations of reading and math
achievement are quite highly correlated, at
.83. The correlation between teacher evaluations
of reading and cognitive evaluation of
reading, at .60, is exactly the same as for
math. Similarly, the teacher evaluation of
math has only a very slightly higher correlation
with the cognitive measure of math, at
.54, than it does with the cognitive measure
of reading, at .51.
In addition, some of the nonacademic
teacher ratings of social skills, notably selfcontrol
and interpersonal skills, are more
highly correlated with the academic ratings
than are the corresponding test scores, which
suggests a possible ?halo? effect among the
teacher ratings. However, the high correlation
of the self-control scale and the interpersonal
skills scale with the teachers? ratings of
academic performance, and to a lesser extent
with the tested academic performance, is also
consistent with Andrew Pellegrini?s theory
A s s e s s m e n t I s s u e s i n t h e Te s t i n g o f C h i l d r e n a t S c h o o l E n t r y
VOL. 15 / NO. 1 / SPRING 2005 23
that social skill development predicts literacy
performance.14
The Size of the Readiness Gap
Various studies have used the tests and data
sources described here to measure the readiness
gap for kindergartners. Table 1 lists
some selected studies that have measured academic
readiness; table 2 presents studies
that have measured social or behavioral
readiness. The first column of each table lists
the authors and the date of the study. The
second column identifies the test used. The
third column comments on the data used.
The final columns list what are called ?raw
gaps? and ?adjusted gaps,? measured in
?standard deviation units.? These terms require
further explanation.
Using Standard Deviation as a Common
Yardstick
The human sciences in general?and psychology
and education in particular?lack
common, shared interchangeable metrics for
expressing differences on many important
constructs, like reading achievement, health
risk, or depression. There are more than 200
nonexchangeable metrics for assessing how
well students read.15 Each reading test reports
in a scale specific to that test?like the
PPVT or the ECLS-K reading scale?but no
tables exist for converting the score on one
reading scale into the metric of another. How
can researchers compare the results of studies
done with different instruments?
To visualize the problem, consider figure 1,
which shows a common pattern that arises in
studies of readiness among black and white
children. The darker line shows the distribution
of scores for black children, the lighter
line that for white children, in a study using
the PPVT as the test and the NLSY79 data.
The test scores have been coded so that the
average score for white and black children
combined is 50. The median score for blacks
(that is, the score that half the children are
above and half below) is 40; the median score
for whites is 52. Most children, however, are
not exactly at the middle, but are rather
above or below it, and so graphs of scores on
readiness tests typically take on a hill, or bell,
shape, with relatively few children at the extremes
and more clustered near the middle
of the distribution. The gap between the median
white and black scores is 12 points?but
who knows what that means compared with
any other vocabulary or readiness scale?
Statisticians have a tool called the standard
deviation for measuring the spread of a bellshaped
distribution.16 A standard deviation
tells how far a distribution is spread out
around the average score?the numerical
scale used to measure the scores doesn?t matter.
To put it another way, imagine that in fig-
D o n a l d A . R o c k a n d A . J a c k s o n S t e n n e r
24 THE FUTURE OF CHILDREN
Table 1. Selected Estimates of the Academic School Readiness Gap
White-black White-Hispanic
Study Test Sample Raw Adjusted Raw Adjusted
Fryer and Levitt ECLS-K 20,000 kindergartners
(2004) Math test (ECLS-K) 0.64 0.09a 0.72 0.20a
ECLS-K
Reading test 0.40 0.12a 0.43 0.06a
ECLS-K
Math teacher assessment 0.28 0.10b 0.24 0.10b
ECLS-K
Reading teacher assessment 0.27 0.07b 0.35 0.18b
Brooks-Gunn, PPVT-R 315 five-year-olds (IHDP)
Klebanov, Smith, Vocabulary 1.63 0.86c
Duncan, and Lee
(2003) WPPSI 315 five-year-olds (IHDP)
IQ 1.21 0.38c
PPVT-R 1,354 five- to six-year-olds
Vocabulary (NLSY child data) 1.15 0.73c
Phillips, Brooks- PPVT-R Five- and six-year-olds
Gunn, Duncan, Vocabulary/IQ (NLSY) 1.14 0.95d
Klebanov, and Crane
(1998) PPVT-R
Vocabulary/IQ Five-year-olds (IHDP) 1.71 0.69d
WPPSI
IQ Five-year-olds (IHDP) 1.28 0.26d
Sources: Roland G. Fryer and Steven D. Levitt, ?Understanding the Black-White Test Score Gap in the First Two Years of School,? Review of
Economics and Statistics, vol. 86, no. 2 (May 2004): 447?64; Jeanne Brooks-Gunn, Pamela K. Klebanov, Judith Smith, Greg J. Duncan,
and Kyunghee Lee, ?The Black-White Test Score Gap in Young Children: Contributions of Test and Family,? Applied Developmental Science
7, no. 4 (2003): 239?52; Meredith Phillips, Jeanne Brooks-Gunn, Greg J. Duncan, Pamela Klebanov, and Jonathan Crane, ?Family Background,
Parenting Practices, and the Black-White Test Score Gap,? in The Black-White Test Score Gap, edited by Christopher Jencks and
Meredith Phillips (Brookings, 1998), pp. 103?45.
Notes: To standardize the score differentials, we used 16 as the standard deviation on the Stanford-Binet and 15 as the standard deviation
on the PPVT-R and the WPPSI, unless the author gave the actual standard deviation for the entire sample. ECLS-K is the Early Childhood
Longitudinal Study-Kindergarten Cohort; IHDP is the Infant Health and Development Program; EHS is the Early Head Start Research and
Evaluation Program; NLSY is the National Longitudinal Survey of Youth Child Supplement.
a. Controls for composite measure of socioeconomic status, a quadratic in the number of children?s books, sex, age attending kindergarten,
birth weight, mother?s age at birth, and WIC participation.
b. Same as note a with the addition of teacher fixed effects.
c. Controls for family income, female headship, mother?s education, mother?s age at birth, and home environment.
d. Controls for family income, female headship, mother?s educational attainment, neighborhood socioeconomic status, home learning environment,
and home warmth.
ure 1, all the scores on the horizontal axis
were multiplied by a factor of 10, or 20, or
any number you choose. The scores themselves
would change, and the measure of the
gap between the peaks of the white and black
distributions would change, but the number
of standard deviations between the two peaks
would be exactly the same. Thus, instead of
expressing the readiness gap in terms of
scores on a particular test, which cannot
readily be compared with scores on other
tests, researchers can express the readiness
gap in terms of standard deviations. In figure
1, the standard deviation is 10 points, so a gap
of 12 points means 1.2 standard deviations.
Using standard deviations to compare distributions
is based on the underlying assumption
that the hill shapes of the distributions
are the same. This assumption is not literally
true. But it remains useful for researchers,
because it creates a ?scale free? measure of
effects that allows comparisons across studies
with different numerical scales.17
Now look back at table 1 and the column
showing the white-black ?raw? gap, the gap
between the averages for white and for black
children before scores are adjusted to take
into account such factors as the age or education
of a child?s mother, family income, or
whether the child was born at low birth
weight. By this measure, the studies listed in
table 1 typically find a white-black gap of
more than 1 standard deviation, with many of
the estimates roughly similar to the gap illustrated
in figure 1. But the estimates of the
white-black raw gap at entrance to kindergarten
using the ECLS-K data are substantially
lower, often hovering at about 0.5 standard
deviation. Finally, the highest estimates
of the raw gap in the table are generated
A s s e s s m e n t I s s u e s i n t h e Te s t i n g o f C h i l d r e n a t S c h o o l E n t r y
VOL. 15 / NO. 1 / SPRING 2005 25
Table 2. Selected Estimates of the Behavioral School Readiness Gap
White-black White-Hispanic
Authors Test Sample Raw Adjusted Raw Adjusted
Magnuson (2004) Approaches to learning 20,000 kindergartners,
teacher reports .36 .21
Self-control (ECLS-K) .38 .13
Externalizing behavior ?.31 .01
Internalizing behavior ?.06 ?.05
Chase-Lansdale, Internalizing behavior 642 five-year-olds, maternal
Gordon, Brooks- (Achenbach CBCL) reports (IHDP) ?.30a
Gunn, and Klebanov
(1997) Externalizing behavior
(Achenbach CBCL) ?.20a
Internalizing behavior 699 five- to six-year-olds,
(BPI) maternal reports (NLSY-CS) ?.01a
Externalizing behavior
(BPI) ?.22a
Sources: Katherine Magnuson, analyses prepared for this article from the Early Childhood Longitudinal Study-Kindergarten Cohort, School
of Social Work, University of Wisconsin (2004); P. Lindsay Chase-Lansdale, Rachel A. Gordon, Jeanne Brooks-Gunn, and Pamela K. Klebanov,
?Neighborhood and Family Influences on the Intellectual and Behavioral Competence of Preschool and Early School-Age Children,?
in Neighborhood Poverty, vol. 1, Context and Consequences for Children, edited by Jeanne Brooks-Gunn, Greg L. Duncan, and J. Lawrence
Aber (New York: Russell Sage, 1997), pp. 79?118.
Notes: ECLS-K is the Early Childhood Longitudinal Study-Kindergarten Cohort; IHDP is the Infant Health and Development Program; NLSYCS
is the National Longitudinal Survey of Youth?Child Supplement.
a. Controls for gender, family income, female headship, mother?s age at birth, mother?s employment, age, and school status.
using the PPVT, some of which are substantially
greater than 1 standard deviation. The
studies listed in table 2 find a much smaller
gap in behavioral readiness, with the raw gap
often in the range of 0.0 to 0.3 standard deviation.
Some measures even find a negative
gap in behavioral readiness, meaning that
black or Hispanic children were more behaviorally
ready for kindergarten on this dimension
than white children.
How Much Does 1 Standard Deviation
Matter?
Should a gap of, say, 1 standard deviation in
reading ability be considered a big difference?
To what extent should policymakers
take note of a white-black achievement gap
that averages 1 standard deviation?
Statisticians often work with what they call a
?normal? distribution, the bell-shaped distribution
produced by many random observations,
such as flipping 100 coins and seeing
how many times heads comes up or rolling
two dice and seeing how often each total
comes up. A rule of thumb for normal distributions
is that 68 percent of all scores will be
within 1 standard deviation above or below
D o n a l d A . R o c k a n d A . J a c k s o n S t e n n e r
26 THE FUTURE OF CHILDREN
Figure 1. Vocabulary Scores for Three- and Four-Year-Olds, by Race
30
25
20
15
10
5
0
Percent
100 95 90 85 80 75 70 65 60 55 50 45 40 35 30 25 20
White
Black
Source: Christopher Jencks and Meredith Phillips, eds., The Black-White Test Score Gap (Brookings, 1998).
Notes: The data are from National Longitudinal Survey of Youth Child study, 1986?94. For blacks, N = 1,134; for whites, N = 2,071. The
figure is based on black and white three- and four-year-olds who took the Peabody Picture Vocabulary Test-Revised. The test is the standardized
residual, coded to a mean of 50 and a standard deviation of 10, from a weighted regression of children?s raw scores on their age in
months, age in months squared, and year-of-testing dummies.
the mean score, while 95 percent of all scores
will be within 2 standard deviations of the
mean. In that spirit, consider the situation in
which the gap between the peak of the hillshaped
distributions of scores for white and
black children is 1 standard deviation. Under
the assumptions that the two distributions
have the same standard deviation and that
both distributions are ?normal,? the following
six statements about the degree of overlap
between the two distributions will all hold
true.18
First, randomly selecting one black child and
one white child and comparing their scores
will show the white child exceeding the black
child 76 percent of the time and the black
child exceeding the white child 24 percent of
the time. Second, 84 percent of white children
will perform better than the average
black child, while 16 percent of black children
will perform better than the average
white child. Third, if a class that is evenly divided
by race is divided into two equal-sized
groups based on ability, then black students
will compose roughly 70 percent, and whites
30 percent, of the students in the lower performing
group. Fourth, if a school district
chooses only the top-scoring 5 percent of students
for ?gifted? courses, such classes will
have thirteen times more whites than blacks.
Fifth, assume that a school district?s student
body mimics the national racial distribution
(17 percent black, 83 percent white and
other). The district chooses the lowestscoring
5 percent of all students for a special
needs program. Although 17 percent of the
district?s children are black, 72 percent of the
special needs students will be black. Finally,
assume that a reading textbook is written so
that the average white student will read it at a
75 percent comprehension rate. The implied
comprehension rate for the average black
student will be 53 percent, virtually guaranteeing
that such a reader will not engage with
the text.19
These statements strongly suggest that a gap
of 1 standard deviation is quite important in
terms of student performance and should be
of serious concern to policymakers. Indeed,
even a gap of 0.5 standard deviation will result
in striking differences between races, especially
in matters like how many students
are assigned to gifted or to remedial classes.
Raw Gap versus Adjusted Gap
Two columns in table 1 are labeled ?raw gap,?
one referring to the gap between whites and
blacks and the other to that between whites
and Hispanics. As noted, the raw gap is calculated
by looking at the distributions for white
students and for either black or Hispanic students
and calculating the difference between
the mean scores, measured in terms of standard
deviations, without making any further
adjustments.
Two other columns are labeled ?adjusted
gap.? The adjusted gap is the raw gap adjusted
statistically to take into account different
factors that might affect scores. For example,
the 2003 study by Jeanne Brooks-
Gunn and others listed in table 1 accounts for
family income, whether a woman is the head
of the family, the mother?s level of education,
the mother?s age at the child?s birth, and aspects
of the home environment. The adjusted
gap calculates how much one would expect a
white and black (or Hispanic) student to differ
even if both had the same family income,
the same type of head of household, mothers
of the same education and age, and the same
home environment. Different studies use different
data on the child and family, so one
study?s adjusted score will account for different
factors than another?s. The specific factors
taken into account in the adjusted scores
are listed in the notes to tables 1 and 2.
The adjusted gap often substantially reduces
the raw gap, although how much it does so
varies across test instruments and studies.
This pattern suggests that influences outside
school, such as family background, health,
and neighborhood, can have important effects
on a child?s academic readiness for school. In
some of the calculations using the ECLS-K
data in table 1, these other factors can almost
completely account for the raw gap in whiteblack
academic scores. In most, however,
some gap in academic scores remains even
after adjustment. In table 2, the adjusted
scores are often near zero or even negative,
suggesting that outside factors can more than
explain any behavioral readiness gap.
Can the Differing Estimates of
Readiness Be Reconciled?
No one would reasonably expect the gaps in
school readiness between white, black, and
Hispanic students to be the same in every
study, regardless of the particular test and the
data used. What factors might help explain
and interpret some of the differences across
tests? In particular, why does the most recent
A s s e s s m e n t I s s u e s i n t h e Te s t i n g o f C h i l d r e n a t S c h o o l E n t r y
VOL. 15 / NO. 1 / SPRING 2005 27
and seemingly up-to-date study, the ECLSK,
produce a substantially lower measure of
the readiness gap than do other tests and
data?
Sample Characteristics
When two studies differ, a first obvious question
is whether they are based on different
data. But the data from both the NLSY and
the ECLS-K are chosen to be nationally representative,
so they should show no systematic
difference. And the IHDP data set, although
it was not chosen to be nationally
representative, is a large enough group and
has been studied for long enough that it is
unlikely to have a buried flaw that would call
results into question. Many of the studies of
kindergarten readiness discussed here struggle
with such issues as how to make good
comparative measurements with children
who do not speak English as a first language,
or are blind, or perhaps have a condition like
cerebral palsy that makes it difficult to finish
the test, and to address these issues they
make various adjustments. But although differences
in the samples certainly explain
some of the variation around the edges, they
seem unlikely to account for substantial
variation.
Racial or Ethnic Bias in the Tests?
A common concern is that the readiness gap
measured between white and minority children
may be caused by systematic bias in the
test; for example, perhaps certain vocabulary
words are more commonly used in white
families than in black or Hispanic ones.
There are many ways to check for racial or
ethnic bias.
One straightforward approach is to look at
groups of white and minority children who
have the same overall scores on the test.
These children should also have essentially
the same breakdown of right and wrong answers
on each question on the test. Otherwise,
?differential item functioning? exists,
and an item on the test may be sorting by
race or ethnicity rather than ability.
A related concept is construct bias; that is,
whether a test measures what it purports to
measure. A test is construct biased if items
tend to be more familiar to one group than
another, so that the characteristics of the test
question help to explain why whites, blacks,
or Hispanics find the questions hard or easy
to answer. More than thirty years of intense
examination of the possibility of construct
bias, with particular focus on white-black differences,
has failed to demonstrate that they
are due to construct bias in achievement
tests.20
Prediction bias might arise if a school district
used a ?school readiness battery? administered
in kindergarten to predict third grade
reading proficiency and found that the ability
of the test to predict later proficiency differed
for blacks, Hispanics, and whites. In
general, though, achievement test items like
reading, vocabulary, mathematics, social
studies, and science function the same for
blacks and whites. That is, test scores on
achievement tests predict similarly for blacks
and whites?and indeed, at the high school
level, they have a slight tendency to overpredict
black outcomes in college grades and
workplace performance (rather than underpredict,
as would be expected if there were
prediction bias).21 Thus, claims of prediction
bias for achievement tests are, for the most
part, not sustainable.
Another possibility is that even if the test instruments
themselves are not racially or ethnically
biased, the broader social context in
which these tests and their uses are embed-
D o n a l d A . R o c k a n d A . J a c k s o n S t e n n e r
28 THE FUTURE OF CHILDREN
ded may lead to racial or ethnic gaps in outcomes.
Claude Steele and Joshua Aronson
have conducted studies that show that calling
a test ?a diagnostic measure of ability? produces
in black students a ?stereotype threat,?
resulting in poorer test performance. The
black-white gap is markedly reduced when
the test does not bear the label ?intellectual
ability.? Steele and Aronson caution against
generalizing these findings beyond high
achievers at a prestigious university and call
for further study of the central hypothesis
and its many implications.22 In particular, it is
not clear whether this issue would affect
kindergartners.
Are the Tests Different Ways of
Measuring a Common Underlying
Readiness?
There is little evidence that distinctions such
as verbal versus nonverbal, group administered
versus individually administered, spatial
versus numerical, or paper-and-pencil
versus performance test explain the pattern
of gap size estimates. Differences in the
readiness gap across the tests can to some extent,
however, be explained by the Spearman
Hypothesis. This hypothesis states that all
tests are imperfect measures of a general
ability construct, commonly known as g. The
more highly a given test correlates with g, the
larger will be the black-white readiness gap.23
Highly specific school-related tasks, like
those involving handwriting or auditory
memory span, have lower correlations with
general ability (g). But tests that involve reasoning
with figures or vocabulary tests like
the PPVT-R correlate highly with g. When a
test combines multiple task types into a composite,
as do all the tests reviewed above
(other than the PPVT-R), the composite
score correlates more highly with g than do
the specific subtests?in keeping with the
Spearman Hypothesis. In effect, composite
scores average out the specific contributions
of particular task types, leaving what is common
among them?that is, general ability, g.
Researchers have tested the Spearman Hypothesis
repeatedly over the past twenty
years by looking at the common factors across
the intelligence tests, and the hypothesis has
successfully predicted the pattern of blackwhite
differences in thirteen studies using a
broad array of cognitive tests.24
But the ?vocabulary? construct measured by
the PPVT-R seems to pose a challenge to the
Spearman Hypothesis. Even though one
would expect vocabulary to be highly correlated
with general ability (g), it is only one
measure and thus should presumably produce
a smaller black-white readiness gap
than do composite scores. But as noted, the
PPVT-R produces some of the highest estimates
of the readiness gap. Further, theories
of vocabulary acquisition emphasize that
words with high frequency in written and oral
discourse are learned first, and words with
low frequency are learned later; that is, children
learn words primarily because they are
exposed to them.25 And the order of vocabulary
acquisition is highly invariant for advantaged
and disadvantaged populations. Perhaps
the greater exposure to words in some
way exaggerates differences in underlying
general ability, but the reasons why vocabulary
tests often produce a larger readiness
gap than composite achievement tests remain
to be investigated.
What about the ECLS-K?
The readiness gap as measured by the ECLS
Kindergarten sample is consistently smaller
than that detected by the other methods,
whether using raw or adjusted scores. Why
might this be so? The ECLS test was designed
more recently, with many useful up-
A s s e s s m e n t I s s u e s i n t h e Te s t i n g o f C h i l d r e n a t S c h o o l E n t r y
VOL. 15 / NO. 1 / SPRING 2005 29
dates in its methodology and administration,
and it has a larger and more recent database.
These factors might contribute to a smaller
measure of the readiness gap.
Another possibility that fits with the Spearman
Hypothesis, however, is that the version
of the ECLS-K test given to kindergarten
students is less correlated with general ability,
g, than is the version given later to, say,
third graders. Remember that the ECLS-K
test evolves and looks different for different
age levels. In kindergarten the ECLS-K
reading test involves basic phonics and decoding
tasks; by third grade, the emphasis of
the reading test has shifted toward comprehension,
with a heavy word-meaning component.
As the ECLS-K assessment moves on
from basic skill processes in kindergarten to
product outcomes in third grade it finds a
larger black-white readiness gap. Indeed, the
ECLS-K readiness gap as of third grade is
much closer to that found by other test instruments.
It is possible that the lower ECLS
readiness gap at the kindergarten level may
reflect the specific way it tests kindergartners.
26 At the same time, student scores on
the ECLS kindergarten test are very highly
correlated with their scores on the test in
third grade, suggesting that the two tests are
not in fact measuring different constructs.
Clearly, the reasons why the ECLS-K test
generates smaller estimates of the racial and
ethnic gaps in school readiness are not well
understood and are worthy of serious future
study, because of the important implications
for education policy.
Future Directions for Research
on the Readiness Gaps
Future research on the school readiness gaps
among black, white, and Hispanic children
will depend to a large extent on the availability
of new data and the uses of new methods.
Data from the ECLS Kindergarten 1998?99
cohort have invigorated research in this area.
And ECLS is also now tracking a sample of
10,600 children born in 2001 whom it plans
to follow through first grade. The new study
seems certain to provide further evidence
about the size and underlying causes of the
racial and ethnic readiness gaps. Researchers
should also be on the lookout for situations in
which a large group of kindergarten-age children,
such as the IHDP group, might usefully
be administered an achievement test.
Another approach is to use different methods.
A relatively new line of thought emphasizes
a kind of cognitive measurement that is
highly correlated with general ability, g.
?Choice reaction time? is the time it takes the
subject to react to a light stimulus by moving
her index finger from a home base to one or
more of eight lights arranged in a semicircle.
Total reaction time is decomposed into the
milliseconds it takes the examinee to remove
her index finger from the home base after the
stimulus light is activated and the time it
takes after removing the index finger to touch
the stimulus switch. The two times are experimentally
independent. The procedure is
simple, can be used for all ages, requires no
memory component, and is highly reliable.
D o n a l d A . R o c k a n d A . J a c k s o n S t e n n e r
30 THE FUTURE OF CHILDREN
Student scores on the ECLS
kindergarten test are very
highly correlated with their
scores on the test in third
grade, suggesting that the two
tests are not measuring
different constructs.
And the time it takes a subject to remove a
finger from the home base is remarkably
highly correlated with cognitive test composites.
27 Some tantalizing links also exist
between reaction time and vocabulary
development.
Most data sets described in this paper are
longitudinal?that is, they track groups of
children over time. Such an approach is obviously
useful for investigating the determinants
and effects of school readiness. But it is
not the only possible approach. For example,
if assessors are interested in a snapshot of the
status of the children at a specific time, a single
cross-sectional study can be less costly
and less complex than a longitudinal study.
Yet another approach is to conduct an experiment
by assigning children to different government
intervention programs and having
each intervention test the children?s school
readiness. For example, the federal government
has supported the Early Head Start Research
and Evaluation Project (EHS), which
has studied seventeen Head Start programs
around the United States since the late 1990s
using a methodology in which 3,000 children
were randomly assigned either to Early Head
Start or to a control group. The first phase of
the study focused on children from birth to
age three, but a second phase from 2001 to
2004 is tracking children from the time they
leave Early Head Start until they enter
kindergarten. The project is evaluating
prekindergarten children using many of the
tools already discussed: the PPVT, the Woodcock
Johnson Psycho-Educational Test Battery,
the Achenbach Child Behavior Checklist,
analysis of videotaped problem-solving
and play sessions, and others. These data will
surely generate a wave of studies of kindergarten
readiness, often with policy implications,
in the next few years. Of course, experimental
evidence of this sort need not be
collected nationwide; such experiments can
also be carried out at the state or metropolitan
levels.
Future research on the readiness gap at
kindergarten will prove useful, but it seems
highly unlikely to overturn the conclusion
that the raw readiness gaps, between white
and black children in particular but also between
white and Hispanic children, are real
and large. The remainder of this issue is devoted
to exploring possible explanations for
this very serious problem, along with their
policy implications.
A s s e s s m e n t I s s u e s i n t h e Te s t i n g o f C h i l d r e n a t S c h o o l E n t r y
VOL. 15 / NO. 1 / SPRING 2005 31
Endnotes
1. Lori Shepherd, Sharon Lynn Kagan, and Emily Wurtz, eds., Principles and Recommendations for Early
Childhood Assessments (Washington: National Education Goals Panel, February 1998).
2. Samuel J. Meisels, ?Can Head Start Pass the Test?? Education Week 22, no. 27 (March 19, 2003): 44; Anthony
D. Pellegrini and Carl D. Glickman, ?Measuring Kindergarteners? Social Competence,? Young Children
(May 1990): 40?44.
3. Sally Atkins-Burnett, Brian Rowan, and Richard Correnti, ?Administering Standardized Achievement Tests
to Young Children: How Mode of Administration Affects the Reliability of Standardized Measures of Student
Achievement in Kindergarten and First Grade,? paper presented at the annual meeting of the American
Educational Research Association, April 2001 (available at www.sii.soe.umich.edu/papers.html).
4. Although the studies reviewed in this issue use the PPVT-R, the test has recently been revised, and studies
now in the field use the PPVT-III. For discussion, see Jeanne Brooks-Gunn, Pamela K. Klebanov, Judith
Smith, Greg J. Duncan, and Kyunghee Lee, ?The Black-White Test Score Gap in Young Children: Contributions
of Test and Family,? Applied Developmental Science 7, no. 4 (2003): 239?52.
5. A. Jackson Stenner, Malbert Smith, and Donald S. Burdick, ?Toward a Theory of Construct Definition,?
Journal of Educational Measurement 20, no. 4 (1983): 304?15.
6. George A. Miller and Patricia M. Gildea, ?How Children Learn Words,? Scientific American 257, no. 3
(1987): 94?99.
7. Elizabeth P. Hagen, Elizabeth A. Delaney, and Thomas F. Hopkins, Stanford-Binet Intelligence Scale?Examiner?s
Handbook: An Expanded Guide for Fourth Edition Users (Chicago: Riverside Publishing Company,
1987).
8. For the genesis of the Behavioral Problems Index, see James L. Peterson and Nicholas Zill, ?Marital Disruption,
Parent-Child Relationships, and Behavioral Problems in Children,? Journal of Marriage and the
Family 48, no. 2 (May 1986). For a discussion of how the BPI is used in the NLS, see Center for Human
Resource Research, NLSY79 Child and Young Adult Data Users Guide (Ohio State University, December
2002), especially pp. 91?94.
9. The article by Jeanne Brooks-Gunn and Lisa Markman in this issue describes in more detail how this approach
was used in one study of 2,000 three-year-olds, with data from the Early Head Start Research and
Evaluation Project (EHS). These data are also discussed further at the end of the present article.
10. National Center for Education Statistics, U.S. Dept. of Education, ECLS-K Base Year Data Files and Electronic
Codebook (2001).
11. Susan M. Benner, Assessing Young Children with Special Needs: An Ecological Perspective (New York:
Longman, 1992); Everett Waters and Alan L. Sroufe, ?Social Competence as a Developmental Construct,?
Developmental Review 3 (1983): 79?97; Anthony Pellegrini, Lee Galda, and Donald L. Rubin, ?Context in
Text: The Development of Oral and Written Language in Two Genres,? Child Development 55 (1984):
1549?55.
12. Frederick M. Lord and Melvin R. Novick, Statistical Theories of Mental Test Scores, with Contributions by
Alan Birnbaum (Reading, Mass.: Addison-Wesley, 1968); Frederick M. Lord, Applications of Item Re-
D o n a l d A . R o c k a n d A . J a c k s o n S t e n n e r
32 THE FUTURE OF CHILDREN
sponse Theory to Practical Testing Problems (Hillsdale, N.J.: Lawrence Erlbaum Associates, 1980). See also
Benner, Assessing Young Children (see note 11).
13. Donald T. Campbell and Donald W. Fiske, ?Convergent and Discriminant Validation by the Multi-Trait
Multi-Method Matrix,? Psychological Bulletin 56 (1959): 81?105.
14. Pellegrini, Galda, and Rubin, ?Context in Text? (see note 11).
15. A. Jackson Stenner and Benjamin D. Wright, ?Readability, Reading Ability, and Comprehension? (paper
presented at the Association of Test Publishers Hall of Fame induction for Benjamin D. Wright, San
Diego, 2002), in Making Measures, edited by Benjamin D. Wright and Mark H. Stone (Chicago: Phaneron
Press, 2004).
16. The mathematical formula for calculating standard deviation works like this: (1) calculate the average of
the scores; (2) calculate the difference between each individual score and the average; (3) square these differences
from the average, and then add them up; (4) take the square root of the total. This calculation will
give the number of points that are equal to 1 standard deviation for this group of scores.
17. Space does not permit a full treatment of the soundness of all assumptions underlying the standard deviation
as a common unit of effect. However, we did compare the standard deviations for five well-known
reading tests that were linked to a common scale and found they ranged from a low of .94 to a high of 1.13.
This modest variability across grades and tests provides a context for evaluating the variability in estimates
of the black-white achievement gap across various studies and instruments reported in this volume.
18. For purposes of this discussion we made the usual simplifying assumptions of bivariate normality, homogeneity
of variance, and equal sample sizes in the two groups. Furthermore, we assume that the 1 standard
deviation difference is in construct measures, not test score performances, which are uncorrected for
measurement error.
19. Using data from the NCES-NAEP website, we estimate that 1.0 standard deviation on NAEP is equivalent
to 220L (220 Lexiles). A back check on this number is to average four norm-referenced achievement test
(NRT) standard deviations. The RMSA standard deviation for the four NRTs is 229L. Comprehension rate
is modeled as the difference between reader ability and text readability. A difference of 225L between a
targeted reader (75 percent comprehension rate) at fourth grade and the average black fourth grader implies
a 53 percent comprehension rate for a ?book bag? of fourth grade textbooks. See Lexile.com, the Lexile
Calculator.
20. For background on construct validity, see Stenner, Smith, and Burdick, ?Toward a Theory of Construct Definition?
(see note 5). For discussion of the evidence, see Richard E. Nisbett, ?Race, Genetics, and IQ,? in
The Black-White Test Score Gap, edited by Christopher Jencks and Meredith Phillips (Brookings, 1998).
The psychometric literature has largely given up on the term ?bias? in favor of the less emotionally charged
terms ?differential item functioning? and ?differential instrument functioning.?
21. See Thomas J. Kane, ?Racial and Ethnic Preferences in College Admissions,? Frederick E. Vars and
William G. Bowen, ?Scholastic Aptitude Test Scores, Race, and Academic Performance in Selective Colleges
and Universities,? and William R. Johnson and Derek Neal, ?Basic Skills and the Black-White Earning
Gap,? in The Black-White Test Score Gap, edited by Jencks and Phillips (see note 20).
A s s e s s m e n t I s s u e s i n t h e Te s t i n g o f C h i l d r e n a t S c h o o l E n t r y
VOL. 15 / NO. 1 / SPRING 2005 33
22. Claude M. Steele, ?Race and the Schooling of Black America,? Atlantic Monthly (April 1992): 68?78;
Claude M. Steele, ?A Threat in the Air: How Stereotypes Shape the Intellectual Identities and Performance
of Women and African Americans,? American Psychologist (June 1997): 613?29; Claude M. Steele
and Joshua Aronson, ?Stereotype Threat and the Intellectual Test Performance of African Americans,?
Journal of Personality and Social Psychology 69, no. 5 (1995): 797?811.
23. Arthur R. Jensen, Bias in Mental Testing (New York: Free Press, 1980), especially p. 146-147; Arthur R.
Jensen, ?Spearman?s Hypothesis Tested with Chronometric Information-Processing Tasks,? Intelligence 17
(1993): 47?77.
24. Arthur R. Jensen, ?Psychometric g and Achievement,? in Policy and Perspectives on Educational Testing,
edited by Bernard R. Gifford (Boston: Kluwer Academic, 1993), pp. 117?227.
25. Betty Hart and Todd R. Risley, The Social World of Children Learning to Talk (Baltimore: Brooks, 1999).
See Stenner, Smith, and Burdick, ?Toward a Theory of Construct Definition? (see note 5), and Miller and
Gildea, ?How Children Learn Words? (see note 6), for introductions to exposure theory. See Jensen, Bias
in Mental Testing (see note 23) for an introduction to education theory.
26. Meredith Phillips, James Crouse, and John Ralph, ?Does the Black-White Test Score Gap Widen after
Children Enter School?? in Jencks and Phillips, The Black-White Test Score Gap (see note 20), pp. 229?71.
27. Jensen, Bias in Mental Testing (see note 23). See also William Hick, ?On the Rate of Information,? Quarterly
Journal of Experimental Psychology 4 (1952): 11?26.
D o n a l d A . R o c k a n d A . J a c k s o n S t e n n e r
34 THE FUTURE OF CHILDREN
Can Family Socioeconomic Resources
Account for Racial and Ethnic
Test Score Gaps?
Greg J. Duncan and Katherine A. Magnuson
Summary
This article considers whether the disparate socioeconomic circumstances of families in which
white, black, and Hispanic children grow up account for the racial and ethnic gaps in school
readiness among American preschoolers. It first reviews why family socioeconomic resources
might matter for children?s school readiness. The authors concentrate on four key components
of parent socioeconomic status that are particularly relevant for children?s well-being?income,
education, family structure, and neighborhood conditions. They survey a range of relevant policies
and programs that might help to close socioeconomic gaps, for example, by increasing family
incomes or maternal educational attainment, strengthening families, and improving poor
neighborhoods.
Their survey of links between socioeconomic resources and test score gaps indicates that resource
differences account for about half of the standard deviation?about 8 points on a test
with a standard deviation of 15?of the differences. Yet, the policy implications of this are far
from clear. They note that although policies are designed to improve aspects of ?socioeconomic
status? (for example, income, education, family structure), no policy improves ?socioeconomic
status? directly. Second, they caution that good policy is based on an understanding of causal
relationships between family background and children outcomes, as well as cost-effectiveness.
They conclude that boosting the family incomes of preschool children may be a promising intervention
to reduce racial and ethnic school readiness gaps. However, given the lack of successful
large-scale interventions, the authors suggest giving only a modest role to programs that
address parents? socioeconomic resources. They suggest that policies that directly target children
may be the most efficient way to narrow school readiness gaps.
VOL. 15 / NO. 1 / SPRING 2005 35
www.future of children.org
Greg J. Duncan is the Edwina S. Tarry Professor of Human Development and Social Policy at Northwestern University. Katherine Magnuson
is an assistant professor of social work at the University of Wisconsin at Madison. The authors are grateful to the Family and Child Well-
Being Research Network of the National Institute of Child Health and Human Development for research support; to Amy Claessens and Mimi
Engel for research assistance; and to the volume editors, Susan Mayer, and Meredith Phillips, for helpful comments.
36 THE FUTURE OF CHILDREN
Greg J. Duncan and Katherine A. Magnuson
in various ways. We then summarize results
from studies that attempt to account for the
racial and ethnic achievement gaps by examining
differences in family socioeconomic
status.
Material Hardship and Family
Socioeconomic Status
Life is very different for a family with a single
parent struggling to make ends meet by
working at two minimum-wage jobs and a
family with one highly paid wage earner and
a second parent at home caring for their children.
One family faces a vast range of material
and psychological hardships, while the
other is largely spared such stressors.3 The
first family, for example, may have a lowerquality
home environment that exposes children
to pollutants and toxins, such as lead,
and provides fewer learning opportunities in
the home or lower-quality child care outside
it. Greater stress may increase the mother?s
irritability and reduce her warmth and responsiveness
to her children. Across racial
and ethnic groups in the United States, such
differences in family resources, particularly
financial resources, are systematic and often
large, prompting researchers to investigate
whether family resource differences may account
for the racial and ethnic differences in
school readiness.
Material Hardship and Household
Resources
The ECLS-K data in figure 2 reveal striking
differences both in a broad range of indicators
of family hardships and in the accumulation
of those disadvantages between poor and
nonpoor children. (Some of the indicators do
not, strictly speaking, point to socioeconomic
status but relate to conditions, such as low
birth weight and depressive symptoms, and
behaviors, like harsh parenting, that are discussed
in other articles in this volume.) The
National tests regularly show
sizable gaps in school readiness
between young white
children and young black and
Hispanic children in the
United States. In the nation?s most comprehensive
assessment of school readiness
among kindergartners, the 1998 Early Childhood
Longitudinal Study (ECLS-K), both
black and Hispanic children scored about
two-thirds of a standard deviation below
whites in math (the equivalent of roughly 10
points on a test with a mean of 100 and a
standard deviation of 15) and just under half
a standard deviation (7?8 points) below
whites in reading (see figure 1).1
What might be causing such gaps? One
prominent possibility is that the historical
racial and ethnic inequalities in the United
States have created disparate socioeconomic
circumstances for the families in which
white, black, and Hispanic children are
reared. As graphed in figure 1, the racial gaps
in family socioeconomic status (SES) of the
children in the ECLS-K closely matched the
gaps in test scores.2 The average socioeconomic
level of black kindergartners was more
than two-thirds of a standard deviation below
that of whites. Hispanic children had even
lower socioeconomic standing relative to
whites.
With such similar racial and ethnic gaps in
test scores and SES, it is tempting to conclude
that equalizing the social and economic
circumstances of white, black, and Hispanic
preschoolers would eliminate most if not all
of the achievement gap. Whether this is likely
is the subject of this article. We begin by considering
theories about why family socioeconomic
resources might matter for children?s
school readiness and reviewing studies of interventions
designed to boost those resources
Figure 1. Racial and Ethnic Gaps in Selected Test Scores and in Family Socioeconomic
Status for Kindergartners
Source: Authors? calculations based on data taken from the ECLS-K.
0
?0.2
?0.4
?0.6
?0.8
?1.0
Standard deviation difference from whites
Hispanic
Blacks
Socioeconomic Index ECLS-K Reading ECLS-K Math
?0.709
?0.445
?0.784
?0.703
?0.389
?0.605
first four items in figure 2 (mother a high
school dropout, single-parent family, mother
with no job or a job with low prestige, and
unsafe neighborhood) are relatively common
indicators of inadequate family economic and
social resources. The next seven items are resource-
related disadvantages often faced by
poor families with children: large family size
(three or more siblings), residential instability
(child moved four or more times before
starting school), harsh discipline (child
spanked two or more times in the past week),
few learning materials (fewer than ten children?s
books in the house), low birth weight
(infant less than 5.5 pounds at birth), young
parents (child born to a teen mother), and
high levels of maternal depressive symptoms.
The contrasts between poor and other children
could hardly be more stark. In almost
every case, more than twice as many poor as
nonpoor children suffer the given hardship,
and for several hardships (high school
dropout mother, bad job, and few children?s
books) the rate is more than three times as
high.
The distribution of hardships differs not only
by poverty status, but also by race and ethnicity
(see table 1). With the exception of residential
instability, black and Hispanic children
are much more likely to experience
hardships than are white children. The
prevalence of single-parent families, low
birth weight, harsh parenting, and maternal
depressive symptoms is highest among black
children. Hispanic children are most likely to
have mothers who did not complete high
school and to have few children?s books in
their homes.
Racial and ethnic differences are also apparent
in the total number of hardships that children
face. The vast majority of black and Hispanic
children suffer at least one hardship,
compared with just over half of white children.
Experiencing four or more hardships is
very rare for white children, but much more
common among Hispanic, and especially
black, children.
Socioeconomic Status or Socioeconomic
Resources?
Some social scientists gather a variety of indicators
of financial and social resources under
the umbrella term of ?socioeconomic status?
(SES). For them, socioeconomic status refers
to one?s social position as well as the privileges
and prestige that derive from access to economic
and social resources. Because it may be
difficult to measure directly a family?s access
to resources or its position in a social hierar-
C a n F a m i l y S o c i o e c o n o m i c R e s o u r c e s A c c o u n t f o r R a c i a l a n d E t h n i c Te s t S c o r e G a p s ?
VOL. 15 / NO. 1 / SPRING 2005 37
chy, analysts often use one indicator (typically
occupation) or combine several indicators (for
example, parental education and occupation)
into scales that indicate families? relative positions
in a social hierarchy.4 The differences in
socioeconomic status shown in figure 1 exemplify
this single-indicator approach. Using a
summary index to measure SES emphasizes
social stratification as an organizing force in
individuals? lives and presumes that one?s social
standing is a more important determinant
of life chances than any of the economic and
social resources that determine it.5
A different approach to measuring SES is
based on the premise that distinct types of
socioeconomic resources contribute to social
inequality and stratification along differing
economic and social dimensions.6 For example,
although parents? educational attainments,
incomes, and occupations are related,
each may affect children in different ways.7
Rather than using a summary measure, proponents
of this approach consider each component
separately, as seen in figure 2. This
method requires a complicated sorting out of
the separate effects of correlated social and
economic disadvantages, which if done incorrectly
may understate the importance of either
the constellation or the accumulation of
household resources. We take this multidimensional
approach throughout this article
by concentrating on four key dimensions of
parental socioeconomic resources?income,
education, family structure, and neighborhood
conditions.8
Greg J. Duncan and Katherine A. Magnuson
38 THE FUTURE OF CHILDREN
Figure 2. Percent of Poor and Nonpoor Children Experiencing Hardships
Source: Same as figure 1.
0 10 20 30 40 50
Poor
children
Nonpoor
children
Mother depressed
Teen mother
Low birth weight
Few children?s books
Spanking
Residential instabilty
Three or more siblings
Low-quality neighboorhood
No or low-prestige job
Single parent
Mother dropout
Are Socioeconomic Resources
Really the Issue?
Before taking a more detailed look at these
resources, we raise a fundamental question:
does SES really determine achievement?
Causation is notoriously difficult to prove in
the social sciences, and just because middleclass
children?s academic achievement exceeds
that of poor children, one should not
necessarily infer that eliminating the income
gap would eliminate the achievement gap.
Maybe what really matters for children?s
achievement is the psychological dispositions of
their parents, including, for example, depression.
As noted, depression is more prevalent
among low- than higher-income parents, as discussed
by Janet Currie in her article in this volume.
Perhaps income and child achievement
are linked because both are higher in the case
of better-adjusted parents. Or maybe the association
between socioeconomic status and
achievement stems from the poorer health and
greater developmental problems of the children,
which can both lower a child?s academic
achievement and reduce a family?s resources by
limiting parents? employment. Moreover, as
pointed out by William Dickens in his article in
this volume, many behavioral geneticists, concluding
that socioeconomic conditions are relatively
unimportant, put forth a different logic.
They argue that genetic endowments of ability
are key determinants of test scores, and children
reared in more affluent families score
higher on achievement tests in part because of
genetic endowments passed on from one generation
to the next.
If parental mental health, child health, or genetic
endowments are what really matter for
children?s achievement, then increasing parents?
income or education without also addressing
these other causes would not boost
achievement. Our discussion of the relationships
between achievement and the four
most important components of SES?income,
education, family structure, and neighborhood?
is mindful of the difficulties of establishing
causal effects.
The best evidence on the effects of socioeconomic
resources on children?s development
comes from experimental studies in which
participants are randomly assigned to a treatment
or a control group. But such studies are
rare in the social sciences. Second-best
strategies involve following large samples of
children for many years and using a host of
statistical strategies to rule out alternative explanations
for the presumed effects.
Household Income
It is easy to see how higher family incomes
might give children a big edge in academic
achievement. Financial resources can enable
C a n F a m i l y S o c i o e c o n o m i c R e s o u r c e s A c c o u n t f o r R a c i a l a n d E t h n i c Te s t S c o r e G a p s ?
VOL. 15 / NO. 1 / SPRING 2005 39
Table 1. Percent of Children Experiencing
Poverty and Hardships, by Race and
Ethnicity
Characteristic White Black Hispanic
Experiencing poverty 10 42 37
Experiencing Hardships
Mother high school dropout 7 18 35
Single parent 15 50 24
No or low-prestige job 8 18 21
Low-quality neighborhood 5 23 21
Three or more siblings 11 21 18
Residential instability 13 12 13
Spanking 7 17 10
Few children?s books 2 20 29
Low birth weight 6 15 8
Teen mother 10 22 19
Mother depressed 11 20 13
One or more hardships 52 87 81
Four or more hardships 4 29 18
Source: Based on data from the ECLS-K study.
parents to secure access to good prenatal
health care and nutrition; rich learning environments,
both in the home and through
child care settings and other opportunities
outside the home; a safe and stimulating
neighborhood; and, for older children, good
schools and a college education.9
But despite abundant evidence of correlations
between income and achievement, the issue
of whether family income is causally linked to
children?s achievement and behavior remains
controversial. A study by Judith Smith and
colleagues compared the achievement of children
in families whose average income fell
below the poverty line between their birth
and age five with that of children in families
whose average income remained above the
poverty line during this period of their childhood.
10 They used statistical techniques to ensure
that any differences in achievement between
poor and nonpoor children were not
due to differences in their mothers? education,
children?s low birth weight, or family
structure. Poverty, they found, accounted for
about 0.30 standard deviation of the gap in
achievement between poor and nonpoor children
(the equivalent of about 4?5 points on a
test with a mean of 100 and a standard deviation
of 15)?enough to explain a substantial
share of the racial gap in achievement. The
achievement gap between middle-income
and higher-income families was not nearly as
large, suggesting that boosting household income
during early childhood would help poor
children more than children from wealthier
families. Children whose families faced deep
and persistent poverty fared the worst and
registered the largest achievement gap, which
again suggests that these children would gain
the most from added income.11
Smith?s study, as well as several others, concludes
that the key advantage bestowed by
higher income is a stimulating learning environment.
The number of books and newspapers
in the home and the access of children
to learning experiences routinely explain
about a third of the poverty ?effect,? as discussed
in the article by Jeanne Brooks-Gunn
and Lisa Markman in this issue.12
Although suggestive of a causal link between
poverty and achievement, this evidence should
not be taken as the final word. A subsequent
study, based on the same data used by Smith
and her coauthors but ruling out a longer list of
alternative explanations for the achievement
gap, estimated a considerably smaller difference
between low- and high-income children.13
A series of experimental welfare reform evaluation
studies during the 1990s made it possible
to observe how increases in family income
affect children?s development. Although all
the experimental programs increased parental
employment, only certain programs increased
family income. Only when income was increased
did preschool and elementary school
children?s academic achievement improve.14
For young children, family income gains of
roughly $1,000 a year translated into achievement
gains of about 0.07 standard deviation,
about 1 point on our reference test. Sustained
over time, even such small gains may be economically
profitable, leading to sizable increases
in lifetime earnings.15
Income, it appears, does matter for children?s
achievement, although perhaps not as much
as some early studies suggested. Estimated at
more than $30,000, the gaps in family income
between white children and black and Hispanic
children are huge. What policies might
begin to close these gaps?
One strategy, embodied in several of the welfare
reform programs described above, is to
Greg J. Duncan and Katherine A. Magnuson
40 THE FUTURE OF CHILDREN
promote low-income parents? participation in
the labor market and reduce their reliance on
welfare. But even the most generous welfare
reform programs boosted average family incomes
by only $1,000 or $2,000 a year. Other
work-oriented interventions, such as low-cost
job search programs, have produced relatively
small absolute income gains for
women?a few hundred dollars over the
course of a year or two.16 More intensive,
training-based programs have netted women
proportionately bigger earnings gains?a few
thousand dollars over several years?but
none created the kind of long-term income
increases that would begin to narrow the income
gap between white families and ethnic
and racial minority families. Employment interventions
for disadvantaged adult men have
had even less encouraging results. Only
about a third of such interventions increased
either employment or earnings, and none
emerged as a panacea.
Another approach is to supplement the incomes
of poor working families through the
earned income tax credit.17 A refundable federal
tax credit for low-income working families
with children, the EITC was expanded
during the 1990s and is now the nation?s
largest cash transfer program for low-income
families. In 2003 the maximum benefit for a
family with two children was about $4,200,
and nearly 19 million families received the
credit.18 In 1997 the program lifted about 2.2
million children out of poverty.19 By providing
income support for low-wage work, the tax
credit also encourages work in single-parent
families. Increases to the EITC in the 1990s
raised the annual employment of poorly educated
single mothers by almost 9 percent.20
Parental Human Capital
Human capital includes parental skills, acquired
both formally and informally, that are
C a n F a m i l y S o c i o e c o n o m i c R e s o u r c e s A c c o u n t f o r R a c i a l a n d E t h n i c Te s t S c o r e G a p s ?
VOL. 15 / NO. 1 / SPRING 2005 41
valuable in the labor market and at home.21
Formal schooling is the most familiar and
most studied form of human capital, and research
confirms that more schooling leads to
better employment and earnings.22 More
schooling may thus indirectly benefit children
by increasing family income, but other
parental skills may also directly enhance child
well-being, for example, by improving parenting
and the ability to accomplish parenting
goals.23
Parents? completed schooling varies widely
by race and ethnicity and is particularly low
among Hispanics, reflecting their immigration
history. Among the ECLS-K sample of
kindergartners, 35 percent of Hispanic mothers
had not completed high school, compared
with only 7 percent of white mothers and 18
percent of black mothers (table 1). At the
other end of the scale, 28 percent of white
mothers had completed a four-year college
program, whereas only 9 percent of black and
8 percent of Hispanic mothers had done so.24
Children with highly educated parents routinely
score higher on cognitive and academic
achievement tests than do children of parents
with less education. Remarkably, the link between
children?s cognitive development and
parental education is evident as early in a
child?s life as three months of age.25 Yet research
has not clearly isolated parental education
as the cause of high child achievement.
26 Few studies are able to disentangle
parents? schooling from other sources of advantage,
such as cognitive endowments, that
may have increased achievement among both
parents and children. The few U.S. studies
that have tried to isolate the effects of
parental education per se typically find positive
but modest effects of maternal and paternal
education on children?s achievement,
with an additional year of schooling linked to
an increase in children?s test scores of about
0.15 standard deviation, or about 2 points on
our reference test.27
It may be that increasing schooling for mothers
who are high school dropouts raises their
children?s achievement more than increasing
education for college-trained mothers.28 According
to a recent study, welfare recipients?
participation in mandated education or training
improved their young children?s school
readiness by as much as a quarter of a standard
deviation, or almost 4 points on our reference
test.29
With large gaps in parental education among
racial and ethnic groups, interventions that
increase rates of high school completion may
have a large payoff for future generations.
But few academic programs developed to increase
high school graduation rates among atrisk
adolescents have been effective so far. A
recent review of sixteen random-assignment
evaluations of dropout-prevention programs
found only one to be successful.30 Rigorous
evaluations of a few intensive teen mentoring
programs have found more promising results,
but nevertheless success is not guaranteed,
particularly when these programs are implemented
on a large scale.31
Studies of low-income populations routinely
report that without any programmatic intervention,
close to 50 percent of disadvantaged
mothers return to school.32 Yet even with
high rates of continued schooling, educational
attainment among economically disadvantaged
parents remains much lower than
among advantaged families. Thus another intervention
approach is to promote educational
activities among parents. For example,
programs targeting teen mothers may provide
support and incentives to stay in school
after the birth of a child, or welfare programs
may make cash benefits contingent on mothers?
participation in education and training.
But evaluations suggest that to date these
types of interventions have not been successful
in boosting mothers? educational activity
above the relatively high level of participation
of control group mothers.33
The high enrollment in further education of
disadvantaged mothers suggests that mothers
might be benefiting from current efforts
to offset the costs of education, particularly
higher education, and to increase access to
educational opportunities. Indeed, expansions
in public spending on higher education,
including more generous financial aid
and an increase in community college funding,
have consistently been linked to higher
levels of college attainment and enrollment.
However, the extent to which educational expenditures
have specifically benefited lowincome
students appears to vary, depending
on the specifics of the spending.34 Still another
approach is to raise the age at which
students may leave school or begin to work.
Such policy changes over the past century
have modestly increased youths? years of
schooling.35
Family Structure
Today about one-third of all children are
born outside marriage, and more than half of
all children will live in a single-parent family
Greg J. Duncan and Katherine A. Magnuson
42 THE FUTURE OF CHILDREN
Income, it appears, does
matter for children?s
achievement, although
perhaps not as much as some
early studies suggested.
at some point in their childhood. This causes
concern because resources can be scarce in
single-parent families.36 Young children living
with single mothers face poverty at five
times the rate of preschoolers in intact families
(50 percent versus 10 percent), and the
declines in income for households with children
after a divorce are dramatic and lasting.
37 Financial and time constraints may
limit a single parent?s ability to supervise and
discipline children and to provide a supportive
and stimulating home environment.38
Furthermore, because fathers are often absent
from single-parent families, children in
these households tend to have fewer male
role models, which may not bode well for
their social development.39
As with education and income, family structure
differences across racial groups are
large. Rates of single-parenthood in the
ECLS-K sample averaged 15 percent for
white children, 24 percent for Hispanic children,
and 50 percent for black children (table
1).40 Black children are more likely to be
born outside marriage; white children, to experience
divorce.
On average, children raised by single parents
have lower social and academic well-being
than the children of intact marriages.41 Most
research on single-parent families has
lumped all varieties of such families together
or focused only on the effects of divorce.42
The few studies that have tried to draw distinctions
find little difference between children
of divorced and never-married parents;
both groups are at greater risk of poor
achievement and behavioral problems than
children from intact families.43
Rates of teenage childbearing have been
steadily falling, dropping 22 percent between
1991 and 2000, from 62.1 births to 48.1
births per 1,000 fifteen- to nineteen-yearolds.
44 Nevertheless, U.S. rates of teen parenthood
continue to exceed those of European
countries. And U.S. teen birth rates
differ substantially by race. As table 1 shows,
about one in five black or Hispanic children
was born to a mother younger than twenty,
nearly twice the rate for white children. Typically,
children of teen mothers face a constellation
of socioeconomic hardships, including
single parenthood, poverty, and lower
maternal educational attainment.45
Although most children from broken families
fare worse than those in intact families, and
children born to teen mothers fare worse
than those born to older mothers, in both
cases it appears that differences in parental
characteristics, such as educational attainment,
rather than family structure or maternal
age per se, account for a portion of the
gaps. Once these differences in family background
are taken into account, growing up
with a single or remarried parent has persistent,
but much more modest negative effects
on children?s achievement.46 For example, a
recent adoption study suggests that differences
in the parental backgrounds of singleand
two-parent families account for a substantial
proportion of children?s achievement
problems after a divorce.47 Similarly, the extent
to which children would benefit from
their mothers? postponing childbearing for a
few years is uncertain, although likely modest.
48
Economic insecurity explains part of the poor
outcomes of children reared in single-parent
or blended families and by young parents.
And parental conflict and strain in divorcing
families may impair children?s development,
particularly with respect to their behavior.49
Finally, children in young and single-parent
families may face many transitions in family
C a n F a m i l y S o c i o e c o n o m i c R e s o u r c e s A c c o u n t f o r R a c i a l a n d E t h n i c Te s t S c o r e G a p s ?
VOL. 15 / NO. 1 / SPRING 2005 43
life, including subsequent cohabitations, remarriages,
separations, and divorces. Such
instability may pose additional risks to child
well-being.50
We know little about whether interventions
can promote marriage and prevent divorce
among disadvantaged populations.51 Yet even
if the current round of federal marriagepromotion
programs succeeds, it is unlikely
to make much of a dent in the huge differences
of family structure between blacks and
whites. Furthermore, it appears that for marriage
to promote children?s achievement substantially,
it must go hand in hand with increases
in family resources, such as income.
Whether higher rates of marriage will improve
other aspects of socioeconomic circumstances
is unclear.52 Evaluations of new
marriage programs should shed light on the
feasibility of increasing marriage rates, as
well as on how doing so will promote children?s
well-being.
Programmatic interventions to prevent teen
childbearing by reducing sexual activity and
promoting contraceptive use among adolescents
have not been very successful. More
often than not, programs designed to postpone
sexual behavior fail to delay its onset or
reduce its frequency.53 Of twenty-eight carefully
evaluated programs focused on abstinence,
sexual education, and HIV prevention,
only ten delayed the age of sexual
initiation. Of the nineteen that measured the
frequency of youths? sexual activity, thirteen
had no significant effect. Nor did the programs
substantially increase contraceptive
use. Only four of the eleven program evaluations
that measured teenagers? use of contraception
found positive effects. A handful of
more intensive interventions that provided
mentoring and constructive after-school activities
had more positive results.54 But
whether these intensive programs can be
replicated on a larger scale is uncertain. As
with dropout-prevention programs, concentrated
intervention is a necessary but not sufficient
condition for success.55
Neighborhoods
Neighborhoods shape children?s development
in many ways, although kindergartners
are probably less susceptible to neighborhood
influences than are adolescents.56 The
risks posed by low-quality neighborhoods are
most striking in high-poverty urban communities
plagued by violence, gangs, drug activity,
old housing stock, and vacant buildings,
where watchful parents may not allow children
to walk to school alone or play outside.57
Such neighborhoods may influence children
through increased stress, perhaps stemming
from community violence; social disorganization,
including a lack of positive role models
and shared values, which may lead to problem
behavior; a lack of institutional resources,
such as strong schools and police
protection; and negative peer influences,
which may spread problem behavior.58 Nevertheless,
studies suggest that neighborhood
characteristics can explain no more than 5
percent of the variation in children?s achievement
and 10 percent of the variation in their
behavior.59
A recent experiment that offered families the
opportunity to move from high-poverty to
low-poverty neighborhoods provides a compelling
test of the extent to which neighborhood
matters for children?s development.
The results are striking. The Moving to Opportunity
(MTO) experiment gave housingproject
residents in five of the nation?s largest
cities a chance to move to low-poverty neighborhoods.
But data collected four to seven
years after the families moved revealed no
differences between program and control
Greg J. Duncan and Katherine A. Magnuson
44 THE FUTURE OF CHILDREN
group children, even among those who were
preschoolers when the program began.60 Despite
dramatic improvements in neighborhood
conditions, children made no gains on
test scores, school success or engagement, or
behaviors. Why not?
One possible explanation is that although the
neighborhoods improved a great deal, the
schools attended by the children did not.61
And although MTO-related neighborhood
advantages appeared to improve the mental
health of mothers, they did not translate into
other kinds of household resources or advantages
that might have promoted children?s
well-being.62 After moving, MTO adults still
resembled their control-group counterparts
in their employment, welfare dependence,
family income, parenting practices, and connections
to their children?s schools and to the
parents of their children?s friends.
Residential mobility programs, then, will not
by themselves remedy the achievement problems
of children in public housing and in
high-poverty neighborhoods. Interventions
focused exclusively on neighborhoods rather
than on influences directly related to the
child, family, and school cannot solve the
myriad problems of children growing up in
high-poverty urban neighborhoods.
Can Family SES Account
for the Gaps?
Both theory and evidence suggest that the
family socioeconomic environments in which
children are reared may account for at least
some differences in school-entry achievement.
Here we review so-called accounting
studies, which estimate the extent to which
socioeconomic differences across groups are
linked to racial and ethnic achievement
gaps.63 We reiterate our warning regarding
causation: accounting studies assume that
SES differences cause achievement differences.
To the extent that this does not hold
true, estimates of the effect of socioeconomic
differences on achievement gaps will likely
overstate the potential of policies to eliminate
differences.
Accounting for the Gaps
Figure 3 shows representative results from
four recent studies of black-white differences
in test scores as children enter school. Math
and reading results (in the left half of the figure)
are taken from the study conducted by
Ronald Fryer and Steven Levitt using data
from the ECLS-K.64 The first bars show the
simple, unadjusted mean racial and ethnic
differences. As noted, black children score
two-thirds of a standard deviation lower than
whites in math and close to half a standard
deviation lower in reading.
To what extent are these gaps due to differences
in socioeconomic resources? A handful
of family and child SES-related measures explain
nearly all of the racial math gap and the
entire racial reading gap. These differences
in family and child background include SES
composite, number of children?s books in the
home, age of entry into kindergarten, birth
weight, age of mother at time of birth, and
whether the mother received the Special
Supplemental Nutrition Program for
Women, Infants, and Children (WIC). The
same characteristics also explain racial and
ethnic gaps in each of the five components of
the math test (for example, counting, relative
size) and the reading test (letter recognition,
beginning sounds) and the gaps for sample
subgroups defined by child gender as well as
the location and racial composition of the
child?s school.65 Figure 4, also using data
drawn from the Fryer and Levitt study, shows
that the same set of SES-related family characteristics
accounts for nearly all of the math
C a n F a m i l y S o c i o e c o n o m i c R e s o u r c e s A c c o u n t f o r R a c i a l a n d E t h n i c Te s t S c o r e G a p s ?
VOL. 15 / NO. 1 / SPRING 2005 45
and reading gaps between Hispanic and
white children.
It is unusual for researchers to find that SES
differences explain all the racial and ethnic
test score gaps. For example, the third set of
bars in figure 3 summarizes results from a
study of gaps in the picture-vocabulary scores
of black and white five- and six-year-olds
from the Children of the National Longitudinal
Study of Youth (CNLSY).66 Not only is
the unadjusted gap much larger in the
CNLSY than in the ECLS-K data?more
than 1 standard deviation, or about 16 points
on our reference test?but a similar collection
of family background measures accounts
Greg J. Duncan and Katherine A. Magnuson
46 THE FUTURE OF CHILDREN
Figure 3. Accounting for Black-White Test Score Gaps with SES
Sources: ECLS-K data are taken from Fryer and Levitt, ?Understanding the Black-White Test Score Gap in the First Two Years of School,? Review
of Economics and Statistics 86 (2004): 447?64; NLSY data are taken from Meredith Phillips, Jeanne Brooks-Gunn, Greg J. Duncan,
Pamela Klebanov, and Jonathan Crane, ?Family Background, Parenting Practices, and the Black-White Test Score Gap,? in The Black-White
Test Score Gap, edited by Christopher Jencks and Meredith Phillips (Brookings, 1998), pp. 103?45; IHDP data are taken from Jeanne
Brooks-Gunn, Pamela K. Klebanov, Judith Smith, Greg J. Duncan, and Kyunghee Lee, ?The Black-White Test Score Gap in Young Children:
Contributions of Test and Family,? Applied Developmental Science 7, no. 4 (2003): 239?52.
Note: Effect sizes calculated by authors using the standard deviation for the sample of white students as the denominator. Variables used
to adjust for SES gap in Fryer and Levitt include an SES composite, number of children?s books in the home, age of entry into kindergarten,
birth weight, age of mother at time of birth, and whether the mother received the Special Supplemental Nutrition Program for Women, Infants,
and Children (WIC). Variables used to adjust for SES in Phillips and others include grandparents? education; grandparents? occupation;
Southern roots; mother?s number of siblings; mother?s number of older siblings; no one in mother?s family subscribed to magazines,
newspapers, or had a library card; percent of white students in mother?s high school; student-teacher ratio in mother?s high school; percent
teacher turnover in mother?s high school; mother?s educational expectations; mother?s self esteem index; two indicators for mother?s sense
of control or mastery; interviewer?s assessment of mother?s attitude toward interview; mother?s education; father?s education; child birth
weight; child birth order; family structure; mother?s age at child?s birth; household size; set of dummy variables for average income; mother?s
AFQT score; mother?s class rank in high school; and interviewer?s assessment of mother?s understanding of interview. For the Brooks-Gunn
and others analyses the SES variables include measures of the income-to-needs ratio averaged over three years, maternal education, family
structure, maternal age at birth, and maternal verbal ability.
0.2
0
?0.2
?0.4
?0.6
?0.8
?1.0
?1.2
Standard deviation difference from whites
Adjusted
for SES
Raw gap
IHDP Verbal CNLSY Picture/Vocabulary ECLS-K Reading ECLS-K Math
?0.638
?0.102
?0.401
0.093
?1.07
?0.502
?0.973
?0.543
Figure 4. Accounting for White-Hispanic Test Score Gaps with SES
Source: Data are taken from Fryer and Levitt, ?Understanding the Black-White Test Score Gap,? table 2.
0
?0.2
?0.4
?0.6
?0.8
Standard deviation difference from whites
Adjusted
for SES
Raw gap
Hispanic Hispanic
ECLS-K Math ECLS-K Reading
?0.722
?0.171
?0.427
?0.076
for only about half of the racial gap, or about
7?8 points.
Figure 3 also presents data on five-year-olds
in the Infant Health and Development Program
(IHDP) study.67 As with the CNLSY,
the IHDP verbal test score gap amounts to
about a full standard deviation (about 15
points), and about half the gap (8 points) appears
to be the result of SES differences between
white and black children. Although
these findings may appear to be contradictory,
an interesting consistency is that SES
explains roughly the same absolute amount of
the gap. In all studies, a collection of SESrelated
measures seems to account for a difference
of about half a standard deviation in
white-black test scores (7?8 points), regardless
of the assessments used or the populations
studied.
Summary
On average, when black and Hispanic children
begin school, their academic skills lag
behind those of whites. Accounting studies
find that differences in socioeconomic status
explain about half a standard deviation of the
initial achievement gaps. But because none
of the accounting studies is able to adjust for
a full set of genetic and other confounding
causes of achievement, we regard them as
providing upper-bound estimates of the role
of family socioeconomic status.
If, indeed, differences in the socioeconomic
backgrounds of young white, black, and Hispanic
children play a causal role in creating
achievement gaps, what are the implications
for policy? The answer is far from clear. First,
no policies address ?socioeconomic status?
directly. They address only its components?
income, parental schooling, family structure,
and the like. Moreover, wise policy decisions
require an understanding of both causal
mechanisms and cost-effective interventions
that can produce desired changes.
To illustrate, suppose that increasing maternal
schooling by one year raises children?s
kindergarten achievement scores by one
quarter of a standard deviation, or roughly 4
points on our reference test. With the
achievement gaps between whites and both
blacks and Hispanics at one-half to threequarters
of a standard deviation (7 to 11
points), a policy that could increase maternal
schooling for all black and Hispanic mothers
by an average of one or two years would significantly
narrow the achievement gap. But
few programmatic interventions can deliver
such gains, and whether further expansions
in educational funding will increase Hispanic
or black mothers? educational attainment will
depend on the specifics of how the money is
spent.
In the case of household income, it appears
that reducing the racial and ethnic differences
in family income by several thousand
dollars would reduce achievement gaps. Political
support for work-based approaches to
boosting income, such as the earned income
tax credit, has increased considerably over
the past decade. Moreover, because income
appears to matter more for preschoolers than
for older children?and much more for poor
children than for others?it seems that an effective
policy would be to adopt childfocused
redistributive efforts using, say, European-
style child allowances or increases in
the EITC with benefits restricted to families
with preschool children. Such programs may
prove politically feasible, because it would be
considerably cheaper to cover only a fraction
of children than to cover all children.68
All in all, given the dearth of successful largescale
interventions, it may be wise to assign
C a n F a m i l y S o c i o e c o n o m i c R e s o u r c e s A c c o u n t f o r R a c i a l a n d E t h n i c Te s t S c o r e G a p s ?
VOL. 15 / NO. 1 / SPRING 2005 47
only a modest role to programs that aim to increase
parents? socioeconomic resources. In
the end, policies that directly target children?s
aptitude or mental and physical health,
discussed in other articles in this issue, may
be the most efficient way to address the gap.
Greg J. Duncan and Katherine A. Magnuson
48 THE FUTURE OF CHILDREN
C a n F a m i l y S o c i o e c o n o m i c R e s o u r c e s A c c o u n t f o r R a c i a l a n d E t h n i c Te s t S c o r e G a p s ?
VOL. 15 / NO. 1 / SPRING 2005 49
Endnotes
1. See the article in this issue by Donald A. Rock and A. Jackson Stenner for a discussion of the different tests
and estimates of the gap.
2. The ECLS-K asked children?s parents about their own schooling, occupations, and household incomes and
then combined these elements into a single socioeconomic status index.
3. Robert Haveman and Barbara Wolfe, Succeeding Generations: On the Effect of Investments in Children
(New York: Russell Sage Foundation, 1994); Gary Evans, ?The Environment of Childhood Poverty,? American
Psychologist 59 (2004): 77?92.
4. Charles Mueller and Toby L. Parcel, ?Measures of Socioeconomic Status: Alternatives and Recommendations,?
Child Development 52 (1981): 13?30.
5. For a discussion of differing approaches to measuring socioeconomic status, see Kenneth A. Bollen, Jennifer
L. Glanville, and Guy Stecklov, ?Socioeconomic Status and Class in Studies of Fertility and Health in
Developing Countries,? Annual Review of Sociology 27 (2001): 153?85; Robert Hauser and John Robert
Warren, ?Socioeconomic Indexes for Occupations: A Review, Update, and Critique,? Sociological Methodology
27: 177?298.
6. Bollen, Glanville, and Stecklov, ?Socioeconomic Status and Class? (see note 5).
7. Greg Duncan and Katherine A. Magnuson, ?Off with Hollingshead: Socioeconomic Resources, Parenting,
and Child Development,? in Socioeconomic Status, Parenting, and Child Development, edited by Marc
Bornstein and Robert Bradley (Mahwah, N.J.: Lawrence Erlbaum Associates, 2001), pp. 83?106.
8. We do not review the literature on the effects of occupation on young children primarily because the research
is sparse and there are no clear interventions that directly target occupation.
9. Greg Duncan and Jeanne Brooks-Gunn, eds., Consequences of Growing Up Poor (New York: Russell Sage,
1997); Evans, ?The Environment of Childhood Poverty? (see note 3); Katherine A. Magnuson and Greg
Duncan, ?Parents in Poverty,? in Handbook of Parenting, edited by Marc Bornstein (Mahwah, N.J.:
Lawrence Erlbaum Associates, 2002), pp. 95?121. Vonnie McLoyd, ?Socioeconomic Disadvantage and
Child Development,? American Psychologist 53 (1998): 185?204.
10. Judith Smith, Jeanne Brooks-Gunn, and Pamela Klebanov, ?The Consequences of Living in Poverty on
Young Children?s Cognitive Development,? in Consequences of Growing Up Poor, edited by Greg Duncan
and Jeanne Brooks-Gunn (New York: Russell Sage, 1997), pp. 132?89.
11. Duncan and Brooks Gunn, Consequences of Growing Up Poor (see note 9); Smith, Brooks-Gunn, and
Klebanov, ?The Consequences of Living in Poverty? (see note 10). See also Eric Dearing, Kathleen Mc-
Cartney, and Beck A. Taylor, ?Change in Family Income-to-Needs Matters More for Children with Less,?
Child Development 72 (2001): 1779?93; Greg Duncan and others, ?How Much Does Childhood Poverty
Affect the Life Chances of Children?? American Sociological Review 63 (1998): 406?23.
12. Smith, Brooks-Gunn, and Klebanov, ?The Consequences of Living in Poverty? (see note 10).
13. Meredith Phillips and others, ?Family Background, Parenting Practices, and the Black-White Test Score
Gap,? in The Black-White Test Score Gap, edited by Christopher Jencks and Meredith Phillips (Brookings,
1998).
Greg J. Duncan and Katherine A. Magnuson
50 THE FUTURE OF CHILDREN
14. Pamela A. Morris and others, How Welfare and Work Policies Affect Children: A Synthesis of Research
(New York: Manpower Demonstration Research Corporation, 2001).
15. Krueger and Whitmore estimate that the one-fifth standard deviation increase in test scores from the Tennessee
STAR class-size experiment could increase future earnings between $5,000 and $50,000, depending
on assumed discount and future earnings growth rates. The .07 effect size, if permanent, would increase
earnings by one-third of these amounts. See Alan Krueger and Diane Whitmore, ?The Effect of Attending
a Small Class in the Early Grades on College Test Taking and Middle School Test Results: Evidence from
Project STAR,? Economic Journal 11 (2001): 1?28.
16. James Heckman, Robert LaLonde, and Jeffrey Smith, ?The Economics and Econometrics of Active Labor
Market Programs,? in Handbook of Labor Economics, edited by Orley Ashenfelter and David Card (New
York: Elsevier, 1999), pp. 1865?2097.
17. Marcia K. Meyers and others, ?Inequality in Early Childhood Education and Care: What Do We Know??
in Social Inequality, edited by Kathryn Neckerman (New York: Russell Sage Foundation, forthcoming).
18. U.S. House of Representatives, Committee on Ways and Means, 2004 Green Book (http://waysandmeans.
house.gov/Documents.asp?section=813 [April 26, 2004]).
19. Council of Economic Advisers, Good News for Low Income Families: Expansions in the Earned Income
Tax Credit and the Minimum Wage (Washington, 1998).
20. Bruce Meyer and Dan T. Rosenbaum, ?Welfare, the Earned Income Tax Credit and the Labor Supply of
Single Mothers,? Quarterly Journal of Economics 116 (2001): 1063?114.
21. Gary Becker, A Treatise on the Family (Harvard University Press, 1981).
22. David Card, ?The Causal Effect of Education on Earnings,? in Handbook of Labor Economics, vol. 3A, edited
by Orley Ashenfelter and David Card (New York: Elsevier, 1999), pp. 1801?63.
23. Erika Hoff, ?The Specificity of Environmental Influence: Socioeconomic Status Affects Early Vocabulary
Development via Maternal Speech,? Child Development 74 (2003): 1368?78; Luis Laosa, ?Maternal Teaching
Strategies in Chicano and Anglo-American Families: The Influence of Culture and Education on Maternal
Behavior,? Child Development 51 (1980): 759?65; Robert T. Michael, The Effect of Education on Efficiency
in Consumption (Columbia University Press, 1972).
24. Four-year college degree attainment is from authors? calculations from the ECLS?K data. For more detailed
information on maternal schooling, see http://nces.ed.gov/programs/coe/2003/section1/tables/
t02_1a.asp [April 13, 2004].
25. Kiki Roe and Robin Bronstein, ?Maternal Education and Cognitive Processing at Three Months as Shown
by Infants? Vocal Response to Mother vs. Stranger,? International Journal of Behavioral Development 11
(1988): 389?95.
26. Pamela Davis-Kean and Katherine Magnuson, ?The Influence of Parental Education on Child Development,?
mimeo, University of Michigan, 2004; James A. Mercy and Lulu Steelman, ?Familial Influence on
Intellectual Attainment of Children,? American Sociological Review 42 (1982): 532?42.
27. See Michelle Neiss and David C. Rowe, ?Parental Education and Child?s Verbal IQ in Adoptive and Biological
Families in the National Longitudinal Study of Adolescent Health,? Behavior Genetics 30 (2000):
C a n F a m i l y S o c i o e c o n o m i c R e s o u r c e s A c c o u n t f o r R a c i a l a n d E t h n i c Te s t S c o r e G a p s ?
VOL. 15 / NO. 1 / SPRING 2005 51
487?95; Mark R. Rosenzweig and Kenneth I. Wolpin, ?Are There Increasing Returns to Intergenerational
Production of Human Capital? Maternal Schooling and Child Intellectual Development,? Journal of
Human Resources 29 (1994): 670?93. Studies using European data have been less conclusive; see Sandra
E. Black, Paul J. Devereux, and Kjell G. Salvanes, ?Why the Apple Doesn?t Fall Far from the Tree,? Working
Paper 10066 (Cambridge, Mass.: National Bureau of Economic Research, 2004); Eric Plug, ?Estimating
the Effects of Mothers? Schooling on Children?s Schooling Using a Sample of Adoptees,? American
Economic Review 94 (2003): 358?68. Many studies attempt to adjust for parent cognitive ability.
28. Robert Haveman and Barbara Wolfe, ?The Determinants of Children?s Attainments: A Review of Methods
and Findings,? Journal of Economic Literature 23 (1995): 1829?78; Katherine Magnuson, ?The Effect of
Increases in Maternal Education on Children?s Academic Trajectories: Evidence from the NLSY,? mimeo,
Columbia University, 2004.
29. The National Evaluation of Welfare-to-Work Strategies Child Outcome Study (NEWWS COS). See
Katherine Magnuson, ?The Effect of Increases in Welfare Mothers? Education on Their Young Children?s
Academic and Behavioral Outcomes: Evidence from the National Evaluation of Welfare-to-Work Strategies
Child Outcomes Study,? Institute for Research on Poverty Discussion Paper (University of Wisconsin,
2003), pp. 1274?303.
30. Mark Dynarski, ?Making Do with Less: Interpreting the Evidence from Recent Federal Evaluations of
Dropout Prevention Programs,? paper presented at the conference Dropouts: Implications and Findings,
held at Harvard University, 2001.
31. Jodie Roth and Jeanne Brooks-Gunn, ?Promoting Healthy Adolescents: Synthesis of Youth Development
Program Evaluations,? Journal of Research on Adolescence 8 (1998): 423?59; David L. Dubois and others,
?Effectiveness of Mentoring Programs for Youth: A Meta-Analytic Review,? Journal of Community Psychology
30 (2002): 157?97.
32. Frank F. Furstenberg, Jeanne Brooks-Gunn, and S. Philip Morgan, Adolescent Mothers in Later Life
(Cambridge University Press, 1987); John M. Love and others, Making a Difference in the Lives of Infants
and Toddlers and Their Families: The Impacts of Early Head Start, vol. 1, Final Technical Report (Princeton:
Mathematica Policy Research, 2002).
33. Sharon M. McGroder and others, National Evaluation of Welfare-to-work Strategies: Impacts on Young
Children and Their Families Two Years after Enrollment: Findings from the Child Outcomes Study (U.S.
Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation,
Administration for Children and Families, 2000); Janet C. Quint, Johannes M. Bos, and Denise F. Polit,
New Chance: Final Report on a Comprehensive Program for Young Mothers in Poverty and Their Children
(New York: Manpower Research Demonstration Corporation, 1997).
34. Susan Dynarksi, ?The Consequences of Lowering the Cost of College,? American Economic Review 92
(2002): 279?85; Neil Sefter and Sarah Turner, ?Back to School: Federal Student Aid Policy and Adult College
Enrollment,? Journal of Human Resources 37 (2002): 336?52; Sarah Turner and John Bound, ?Closing
the Gap or Widening the Divide? The Effects of the G.I. Bill and World War II on the Educational Outcomes
of Black Americans,? Journal of Economic History 63 (2003): 145?77.
35. Daron Acemoglu and Joshua Angrist, ?How Large Are the Social Returns to Compulsory Schooling Effects?
Evidence from Compulsory Schooling Laws,? Working Paper 7444 (Cambridge, Mass.: National BuGreg
J. Duncan and Katherine A. Magnuson
52 THE FUTURE OF CHILDREN
reau of Economic Research, 1999); Philip Oreopoulus, Marianne Page, and Anne H. Stevens, ?Does
Human Capital Transfer from Parent to Child? The Intergenerational Effects of Compulsory Schooling,?
Working Paper 10164 (Cambridge, Mass.: National Bureau of Economic Research, 2003).
36. Sara McLanahan and Gary Sandefur, Growing Up with a Single Parent: What Hurts, What Helps (Harvard
University Press, 1994).
37. U. S. Census Bureau. http://ferret.bls.census.gov/macro/032003/pov/new03_100_01.htm accessed April 26,
2004.
38. Paul R. Amato, ?Children?s Adjustment to Divorce: Theories, Hypotheses and Empirical Support,? Journal
of Marriage and the Family 55 (1993): 23?38; K. Alison Clarke-Stewart and others, ?Effects of Parental
Separation and Divorce on Very Young Children,? Journal of Family Psychology 14 (2000): 304?26; Rebecca
L. Coley, ?Children?s Socialization Experiences and Functioning in Single Mother Households: The
Importance of Fathers and Other Men,? Child Development 69 (1998): 219?30; Jane E. Miller and Diane
Davis, ?Poverty History, Marital History, and Quality of Children?s Home Environments,? Journal of Marriage
and the Family 59 (1997): 996?1007.
39. Maureen Black, Howard Dubowitz, and Raymond H. Starr, ?African American Fathers in Low Income,
Urban Families: Development, Behavior, and Home Environment of Three-Year-Old Children,? Child Development
70 (1999): 967?78; Coley, ?Children?s Socialization Experiences? (see note 38).
40. For data on all children, see www.childtrendsdatabank.org/tables/59_Table_1.htm [March 24, 2004].
41. McLanahan and Sandefur, Growing Up with a Single Parent (see note 36).
42. Sara McLanahan, ?Parent Absence or Poverty: Which Matters More?? in Consequences of Growing Up
Poor, edited by Greg Duncan and Jeanne Brooks-Gunn (New York: Russell Sage Foundation, 1997), pp.
35?48.
43. Marcia J. Carlson and Mary E. Corcoran, ?Family Structure and Children?s Behavioral and Cognitive Outcomes,?
Journal of Marriage and the Family 63 (2001): 779?92; Elizabeth Cooksey, ?Consequences of
Young Mothers? Marital Histories for Children?s Cognitive Development,? Journal of Marriage and the
Family 59 (1997): 245?61; McLanahan, ?Parent Absence or Poverty? (see note 42); Thomas DeLeire and
Ariel Kalil, ?Good Things Come in 3?s: Multigenerational Coresidence and Adolescent Adjustment,? Demography
39 (2002): 393?413.
44. Stephanie J. Ventura, T. J. Mathews, and Brady E. Hamilton, ?Teenage Births in the United States: State
Trends, 1991?2000, an Update,? National Vital Statistics Reports 50 (2002): 1?2.
45. Rebecca L. Coley and P. Lindsay Chase-Lansdale, ?Adolescent Pregnancy and Parenthood: Recent Evidence
and Future Directions,? American Psychologist 53 (1998): 152?66.
46. Carlson and Corcoran, ?Family Structure and Children?s Behavioral and Cognitive Outcomes? (see note
43); Clarke-Stewart and others, ?Effects of Parental Separation and Divorce? (see note 38).
47. Thomas G. O?Connor and others, ?Are Associations between Parental Divorce and Children?s Adjustment
Genetically Mediated? An Adoption Study,? Developmental Psychology 36 (2000): 429?37.
48. Sara Jaffee and others, ?Why Are Children Born to Teen Mothers at Risk for Adverse Outcomes in Young
Adulthood? Results from a 20-Year Longitudinal Study,? Development and Psychopathology 13 (2001):
C a n F a m i l y S o c i o e c o n o m i c R e s o u r c e s A c c o u n t f o r R a c i a l a n d E t h n i c Te s t S c o r e G a p s ?
VOL. 15 / NO. 1 / SPRING 2005 53
377?97; Judith A. Levine, Harold Pollack, and Maureen Comfort, ?Academic and Behavioral Outcomes
among Children of Young Mothers,? Journal of Marriage and the Family 63 (2001): 355?69; Ruth N. L.
Turley, ?Are Children of Young Mothers Disadvantaged Because of Their Mother?s Age or Family Background??
Child Development 74 (2003): 465?74.
49. O?Connor and others, ?Are Associations between Parental Divorce and Children?s Adjustment Genetically
Mediated?? (see note 47).
50. Jaffee and others, ?Why Are Children Born to Teen Mothers at Risk? (see note 48); Wendy Sigle-Rushton
and Sara McLanahan, ?For Richer or Poorer: Marriage as an Anti-Poverty Strategy in the United States,?
Working Paper 01?17?FF (Princeton University, Center for Research on Child Wellbeing, 2003).
51. M. Robin Dion and others, ?Helping Unwed Parents Build Strong and Healthy Marriages: A Conceptual
Framework for Interventions? (Princeton: Mathematica Policy Research, 2002).
52. Sigle-Rushton and McLanahan, ?For Richer or Poorer? (see note 50).
53. Douglas Kirby, Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy (Washington:
National Campaign to Prevent Teen Pregnancy, 2001).
54. Coley and Chase-Lansdale, ?Adolescent Pregnancy and Parenthood? (see note 45); Andrea Kane and Isabel
V. Sawhill, ?Preventing Teen Childbearing? in One Percent for the Kids, edited by Sawhill (Brookings,
2003), pp. 56?75.
55. Coley and Chase-Lansdale, ?Adolescent Pregnancy and Parenthood? (see note 45).
56. Jeanne Brooks-Gunn, Greg J. Duncan, and J. Lawrence Aber, Neighborhood Poverty: Context and Consequences
for Children, vols. 1 and 2 (New York: Russell Sage, 1997).
57. Robin Jarrett, ?African American Family and Parenting Strategies in Impoverished Neighborhoods,? Qualitative
Sociology 20 (1997): 275?88.
58. Christopher Jencks and Susan Mayer, ?The Social Consequences of Growing Up in a Poor Neighborhood,?
in Inner-City Poverty in the United States, edited by Laurence E. Lynn and Michael G. H. McGeary
(Washington: National Academy Press., 1990), pp. 111?86; Robert J. Sampson, Jeffrey D. Morenoff, and
Thomas Ganon-Rowley, ?Assessing Neighborhood Effects: Social Processes and New Directions in Research,?
Annual Review of Sociology 94 (2002): 774?80.
59. Tama Leventhal and Jeanne Brooks-Gunn, ?The Neighborhoods They Live In: The Effects of Neighborhood
Residence on Child and Adolescent Outcomes,? Psychological Bulletin 126 (2000): 309?37.
60. Larry Orr and others, Moving to Opportunity Interim Impacts Evaluation (U.S. Department of Housing
and Urban Development, Office of Policy Development and Research, 2003), pp. 1?178; Lisa Sanbonmatsu
and others, ?Neighborhoods and Academic Achievement: Results from the Moving to Opportunity
Experiment,? Industrial Relations Section Working Paper (Princeton University, 2004).
61. Sanbonmatsu and others, ?Neighborhoods and Academic Achievement? (see note 60).
62. Orr and others, Moving to Opportunity (see note 60); Jeffrey Kling and others, ?Moving to Opportunity and
Tranquility: Neighborhood Effects on Adult Economic Self-Sufficiency and Health from a Randomized
Housing Voucher Experiment,? Industrial Relations Section Working Paper (Princeton University, 2004).
Greg J. Duncan and Katherine A. Magnuson
54 THE FUTURE OF CHILDREN
63. Whereas all of the accounting studies do a good job of measuring family components of SES, few measure
neighborhood conditions very well. An exception is Phillips and others, ?Family Background, Parenting
Practices, and the Black-White Test Score Gap? (see note 13), whose look at racial gaps at age five includes
an assessment of the role of conditions in the neighborhoods in which the children are raised. While they
find considerable racial differences in neighborhood conditions, these appear inconsequential for the racial
gap. This is not surprising, given that a general finding from the neighborhood effects literature is that
neighborhood conditions add little to the explanation of child outcomes once family conditions are taken
into account; see Brooks-Gunn, Duncan, and Aber, Neighborhood Poverty (see note 56). In the accounting
exercises, this translates into the finding that racial differences in neighborhood conditions account for
little of the gap, once racial differences in family conditions are taken into account.
64. Ronald Fryer and Steven D. Levitt, ?Understanding the Black-White Test Score Gap in the First Two Years
of School,? Review of Economics and Statistics 86 (2004): 447?64.
65. Using the same data, other scholars have reached similar conclusions. See Valerie Lee and David Burkam,
Inequality at the Starting Gate: Social Background Differences in Achievement as Children Begin School
(Washington: Economic Policy Institute, 2002).
66. Phillips and others, ?Family Background, Parenting Practices, and the Black-White Test Score Gap? (see
note 13).
67. Jeanne Brooks-Gunn and others, ?The Black-White Test Score Gap in Young Children: Contributions of
Test and Family Characteristics,? Applied Developmental Science 7 (2003): 239?52.
68. Greg Duncan and Katherine Magnuson, ?Promoting the Healthy Development of Young Children,? in One
Percent for the Kids, edited by Isabel Sawhill (Brookings, 2003), pp. 16?39.
Genetic Differences and School Readiness
William T. Dickens
The author considers whether differences in genetic endowment may account for racial and
ethnic differences in school readiness. While acknowledging an important role for genes in explaining
differences within races, he nevertheless argues that environment explains most of the
gap between blacks and whites, leaving little role for genetics.
Based on a wide range of direct and indirect evidence, particularly work by Klaus Eyferth and
James Flynn, the author concludes that the black-white gap is not substantially genetic in orgin.
In studies in 1959 and 1961, Eyferth first pointed to the near-disappearance of the black-white
gap among children of black and white servicemen raised by German mothers after World War
II. In the author?s view, Flynn?s exhaustive 1980 analysis of Eyferth?s work provides close to definitive
evidence that the black disadvantage is not genetic to any important degree.
But even studies showing an important role for genes in explaining within-group differences,
he says, do not rule out the possibility of improving the school performance of disadvantaged
children through interventions aimed at improving their school readiness. Such interventions,
he argues, should stand or fall on their own costs and benefits. And behavioral genetics offers
some lessons in designing and evaluating interventions. Because normal differences in preschool
resources or parenting practices in working- and middle-class families have only limited
effects on school readiness, interventions can have large effects only if they significantly change
the allocation of resources or the nature of parenting practices.
The effects of most interventions on cognitive ability resemble the effect of exercise on physical
conditioning: they are profound but short-lived. But if interventions make even small permanent
changes in behavior that support improved cognitive ability, they can set off multiplier
processes, with improved ability leading to more stimulating environments and still further improvements
in ability. The best interventions, argues the author, would saturate a social group
and reinforce individual multiplier effects by social multipliers and feedback effects. The aim
of preschool programs, for example, should be to get students to continue to seek out the cognitive
stimulation the program provides even after it ends.
VOL. 15 / NO. 1 / SPRING 2005 55
www.future of children.org
William T. Dickens is a senior fellow in the Brookings Economic Studies program. He acknowledges the excellent research assistance of
Rebecca Vichniac and Jennifer Doleac.
In national tests of school readiness,
black preschoolers in the United
States are not doing as well as white
preschoolers. Researchers find blackwhite
gaps not only in achievement
and cognitive tests, but also in measures of
readiness-related behaviors such as impulse
control and ability to pay attention. Could
some of these differences in school readiness
be the consequence of differences in genetic
endowment? In what follows I will review research
evidence on this question.1
Evidence on the Role
of Genetic Differences
To evaluate the research findings on the role
of genetic differences in cognitive ability, I
begin by drawing a clear distinction between
evidence that genetic endowment explains a
large fraction of differences within races and
evidence that it explains differences between
races and ethnic groups. There can be little
doubt that genetic differences are an important
determinant of differences in academic
achievement within racial and ethnic groups,
though the size of that effect is not known
precisely. Depending on the measure of
achievement used, the sample studied, and
the age of the subjects, estimates of the share
of variance explained by genetic differences
within racial and ethnic groups range from as
low as 20 percent to upward of 75 percent.
However, most estimates, particularly those
for younger children, seem to cluster in the
range of 30 to 40 percent. The fraction of
variance explained by genetic differences in a
population is termed the heritability of the
trait for that population.2
But the heritability of academic achievement
within racial or ethnic groups says little about
whether genes play a role in explaining differences
between racial groups. Suppose one
scatters a handful of genetically diverse seed
corn in a field in Iowa and another in the Mojave
Desert. Nearly all the variance in size
within each group of seedlings could be due
to genetic differences between the plants,
but the difference between the average for
those growing in the Mojave and those growing
in Iowa would be almost entirely due to
their different environments.
If researchers were able to identify all the
genes that cause individual differences in
school readiness, understand the mechanism
by which they affect readiness and the magnitude
of those effects, and assess the relative
frequency of those genes in the black and
white populations, they would know precisely
the extent to which genetic differences explain
the black-white gap. But only a few
genes that influence cognitive ability or other
behaviors relevant to school readiness have
been tentatively identified, and nothing is
known about their frequency in different populations.
Nor are such discoveries imminent.
Although genetic effects on several different
learning and school-related behavior disorders
have been identified and many aspects of
personality are known to have a genetic component,
genes have their primary effect on
school readiness through their effect on cognitive
ability.3 Experts believe that a hundred
or more genes are responsible for individual
differences in cognitive ability. Many of these
genes are likely to have weak and indirect effects
that will be difficult to detect. It could
be decades before enough genes are identified,
and their frequencies estimated, to make
it possible to determine what role, if any, they
play in explaining group differences.
So it is necessary to turn to less direct ways of
answering the question. Much has been written
on this topic in the past fifty years. James
Flynn?s Race, IQ, and Jensen, published in
1980, remains the most thoughtful and thor-
Wi l l i a m T. D i c k e n s
56 THE FUTURE OF CHILDREN
ough treatment available.4 More recently
Richard Nisbett wrote a shorter review of
this literature.5 Both Flynn and Nisbett take
the view, as do I, that genetic differences
probably do not play an important role in explaining
differences between the races, but
the point remains controversial, and Arthur
Jensen provides a recent discussion from a
hereditarian perspective.6 Here I will review
the major types of evidence and explain why
I think they suggest that environmental differences
likely explain most, if not all, of the
black-white gap in school readiness. I will
concentrate entirely on the evidence on cognitive
ability, as it is the most studied trait
that influences school readiness, and genetically
induced differences in cognitive ability
account for the vast majority of genetically
induced differences in school readiness
within ethnic groups. Almost no studies have
been done of racial differences in other traits
G e n e t i c D i f f e r e n c e s a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 57
that might influence school readiness. And I
choose to focus on the black-white gap rather
than to consider the role of genetic differences
in determining the academic readiness
of disadvantaged groups more generally,
again, because it is a topic that has been more
thoroughly studied.
Direct Evidence on the Role
of Genes: European Ancestry
and Cognitive Ability
Blacks in the United States have widely varying
degrees of African and European ancestry.
If their genetic endowment from their
African ancestors is, on average, inferior to
that from their European ancestors, then
their cognitive ability would be expected to
vary directly in proportion to the extent of
their European ancestry. Some early attempts
to assess this hypothesis linked skin
color with test scores and found that lighter-
Clearing Up a Confusion
It is difficult to discuss genetic causation of the black-white test score gap. The reason, I believe,
is that people confuse genetic causation with intractability. Suppose that the entire black-white
gap in school readiness were genetic in origin, but that a shot could be given to black babies at
birth to offset completely the effects of the genetic difference. Would anyone care about the genetic
component of the racial gap? If it is possible to remedy or ameliorate the black-white difference,
the only question is how much it would cost and whether society is willing to pay the price.
As this article explains, genetic causation is nearly irrelevant to the question of how malleable a
trait is.
Some argue that a genetic cause for black-white differences would lessen the moral imperative
for removing them, but as the example of the shot illustrates, this is not the case. It would be
hard to argue that the fact that the differences were genetic rather than environmental in origin
would make it any less of an imperative for society to be sure that every black child got the shot.
Some would say that the fact that the cause is beyond the child?s control would make it more
important. Jessica L. Cohen and I have made this argument in more detail in ?Instinct and
Choice: A Framework for Analysis,? in Nature and Nurture: The Complex Interplay of Genetic and
Environmental Influences on Human Behavior and Development, edited by Cynthia Garcia Coll,
Elaine L. Bearer, and Richard Lerner (Hillsdale, N.J.: Lawrence Erlbaum and Associates, 2003),
pp. 145?70.
skinned blacks typically had higher scores.
But skin color is not strongly related to degree
of European ancestry, while socioeconomic
status clearly is. Thus the differences
might reflect environmental rather than genetic
causes. Nearly all commentators agree
that these early studies are not probative.
More recent studies have looked at measures
of European ancestry, such as blood groups
or reported ancestry, that are not visible.
Such studies have found little or no correlation
between the measure of ancestry and
cognitive ability, though all are subject to
methodological criticisms that could explain
their failure to find such a link. Thus although
these studies do not provide evidence
for a role for genes in explaining black-white
differences, they do not provide strong evidence
against it.
Direct Evidence on the Role
of Environment: Adoption and
Cross-Fostering
If there is no direct evidence of a role for
genes in explaining the black-white gap, perhaps
there is direct evidence that environment
can or cannot account for the whole
difference between blacks and whites. Several
studies have shown that environmental
differences between blacks and whites can, in
a statistical sense, ?explain? nearly all of the
difference in cognitive ability between black
and white children.7 But because the studies
do not completely control for the genetic endowment
of either the child or the parents
and because many of the variables used to explain
the difference are themselves subject to
genetic influence, the effect being attributed
to environment may in reality be due to genetic
differences.
What is needed is a way to see the effect of
environment without confusing it with the effect
of genetic endowment. For example,
randomly choosing white and black children
at birth and assigning them to be fostered in
either black or white families would ensure
that the children?s environments were not
correlated with their genetic potential and
would show how much difference environment
makes. No existing study replicates the
conditions of this experiment exactly, but
some come close. The strongest evidence for
both the environmentalist and hereditarian
perspectives is of this sort.
After the end of World War II both black and
white soldiers in the occupying armies in
Germany fathered children with white German
women. Klaus Eyferth gathered data on
a large number of these children, of mainly
working-class mothers, and gave the children
intelligence tests.8 He found almost no difference
between the children of white fathers
and those of black fathers. The finding
is remarkable given that the black children
faced a somewhat more hostile environment
than the white children. Hereditarians have
challenged these findings by appealing to the
possibility that the black soldiers who fathered
these children might have been a particularly
elite group. Flynn has researched
the plausibility of this explanation and concludes
that such selection did not play more
than a small role.9 Thus Eyferth?s study suggests
that the black-white gap is largely, and
possibly entirely, environmental.
A study similar to Eyferth?s found the cognitive
ability of black children raised in an orphanage
in England to be slightly higher than
that of white children raised there.10 Again,
critics have raised the possibility that the black
children were genetically advantaged relative
to other blacks, and the whites disadvantaged
relative to other whites. And again, Flynn
finds it unlikely that this contention explains
Wi l l i a m T. D i c k e n s
58 THE FUTURE OF CHILDREN
much of the disappearance of the black-white
gap.11 This study, too, suggests that the blackwhite
gap is mainly environmental.
If the black-white gap is mainly genetic in
origin, children?s cognitive ability should not
depend on the race of their primary caregiver,
comparing those of the same race. Yet
two studies comparing the experience of
black children raised by black or white mothers
suggest that it does.12 Here too, because
the children were not randomly assigned to
their caregivers, it is possible that the children
raised by black mothers were of lower
genetic potential, but it would be hard to
make such a selection story explain more
than a small fraction of the apparent environmental
effect.
Another transracial adoption study provides
mixed evidence, but some of the strongest
that genes play a role in explaining the blackwhite
gap.13 A group of children, some with
two black parents and some with one white
and one black parent, were raised in white
middle-class families. When the children?s
cognitive ability was tested at age seven, the
children with two black parents scored 95,
higher than the average black child in the
state (89) and only slightly below the national
average for whites, while the mixed-race children
scored 110, which was considerably
above it.14 On the one hand, this finding suggests
a huge effect of environment on the
cognitive ability of the adopted black and
mixed-race children. On the other hand, the
higher scores of the mixed-race children suggest
that parents? genes may account for
some of the difference from the black children,
and that the mixed-race children may
have had a better inheritance by virtue of
having one white parent. Both black and
mixed-race children scored worse than the
biological children of their adoptive parents
(who scored 116), an expected finding because
the adopting parents were an elite
group and likely passed on above-average genetic
potential to their children. But they also
scored considerably below the average of 118
for comparison white children adopted into
similar homes.
When the same children were retested ten
years later, the results were different.15 The
scores of the children with two black parents
had dropped to about the average for blacks
in the state where they lived before they
were adopted (89). The scores of the mixedrace
children had dropped too (99), but remained
intermediate between those of the
children with two black parents and those of
the adoptive parents? biological children,
which had also declined, to 109. The scores
of the white children raised in adoptive
homes had dropped the most, falling to 106.
The disappearance of the salutary effect of
the adoptive home, however, does not mean
that genes determine black-white differences.
We can assume that as the children
aged and moved out into the world, the effect
of the home environment diminished, and
both whites and blacks tended to the average
for their own population because of either
genetic or environmental effects. By showing
how the effect of a child?s home environment
disappears by adolescence, this study sug-
G e n e t i c D i f f e r e n c e s a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 59
The disappearance of the
salutary effect of the adoptive
home, however, does not
mean that genes determine
black-white differences.
gests that environmental disadvantages experienced
by blacks as children cannot explain
the deficit in their cognitive ability as adolescents
and adults. But environmental disadvantages
facing black adolescents and adults
could still explain those deficits. The transience
of environmental effects on cognitive
ability is a theme to which I shall return. The
persistence of the advantage of the mixedrace
children over the children with two
black parents is suggestive of a role for genes.
It is not, though, definitive: several other explanations
have been offered, including the
late adoption of the children with two black
parents and parental selection effects unrelated
to race.16
Indirect Evidence on the Role
of Genetic Differences
Although the direct evidence on the role of
environment is not definitive, it mostly suggests
that genetic differences are not necessary
to explain racial differences. Advocates
of the hereditarian position have therefore
turned to indirect evidence.17
Several authors have argued that estimates of
the heritability of cognitive ability put limits
on the plausible role of environment.18 The
argument is normally made in a mathematical
form, but it boils down to this. First, it is now
widely accepted that differences in genetic
endowment explain at least 60 percent of the
variance in cognitive ability among adults in
the white population in the United States.19 If
all the environmental variation among U.S.
whites can explain only 40 percent of the variance
among whites, how could environmental
differences explain the huge gap between
blacks and whites? The mathematical argument
implies that the average black environment
would have to be worse than at least 95
percent of white environments, but observable
characteristics of blacks and whites are
not that different. For example, black deficits
in education or in socioeconomic status place
the average black below only about 60 to 70
percent of whites.20
The heritability of cognitive ability is also
crucial to a second type of indirect evidence
for a role of genetic differences in explaining
the black-white gap. Arthur Jensen has advanced
what he calls ?Spearman?s Hypothesis,?
after the late intelligence researcher
Charles Spearman, who observed that people
who had large vocabularies were good at solving
mazes and logic problems and were also
more likely to have command of a wide range
of facts. Spearman posited that a single,
largely genetic, mental ability that he called g
(for general mental ability) explained the correlation
of people?s performance across a
wide range of tests of mental ability. Researchers
now know that a single underlying
ability cannot explain all the tendency of people
who do well on one type of test to do well
on another.21 But it is possible to interpret
the evidence as indicating that there is a single
ability that differs among people, that is
subject to genetic influence, and that explains
much of the correlation across tests. Other
interpretations are also possible, but this one
cannot be discounted. In a series of studies
Jensen and Rushton have argued that different
types of tests tap this general ability to
different degrees; that the more a test taps g,
the more it is subject to genetic influence;
and that black-white differences are largest
on the tests most reflective of the underlying
general ability, g.22
Using several restrictive assumptions about
the nature of genetic and environmental influence
on genetic ability, researchers can
use this information to estimate the fraction
of the black-white gap that is due to differences
in genetic endowment. The more the
Wi l l i a m T. D i c k e n s
60 THE FUTURE OF CHILDREN
pattern of black-white differences across different
tests resembles the pattern of genetic
influence on different tests, the more the statistical
procedure will attribute the blackwhite
differences to genetic differences.
Using this method, David Rowe and Jensen
have independently estimated that from onehalf
to two-thirds of the black-white gap is
genetic in origin.23
A Problem for the Indirect
Arguments: Gains in Cognitive
Ability over Time
Over the past century, dozens of countries
around the world have seen increases in
measured cognitive ability over time as large
as or even larger than the black-white gap.24
The phenomenon has been christened the
?Flynn Effect,? after James Flynn, who did
the most to investigate and popularize this
worldwide trend. The score gains have been
documented even between a large group of
fathers and sons taking the same test only
decades apart, making it impossible that the
gains are due to changes in genes. Clearly environmental
changes can cause huge leaps in
measured cognitive ability. Although it might
not seem plausible that the average black environment
today is below the 5th percentile of
the white distribution of environments, it is
certainly plausible that the average black environment
in the United States today is as deprived
as the average white environment of
thirty to fifty years ago?the time it took for
cognitive ability to rise by an amount equal to
the black-white gap in many countries. These
gains in measured cognitive ability over time
point to a problem in the argument that high
heritability estimates for cognitive ability preclude
large environmental effects.
Gains in cognitive ability over time also challenge
the logic of Jensen?s genetic explanation
for the pattern of black-white differences
across different types of tests. All studies
show that gains on different tests are positively
correlated with measures of test score
heritability, and most studies show that gains
are positively correlated with the extent to
which a test taps the hypothesized general
cognitive ability.25 There is little doubt that
applying the same method as Rowe and
Jensen used to data on gains in cognitive ability
over time would show them to be partially
genetic in origin, something we know cannot
be true.
So, what is it that is wrong with the logic of
these two arguments, that the high heritability
of cognitive ability limits the possible effect
of the environment and that the pattern
of black-white differences across different
tests shows those differences to be genetic in
origin? And in particular, where is the problem
in the first?
It is important to detect the flaw, because if
the logic of the argument were sound, the
case for environmental causes of black-white
differences would be difficult to make, and
the possibility of remedying those differences
would be remote. But before I explain, I
want to cite two other pieces of evidence
marshaled by advocates of the hereditarian
position that suggest the limited power of the
environment to change cognitive ability (and
therefore to explain the entire black-white
gap). The first is that the heritability of cognitive
ability rises with age. It does so at the expense
of the effect of family environment,
which disappears nearly completely in most
studies of late adolescents and adults.26 The
disappearance of the effect on black children
of being raised in white families, which I
have already noted, is just one case of a general
finding from several different types of
studies. A second piece of evidence is the
fade-out of the effect of preschool programs
G e n e t i c D i f f e r e n c e s a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 61
on cognitive ability. Although such programs
have been shown to have profound effects on
the measured ability of children, the effects
fade once the programs end, leaving little evidence
of any effect by adolescence.27 Is it
possible to reconcile the high heritability of
cognitive ability with large, but transient, environmental
effects?
The Interplay of Genes
and the Environment
To explain this puzzle, James Flynn and I
have proposed a formal model in which
genes and environment work together, rather
than independently, in developing a person?s
cognitive ability.28 The solution involves
three aspects of the process by which individual
ability is molded that are overlooked by
the logic that implies small environmental effects.
We illustrate our argument with a basketball
analogy.
How can genes and environment both be
powerful in shaping ability? Consider a
young man with a small genetic predisposition
toward greater height and faster reflexes.
When he is young, he is likely to be slightly
better than his playmates at basketball. His
reflexes will make him generally better at
sports, and his height will be a particular advantage
when it comes to passing, catching,
and rebounding. These advantages by themselves
confer only a small edge, but they may
be enough to make the game more rewarding
for him than for the average person and get
him to play more than his friends and to improve
his play more over time. After a while,
he will be considerably better than the average
player his age, making it likely that he
will be picked first for teams and perhaps receive
more attention from gym teachers.
Eventually, he joins a school team where he
gets exhaustive practice and professional
coaching. His basketball ability is now far superior
to that of his old playmates. Through a
series of feedback loops, his initial minor
physical advantage has been multiplied into a
huge overall advantage. In contrast, a child
who started life with a predisposition to be
pudgy, slow, and small would be very unlikely
to enjoy playing basketball, get much practice,
or receive coaching. He would therefore
be unlikely to improve his skills. Assuming
children with a range of experience between
these two extremes, scientists would find that
a large fraction of the variance of basketball
playing ability would be explained by differences
in genetic endowment?that basketball
ability was highly heritable. And they would
be right to do so. But that most certainly
would not mean that short kids without lightning
reflexes could not improve their basketball
skills enormously with practice and
coaching.
The basketball analogy so far illustrates two
of the considerations that Flynn and I believe
are important for understanding the implications
of behavioral genetic studies of cognitive
ability. First, genes tend to get matched
to complimentary environments. When that
happens, some of the power of environment
is attributed to genes. Only effects of environment
shared by all children in the same
family and effects of environment uncorrelated
with genes get counted as environmental.
Second, the effect of genetic differences
gets multiplied by positive feedback loops.
Small initial differences are multiplied by
processes where people?s initially varying
abilities are matched to complimentary environments
that cause their abilities to diverge
further.
In theory this same multiplier process could
be driven by small environmental differences.
But to drive the multiplier to its maximum,
the environmental advantage would
Wi l l i a m T. D i c k e n s
62 THE FUTURE OF CHILDREN
have to be as constant over time as the genetic
difference, because in the absence of
the initial advantage there will be a tendency
for the whole process to unwind. For example,
suppose that midway through high
school the basketball enthusiast injures a leg,
which makes him less steady and offsets his
initial advantage in height and reflexes. Because
of all his practice and learning, he will
still be a superior player. But his small decrement
in performance could mean discouragement,
more bench time, or not making
the cut for the varsity team. This could lead
to a further deterioration of his skills and further
discouragement, until he gives up playing
on the team entirely. Although each individual?s
experience will differ, the theory that
Flynn and I lay out would have people with
average physical potential reverting to average
ability over time, on average.
The transitory nature of most environmental
effects not driven by genetic differences
helps explain why environmental differences
do not typically drive large multipliers and
produce the same large effects as genetic differences.
That same transience helps explain
why environment can be potent but still
cause a relatively small share of the variance
of cognitive ability in adults.29
Social Multipliers and
the Effect of Averaging
If most external environmental influences are
transitory and transitory environmental effects
are unable to drive multipliers, what explains
the large gains in cognitive ability over
the past century? That question has two answers.
One is the social multiplier process.
The other is that many random transient environmental
effects that lean in one direction
when averaged together can substitute for a
single persistent environmental cause. This is
the third point missed by the argument that
claims that high heritability implies small environmental
effects.
Another basketball analogy will help explain
social multipliers. During the 1950s television
entered many U.S. homes. Professional
basketball, with its small arena, could not
reach as wide an audience as baseball, but
basketball translated much better to the
small screen. Thus public interest in basketball
began to grow. The increased interest
made it easier for enthusiasts to find others
to play with, thus increasing the opportunities
to improve skills. As skills improved,
standards of play rose, with players learning
moves and skills from each other. As more
people played and watched the game, interest
increased still further. More resources
were devoted to coaching basketball and developing
basketball programs, providing yet
more opportunities for players to improve
their skills. In the end, the small impetus provided
by the introduction of television had a
huge impact on basketball skills.
A similar process may well be at work for cognitive
ability. An outpouring of studies in recent
years suggests that social effects have an
important influence on school performance.30
One study of an experimental reduction in
school class size resulting in major achievement
score gains suggests that a very large
fraction of the gains came through the children?s
extended association with their peers,
who shared the experience of small class
sizes.31 In this case an arguably minor intervention
had large and long-lasting effects
largely owing to a social multiplier effect.
But improvements in cognitive ability could
have many triggers, rather than a single one.
Many such triggers over the past half-century
averaged together could be acting to raise
cognitive ability. Increasing cognitive de-
G e n e t i c D i f f e r e n c e s a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 63
mands from more professional, technical,
and managerial jobs; increased leisure time;
changing cognitive demands of personal interactions;
or changing attitudes toward intellectual
activity could all be playing a role.
And small initial changes along any of these
dimensions would be magnified by individual
and social multipliers.
Genes and Environment
and the Black-White Gap
The black-white gap in measured cognitive
ability may come about in a similar way, but it
could have even more triggers. Segregation
and discrimination have caused many aspects
of blacks? environment to be inferior to that
of whites. Averaged together, the total impact
can be large, even if each individual effect is
small. Suppose, for example, that environment
relevant to the formation of cognitive
ability consists of 100 factors, each with an
equal effect. If for each of these 100 factors
the average black were worse off than 65 percent
of whites, he would be worse off than 90
percent of whites when the effects of all the
environmental factors were considered together.
(The disparity is the necessary result
of accumulating a large number of effects
when two groups have slightly different
means for all the effects.)32 Taking the total
effect of environment in this way, considering
the underestimate of the total effect of environment
because some of its power is attributed
to genes, and considering individual and
social multipliers, a purely environmental explanation
for black-white differences becomes
plausible despite high estimates for
the heritability of cognitive ability.
Moreover, our model also has explanations
for the correlation of the heritability of scores
on different tests with the size of the blackwhite
gap on those tests and the anomalous
correlation of the size of gains in cognitive
ability over time on different tests with the
heritability of those test scores. Those cognitive
abilities for which multiplier processes
are most important will be the ones that show
the largest heritability, because of the environmental
augmentation of the genetic differences.
But they will also be the ones on
which a persistent change in environment
will have the biggest influence. Thus we
might expect that persistent environmental
differences between blacks and whites, as
well as between generations, could cause a
positive correlation between test score heritabilities
and test differences.33 Rushton and
Jensen?s indirect evidence of a genetic role in
black-white differences is, therefore, not
probative.
Implications and Conclusions
The indirect evidence on the role of genes in
explaining the black-white gap does not tell
us how much of the gap genes explain and
may be of no value at all in deciding whether
genes do play a role. Because the direct evidence
on ancestry, adoption, and crossfostering
is most consistent with little or no
role for genes, it is unlikely that the blackwhite
gap has a large genetic component.
But what if it does? What would be the implications
for the school readiness of children?
Much of the variance in human behavior, including
cognitive ability and achievement
test scores, can be traced to differences in individuals?
genetic endowments. But as indisputable
as is the role of genes in shaping differences
in outcomes within races, so is the
role of environment. Studies of young children
show that environmental differences explain
more variation than do genetic differences.
And even studies showing an
important role for genes in no way rule out
the possibility of improving the school performance
of disadvantaged children through
Wi l l i a m T. D i c k e n s
64 THE FUTURE OF CHILDREN
interventions aimed at enhancing their
school readiness. Interventions should stand
or fall on their own costs and benefits and not
be prejudged on the basis of genetic
pessimism.
In fact, studies of the role of genes and environment
in determining school readiness
offer some useful lessons in designing and
evaluating interventions. These studies show
that normally occurring differences in preschool
resources or parenting practices in
working- and middle-class families have only
limited effects on school readiness once the
correlation due to parents? and children?s
genes is taken out of play.34 Thus small interventions
that make only modest changes in
the allocation of resources or the nature of
parenting practices will have limited to modest
effects at best. Effects will likely be somewhat
larger if interventions target very disadvantaged
families, probably because the
room for improvement is greater.35
Achieving permanent effects on cognitive
ability is harder than achieving large effects.
Most environmental effects on cognitive ability
seem to be like the effect of exercise on
physical conditioning: profound but shortlived.
But even short-lived improvements in
cognitive ability can be valuable if they mediate
longer-term changes in achievement?for
example, if improved cognitive ability for
some period of time allows students to learn
to read more quickly, putting them on a permanently
higher achievement path. And evidence
suggests that programs aimed at improving
cognitive ability do have long-term
effects on achievement even if they have no
significant long-term effects on cognitive
ability. However, if interventions make even
small permanent changes in behavior that
support improved cognitive ability, they can
set off multiplier processes, with improved
ability leading to better environments and
still further improvements in ability. If we
knew what aspects of preschool programs
help elevate cognitive ability, and if we could
get children to continue to seek out such
stimulation after they leave preschool programs,
their increased ability could lead them
to associate with more able peers, to have the
confidence to take on more demanding academic
challenges, and to get the further advantage
of yet more positive stimulation from
these activities. This, in turn, could further
develop their cognitive ability. Long-lived effects
are more likely to be large effects.
Effects are particularly likely to be large if an
intervention saturates a social group and allows
the individual multiplier effects to be reinforced
by social multipliers or feedback effects.
If students find themselves among
others with greater ability, individual interactions
and group activities are more likely to
give rise to further improvements in cognitive
ability. In this same vein, evaluations that
do not take into account the social effects of
the intervention on children who did not directly
take part may be missing an important
aspect of the effects of an intervention.
Although much of normal environmentally
induced variance in cognitive ability seems to
be transient, if interventions could induce
even small long-lasting changes in behavior,
they might produce very large effects
through the multiplier process. Taking advantage
of such processes may make it possible
to overcome the black-white gap and put
black and white children on an even footing.
G e n e t i c D i f f e r e n c e s a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 65
Endnotes
1. The review necessarily highlights only the most important studies; a complete review of all the arguments
on both sides of this debate would require hundreds of pages and be beyond the scope of this article.
2. Heritability is estimated by examining the similarity of people with different degrees of genetic similarity
raised in similar sorts of environments, and there is some reason to believe that most estimates are somewhat
overstated by existing methods. Robert Plomin and others, Behavioral Genetics, 4th ed. (New York:
Worth Publishers, 2001), in chapter 5 and the appendix, provide a thorough discussion of the methods used
to estimate heritability. Mike Stoolmiller, ?Implications of the Restricted Range of Family Environments
for Estimates of Heritability and Nonshared Environment in Behavior-Genetic Adoption Studies,? Psychological
Bulletin 125 (1999): 392?409, shows that adoption studies probably overstate the degree of heritability
and speculates on reasons why some other methods may as well.
3. Robert Plomin and others, Behavioral Genetics (see note 2), examine learning disorders on pp. 145?49,
ADHD on pp. 227?29, and personality in chapter 12. For the effects of genes on cognitive ability, see Marcie
L. Chambers and others, ?Variation in Academic Achievement and IQ in Twin Pairs,? Intelligence
(forthcoming); Lee Anne Thompson and others, ?Associations between Cognitive Abilities and Scholastic
Achievement: Genetic Overlap but Environmental Differences,? Psychological Science 2 (1991): 158?65;
and Sally J. Wadsworth, ?School Achievement,? in Nature and Nurture during Middle Childhood, edited
by John C. DeFries, Robert Plomin, and David W. Fulker (Oxford: Blackwell, 1994), pp. 86?101.
4. James R. Flynn, Race, IQ, and Jensen (London: Routledge, 1980).
5. Richard Nisbett, ?Race, Genetics, and IQ,? in The Black-White Test Score Gap, edited by Christopher
Jencks and Meredith Phillips (Brookings, 1998), pp. 86?102.
6. Arthur Jensen, The g Factor (Westport, Conn.: Praeger, 1998), pp 350?531.
7. Jane R. Mercer, ?What Is a Racially and Culturally Nondiscriminatory Test? A Sociological and Pluralistic
Perspective,? in Perspectives on ?Bias in Mental Testing,? edited by Cecil R. Reynolds and Robert T. Brown
(New York: Plenum Press, 1984); Jonathan Crane, ?Race and Children?s Cognitive Test Scores: Empirical
Evidence That Environment Explains the Entire Gap,? mimeo, University of Illinois at Chicago, 1994; and
Jeanne Brooks-Gunn and others, ?Ethnic Differences in Children?s Intelligence Test Scores: Role of Economic
Deprivation, Home Environment, and Maternal Characteristics,? Child Development 67, no. 2
(1996): 396?408.
8. This is based on the account by James R. Flynn (Race, IQ, and Jensen, pp. 84?87; see note 4) of Klaus
Eyferth, ?Leistungen verschiedener Gruppen von Besatzungskindern in Hamburg-Wechsler Intelligenztest
fur Kinder (HAWIK),? Archiv fur die gesamte Psychologie 113 (1961): 222?41.
9. Flynn, Race, IQ, and Jensen, pp. 84?102 (see note 4).
10. Barbara Tizard, ?IQ and Race,? Nature 247, no. 5439 (February 1, 1974).
11. Flynn, Race, IQ, and Jensen, pp. 108?11 (see note 4).
12. Elsie G. J. Moore, ?Family Socialization and the IQ Test Performance of Traditionally and Transracially
Adopted Black Children,? Developmental Psychology 22 (1986): 317?26; and Lee Willerman and others,
?Intellectual Development of Children from Interracial Matings: Performance in Infancy and at 4 Years,?
Behavioral Genetics 4 (1974): 84?88.
Wi l l i a m T. D i c k e n s
66 THE FUTURE OF CHILDREN
13. Sandra Scarr and Richard A. Weinberg, ?IQ Test Performance of Black Children Adopted by White Families,?
American Psychologist 31 (1976): 726?39; and Sandra Scarr and Richard A. Weinberg, ?The Minnesota
Adoption Studies: Genetic Differences and Malleability,? Child Development 54 (1983): 260?67.
14. These are IQ scores, which have a mean of 100 and a standard deviation of 15 in the U.S. population.
15. Sandra Scarr and others, ?The Minnesota Transracial Adoption Study: A Follow-Up of IQ Test Performance
at Adolescence,? Intelligence 16 (1992): 117?35.
16. But see Arthur Jensen, The g Factor, pp. 477?78 (see note 6), on whether late adoption can explain the
difference.
17. One body of evidence is difficult to judge. See J. Philippe Rushton, Race, Evolution, and Behavior: A Life
History Perspective, 3rd ed. (Port Huron, Mich.: Charles Darwin Research Institute, 2000). Rushton has
proposed a theoretical framework that would explain a genetic gap in cognitive ability between blacks and
whites and has marshaled evidence for it. But because much of the evidence was known before the theory
was proposed, some view the theory as nothing more than post hoc rationalization for hereditarian views on
the black-white gap. At most it suggests that some of the black-white gap may be genetic, but it does not
suggest how much.
18. Arthur Jensen, Educability and Group Differences (New York: Harper and Row, 1973), pp. 135?39,
161?73, 186?90; Arthur Jensen, Educational Differences (London: Methuen, 1973), pp. 408?12; Jensen,
The g Factor, pp. 445?58 (see note 6); and Richard Herrnstein and Charles Murray, The Bell Curve: Intelligence
and Class Structure in American Life (New York: Simon and Schuster, 1994), pp. 298?99.
19. Plomin and others, Behavioral Genetics, p. 177 (see note 2); and Ulric Neisser and others, ?Intelligence:
Knowns and Unknowns,? American Psychologist 51, no. 2(1996): 85.
20. Author?s calculations from the 1979 National Longitudinal Survey of Youth.
21. John B. Carol, Human Cognitive Abilities: A Survey of Factor-Analytic Studies (Cambridge University
Press, 1993), is the most comprehensive survey of what is known about the correlation of scores on different
types of mental tests.
22. See J. Philippe Rushton and Arthur Jensen, ?Thirty Years of Research on Race Differences in Cognitive
Ability,? Psychology, Public Policy, and Law (forthcoming), for a review of this evidence and citations to
the original studies.
23. David Rowe, Alexander Vazsonyi, and Daniel Flannery, ?Ethnic and Racial Similarity in Developmental
Process: A Study of Academic Achievement,? Psychological Review 101, no. 3 (1994): 396?413; Jensen, The
g Factor, pp. 464?67 (see note 6).
24. James R. Flynn, ?Massive Gains in 14 Nations: What IQ Tests Really Measure,? Psychological Bulletin 101
(1987): 171?91; James R. Flynn, ?IQ Gains over Time,? in Encyclopedia of Human Intelligence, edited by
Robert J. Sternberg (New York: Macmillan, 1994), pp. 617?23; James R. Flynn, ?IQ Gains over Time: Toward
Finding the Causes,? in The Rising Curve: Long-Term Gains in IQ and Related Measures, edited by
Ulric Neisser (Washington: American Psychological Association, 1998), pp. 551?53.
25. Existing evidence suggests that IQ gains across subtests are probably positively correlated with g loading.
See Roberto Colom, Manuel Juan-Espinosa, and Lu? F. Garc?, ?The Secular Increase in Test Scores Is a
G e n e t i c D i f f e r e n c e s a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 67
?Jensen effect,?? Personality and Individual Differences 30 (2001): 553?58; and Manuel Juan-Espinosa and
others, ?Individual Differences in Large-Spaces Orientation: g and Beyond?? Personality and Individual
Differences 29 (2000): 85?98, for much stronger correlations between g loadings and IQ gains. Jensen, The
g Factor, pp. 320?21 (see note 6), reviews a number of studies of the relation between subtests gains and g
loadings, all of which show weak positive correlations. J. Philippe Rushton, ?Secular Gains in IQ Not Related
to the g Factor and Inbreeding Depression?unlike Black-White Differences: A Reply to Flynn,?
Personality and Individual Differences 26 (1999): 381?89, finds that a measure of g developed on the
Wechsler Intelligence Scale for Children has loadings that are negatively correlated with subtest gains in
several countries. But see James R. Flynn, ?The History of the American Mind in the 20th Century: A Scenario
to Explain IQ Gains over Time and a Case for the Irrelevance of g,? in Extending Intelligence: Enhancement
and New Constructs, edited by P. C. Kyllonon, R. D. Roberts, and L. Stankov (Hillsdale, N.J.:
Erlbaum, forthcoming.) for an argument that IQ gains are greatest on tests of fluid g rather than crystallized
g. He finds a positive (though statistically insignificant) correlation between a measure of fluid g he
develops and IQ gains in the data used by Rushton. Olev Must, Aasa Must, and Vilve Raudik, ?The Flynn
Effect for Gains in Literacy Found in Estonia Is Not a Jensen Effect,? Personality and Individual Differences
33 (2001); and Olev Must, Aasa Must, and Vilve Raudik, ?The Secular Rise in IQs: In Estonia the
Flynn Effect Is Not a Jensen Effect,? Intelligence 31 (2003): 461?71, find no correlation between g loadings
and gains on two tests in Estonia, but these are achievement tests with a strong crystallized bias.
26. Plomin and others, Behavioral Genetics, pp. 173?77 (see note 2).
27. Irving Lazar and Richard Darlington, ?Lasting Effects of Early Education: A Report from the Consortium
for Longitudinal Studies,? Monographs of the Society for Research in Child Development 47, nos. 2?3
(1982).
28. William T. Dickens and James Flynn, ?Heritability Estimates versus Large Environmental Effects,? Psychological
Review 108, no. 2 ( 2001).
29. This is not to say that there are no permanent or long-lasting environmental effects on cognitive ability. The
effects of brain damage can be severe and permanent. However, such permanent environmental effects evidently
explain only a small fraction of normal variation in cognitive ability. Shared family environment
plays a large role in explaining variance in cognitive ability when children are spending most of their time
in the home, with their activities strongly influenced by their parents. But that effect fades as they spend
more of their time away from home and in self-directed activities.
30. Eric A. Hanushek and others, ?Does Peer Ability Affect Student Achievement?? Working Paper 8502
(Cambridge, Mass.: National Bureau of Economic Research, 2001); Caroline Hoxby, ?Peer Effects in the
Classroom: Learning from Gender and Race Variation,? Working Paper 7867 (Cambridge, Mass.: National
Bureau of Economic Research, 2001); Dan M. Levy, ?Family Income and Peer Effects as Determinants of
Educational Outcomes,? Ph.D. diss., Northwestern University, 2000; Donald Robertson and James
Symons, ?Do Peer Groups Matter? Peer Group versus Schooling Effects on Academic Achievement,? Economica
70 (2003): 31?53; Bruce Sacerdote, ?Peer Effects with Random Assignment: Results from Dartmouth
Roommates,? Quarterly Journal of Economics (May 2001): 681?704; David J. Zimmerman, ?Peer
Effects in Academic Outcomes: Evidence from a Natural Experiment,? Review of Economics and Statistics
85 (2003): 9?23.
Wi l l i a m T. D i c k e n s
68 THE FUTURE OF CHILDREN
31. Michael A. Boozer and Stephen E. Cacciola, ?Inside the ?Black Box? of Project STAR: Estimation of Peer
Effects Using Experimental Data,? Discussion Paper 832 (Economic Growth Center, Yale University, 2001).
32. In statistics this is referred to as the law of large numbers?that the variance of a mean falls as the number
of items being averaged goes up. See Eugene Lukacs, Probability and Mathematics Statistics: An Introduction
(New York: Academic Press, 1972). It applies whether or not the weights being put on the elements
are equal. Because the variance and standard deviation of the mean fall, while the average difference stays
the same, the difference in standard deviations grows. The example assumes that the effects are all uncorrelated
with each other and that each has a normal distribution in the white and the black populations. If
the effects were assumed to be correlated or the weights unequal, the results would be less dramatic, but
with observed values for correlations of environmental factors, increasing the number of items to be averaged
could produce the same results.
33. Dickens and Flynn, ?Heritability Estimates vs. Large Environmental Effects? (see note 28).
34. Plomin and others, Behavioral Genetics, p. 201 (see note 2).
35. Eric Turkheimer and others, ?Socioeconomic Status Modifies Heritability of IQ in Young Children,? Psychological
Science 14, no. 6 (2003). Their own study finds that shared family environment explains 60 percent
of the variance of an IQ test score in low-socioeconomic-status seven-year-olds, which is a much larger
share than other studies have found. For example, see Kathryn Asbury and others, ?Environmental Moderators
of Genetic Influence on Verbal and Nonverbal Abilities in Early Childhood? (Institute of Psychiatry,
De Crespigny Park, London, 2004).
G e n e t i c D i f f e r e n c e s a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 69
Neuroscience Perspectives on
Disparities in School Readiness
and Cognitive Achievement
Kimberly G. Noble, Nim Tottenham, and B. J. Casey
Summary
This article allows readers to look at racial and ethnic disparities in school readiness from a
neuroscience perspective. Although researchers have traditionally measured gaps in school
readiness using broad achievement tests, they can now assess readiness in terms of more specific
brain-based cognitive functions. Three neurocognitive systems?cognitive control, learning
and memory, and reading?are essential for success in school. Thanks to recent advances in
brain imaging, it is now possible to examine these three systems, each located in specific areas
of the brain, by observing them in action as children engage in particular tasks.
Socioeconomic status?already linked with how well children do on skills tests generally?is
particularly closely linked with how well they perform on tasks involving these crucial neurocognitive
systems. Moreover, children?s life experiences can influence their neurocognitive
development and lead to functional and anatomical changes in their brains. Noting that chronic
stress or abuse in childhood can impair development of the brain region involved in learning
and memory, the authors show how the extreme stress of being placed in an orphanage leads to
abnormal brain development and decreased cognitive functioning.
More optimistically, the authors explain that children?s brains remain plastic and capable of
growth and development. Targeted educational interventions thus have the promise of improving
both brain function and behavior. Several such interventions, for example, both raise children?s
scores in tests of reading and increase activity in the brain regions most closely linked
with reading. The brain regions most crucial for school readiness may prove quite responsive to
effective therapeutic interventions?even making it possible to tailor particular interventions
for individual children. The authors look ahead to the day when effective educational interventions
can begin to close racial and socioeconomic gaps in readiness and achievement.
VOL. 15 / NO. 1 / SPRING 2005 71
www.future of children.org
Kimberly G. Noble is an M.D./Ph.D. candidate at the Center for Cognitive Neuroscience, University of Pennsylvania, and a visiting graduate
student at the Sackler Institute for Developmental Psychobiology of Weill Medical College of Cornell University. Nim Tottenham is a Ph.D.
candidate at the Institute of Child Development, University of Minnesota, and a visiting graduate student at the Sackler Institute for Developmental
Psychobiology. B. J. Casey is the director of the Sackler Institute for Developmental Psychobiology and the Sackler Professor of
Psychobiology at Weill Medical College.
Racial disparities in school readiness
among America?s preschoolers
are strong and persistent.
As elaborated elsewhere in
this volume, many aspects of
childhood experience, including health, parenting,
stress, violence, and access to resources,
contribute to these disparities. Many
of these same experiences, including chronic
stress and cognitive stimulation, also affect
brain development in both animals and humans,
suggesting a possible pathway between
experience and ability.
To show how differences in brain development
may ultimately link experience and academic
achievement, we focus in this article on
three core neurocognitive systems that are
crucial for school readiness. Typical measures
of school readiness such as achievement tests
or even IQ tests are quite imprecise from the
perspective of brain science.1 These tests assess
a diverse set of mental processes, involving
many neural systems, without telling
much about the specific systems of the child?s
mind and brain that are most involved in
school readiness. Recent work in the field of
cognitive neuroscience, however, has made it
possible to assess the specific neurocognitive
systems or brain regions involved in particular
cognitive skills. Using new neuroimaging
methods, researchers can design cognitive
tests that assess a single system, enabling
them to understand more precisely the cognitive
processes and underlying brain regions
whose development contributes to differences
in achievement. Ultimately, specific neurocognitive
systems might be differentially
targeted by early educational interventions.
We begin by introducing the three neurocognitive
systems, including the cognitive processes
involved, the types of tests used for assessment,
and the brain regions implicated.
We touch on the limited research into racial
differences across these systems and discuss
some links between socioeconomic background
and neurocognitive performance. We
then discuss research findings about how experience
can influence development of these
systems. We conclude by drawing implications
for educational interventions on early
brain and cognitive development in these
systems.
Three Core Neurocognitive
Systems
To illustrate how brain development can inform
notions of readiness and achievement,
we briefly describe three key neurocognitive
systems involved in cognitive skills necessary
for school success. Cognitive control, the
ability to override inappropriate thoughts and
behaviors, is associated with the prefrontal
cortex, located in the front of the brain.
Learning and memory involve the hippocampus,
buried deep within the brain?s temporal
lobe. And reading (and its precursors in preliterate
children) is associated with the temporo-
parietal and temporo-occipital cortex,
located on the left surface of the brain. Each
of these brain regions changes and matures
throughout childhood, and researchers are
currently trying to understand how children?s
experiences influence such brain development.
Scientists hope that this research will
lead to insights that are promising for the design
of specific educational interventions.
Cognitive Control
Cognitive processes attributed to the prefrontal
cortex include the ability to allocate
attention, to hold something ?online? in
memory, and to withhold an inappropriate
response.2 Such processes, collectively
known as cognitive control, are important developmentally,
as they underlie cognitive and
social skills essential to academic success,
Kimberly G. Noble, Nim To t t e n h a m , a n d B . J . C a s e y
72 THE FUTURE OF CHILDREN
such as the ability to ignore distracting events
inside and outside the classroom. In the laboratory,
researchers can design behavioral
tasks to assess a child?s ability to inhibit an inappropriate
response. For example, a widely
used paradigm known as the Go?No Go task
presents a child with many ?go? stimuli that
require a rote button-press response, along
with an occasional ?no go? stimulus that requires
the child to withhold a response.3
Now, thanks largely to developments in imaging
methods, like magnetic resonance imaging
(MRI), researchers can study cognitive skills
in the developing human brain. More than a
decade ago, Kenneth Kwong, Seji Ogawa, and
others showed that magnetic resonance is sensitive
to blood oxygenation changes in the
brain that may reflect changes in blood flow
and neuronal activity.4 The discovery that
MRI can assess activity in the human brain
without the need for radioactive tracers required
by other forms of brain imaging
opened a new era in the study of human brain
development and behavior. Since then, numerous
functional magnetic resonance imaging
(fMRI) studies have examined children
engaged in cognitive control tasks and have
found a characteristic age-related pattern in
the development of neural activity in the prefrontal
cortex.5 In young children, cognitive
control tasks are associated with diffuse patterns
of prefrontal cortex activity, whereas by
adolescence the pattern of activity is both
more focal and more intense. In adulthood,
activity remains focal, but somewhat less intense.
Because increasing age is also linked
with accuracy in performing a task, with experience,
and with learning, one possible interpretation
of these findings is that the age-related
decrease in brain activity could reflect
reduced recruitment of brain tissue as the task
becomes easier. But studies that have
matched children and adults on accuracy on
the Go?No Go task show that prefrontal activity
differences represent maturational change,
not difference in ability.6
Memory and Learning
The development of memory and learning is
also clearly important to academic success.
One aspect of learning is the ability to form
new associations among events. In laboratory
tasks that test the learning of new memories,
children typically see or hear lists of words,
stories, or scenes and then try to recollect the
presented stimuli.7 For very young children,
for whom a nonverbal memory assessment is
preferable, researchers first familiarize the
child with a stimulus and then present him or
her with test trials pairing the familiar stimulus
with a new one. Infants? known preference
for novelty allows researchers to infer
that an infant who spends a longer time looking
at the new stimulus recognizes the familiar
one.8
The ability to learn and remember is supported
in part by the hippocampus, located
deep inside the brain?s temporal lobe.9 A
child?s hippocampus increases in size with
age, with a particularly sharp increase before
the age of two.10 During the course of those
two years, a child?s ability to learn and re-
N e u r o s c i e n c e P e r s p e c t i v e s o n D i s p a r i t i e s i n S c h o o l R e a d i n e s s a n d C o g n i t i v e A c h i e v e m e n t
VOL. 15 / NO. 1 / SPRING 2005 73
Now, thanks largely to
developments in imaging
methods, like magnetic
resonance imaging (MRI),
researchers can study
cognitive skills in the
developing human brain.
member associations matures in terms of both
how much information is remembered and
how long it is retained.11 Although research
into the link between a child?s memory and
the functional neuroanatomical development
of the hippocampus is still in its early stages, a
recent imaging study showed that in both
children and adults, the speed of learning a
new association was correlated with hippocampal
activity.12 Interestingly, as with cognitive
control and the prefrontal cortex, the
activity associated with forming and remembering
new associations was more diffuse and
less focal in children than it was in adults.
Language and Reading
Both cognitive control and memory and
learning are general cognitive abilities that a
child brings to the academic environment. A
more specific cognitive ability?one that is
key to understanding the gap in school readiness?
is reading, along with the precursor
language skills that are critical for the development
of reading. Ample evidence has
shown that phonological awareness, or an understanding
of the sounds of language, is crucial
for reading.13 Not only do preliterate
children with better phonological awareness
learn to read more quickly than children with
less such awareness, but kindergarten phonological
awareness predicts teenage reading
ability better than kindergarten reading skill
does.14 Phonological awareness is measured
behaviorally by tasks such as rhyming, blending
sounds, and word-sound games that assess
the ability to manipulate syllables or
smaller units of speech known as phonemes.
A large swath of cortex known as the perisylvian
region stretches along the left side of the
brain and underlies most language functioning.
Within this larger area, two regions are
primarily responsible for the normal development
of reading.15 The first region, the superior
temporal gyrus, is involved in phonological
processing in normally reading adults and
children.16 Later childhood brings anatomical
maturation of this region as measured by
size, symmetry, and connectivity.17 The second
region, the fusiform gyrus, located along
the bottom-left side of the brain, has been associated
with the ability of skilled readers to
perceive automatically a written word. Activity
in the fusiform gyrus is positively correlated
with both reading ability and age.18 The
two regions are functionally linked in that the
development of the fusiform gyrus is thought
to be influenced by phonological processing
in the preliterate child.19
This sketch of these three neurocognitive systems
illustrates how researchers have begun
to understand the developmental course of
several cognitive processes and their neural
underpinnings. The challenge is to understand
how an individual child?s experiences,
many of which may vary according to racial,
ethnic, or socioeconomic background, may
affect the developing brain. Focusing on
these specific neurocognitive systems, rather
than on the multiple systems measured by
achievement tests, may make it possible both
to understand the link between experience
and brain development and to address the
racial gap in school readiness by directly targeting
the specific systems with interventions.
Racial and Socioeconomic
Disparities in Neurocognitive
Performance
Few researchers as yet have examined racial
disparities in academic achievement in terms
of specific neurocognitive systems. In fact,
few studies of cognitive development explicitly
examine race at all. One notable recent
exception, a study of cognitive control, investigated
a child?s ability to suppress an inappropriate
response as measured in a labora-
Kimberly G. Noble, Nim To t t e n h a m , a n d B . J . C a s e y
74 THE FUTURE OF CHILDREN
crease with age: a child?s cognitive ability at
age ten is more closely linked to his socioeconomic
status at age two than to his cognitive
ability at age two.25 But despite extensive
work on the connection between socioeconomic
status and cognitive performance as
measured by standardized testing, researchers
are only beginning to focus on the
specific brain functions that link childhood
experience and cognitive performance.
To address this gap in research, we recently
examined the neurocognitive functioning of
African American kindergartners from different
socioeconomic backgrounds, using tasks
from the cognitive neuroscience literature to
explore how childhood SES helps account for
the normal variance in performance across
different neurocognitive systems.26 We recruited
thirty middle-SES children and thirty
low-SES children from public kindergarten
classes in Philadelphia to participate in a battery
of behavioral tasks, each specific to a
particular neurocognitive system. The tasks
were designed to assess the language, cognitive
control, and memory systems, along with
several others. The systems we selected were
relatively independent of one another, had
correspondingly distinct locations in the
N e u r o s c i e n c e P e r s p e c t i v e s o n D i s p a r i t i e s i n S c h o o l R e a d i n e s s a n d C o g n i t i v e A c h i e v e m e n t
VOL. 15 / NO. 1 / SPRING 2005 75
tory task.20 The study found that children
from higher socioeconomic backgrounds
generally performed better on the test. It also
found, after controlling for socioeconomic
status, that African American and Hispanic
children resisted the interference of competing
demands better than white children did.
Although this study needs to be replicated to
confirm its findings, a preliminary interpretation
might be that racial disparities in
achievement, or at least in cognitive control,
are in fact mediated by socioeconomic differences
(and with associated differences in access
to resources).
The suggestion that socioeconomic differences
underlie racial differences in academic
performance is supported by the fact that minorities
are at much greater risk for growing
up in poverty.21 As detailed elsewhere in this
volume, children from impoverished backgrounds
are at heightened risk for poor academic
readiness and achievement because of
differences in their physical health, the quality
of the cognitive and emotional stimulation
they receive at home, their parenting, and
their early childhood education.22 Thus, although
work on racial differences in cognitive
development is limited as yet, researchers
are beginning to examine the link
between socioeconomic status (SES) and
neurocognitive achievement.
So far this research has documented a strong
and persistent connection between socioeconomic
status?most commonly measured
using education, occupation, and income?
and childhood cognitive ability and achievement
as measured by IQ, achievement test
scores, and functional literacy.23 In one study,
for example, socioeconomic status accounted
for some 20 percent of the variation in childhood
IQ.24 Another found that disparities in
achievement due to socioeconomic status in-
The suggestion that
socioeconomic differences
underlie racial differences in
academic performance is
supported by the fact that
minorities are at much
greater risk for growing up
in poverty.
brain, and had substantial roles in cognition
and school performance. We found that socioeconomic
status was generally correlated
with the children?s performance on the battery
of tasks as a whole, thus replicating the
well-documented socioeconomic gap in
global measures of cognitive performance.
But we also found that socioeconomic status
was disproportionately correlated with performance
in certain systems. Specifically,
children?s performance in tasks tapping the
left perisylvian (language) system and the
prefrontal (cognitive control) system varied
widely according to their socioeconomic status,
while their performance in tasks involving
other systems showed either no differences
or nonsignificant trends. The effects on
the language and cognitive control systems
were quite large. For the left perisylvian (language)
system, the mean score of the group
of middle-class children was 1.1 standard deviations
higher than the mean score of the
poorer children; for the prefrontal (cognitive
control) system, the difference was 0.68 standard
deviation.
When we replicated our preliminary study in
a larger sample of 150 multiracial children,
we largely confirmed our original findings.27
Socioeconomic status accounted for the most
variance in performance in the language system.
It also accounted for a good portion of
the variance in performance in different aspects
of cognitive control and in tasks involving
several other systems, including learning
and memory.
These two studies are the first ever to compare
directly the extent to which socioeconomic
factors account for the variance in
children?s performance on tasks involving different
neurocognitive systems. Both found
that the effect of socioeconomic status was
not uniform, that it differs from system to
system. In some systems, the effect was negligible.
Effects were greatest on variations in
language skills, but socioeconomic status also
accounts for some of the variation in other
systems, including cognitive control and possibly
learning and memory, among others.
Because of the exceptional importance of
reading skill for academic and life achievement,
we were particularly interested in examining
how socioeconomic status affects
that particular aspect of language development.
Correlations between socioeconomic
background and word reading ability are typically
fairly strong (they fall within the range
of 0.3 to 0.7, with 1 being a perfect correlation).
28 Often, researchers attribute this close
relationship to the link between socioeconomic
status and reading-related experiences,
such as the home literacy environment,
degree of early print exposure, and
quality of early schooling.29 But, as noted, a
largely separate line of research has provided
abundant evidence that phonological awareness
is causally related to reading development.
30 Despite independent work showing
that socioeconomic background and phonological
awareness are each associated with
Kimberly G. Noble, Nim To t t e n h a m , a n d B . J . C a s e y
76 THE FUTURE OF CHILDREN
Socioeconomic status
accounted for a good
portion of the variance in
performance in different
aspects of cognitive control
and in tasks involving
several other systems,
including learning and
memory.
reading achievement, surprisingly few studies
have explored how socioeconomic status relates
to phonemic awareness in predicting individual
differences in reading ability.31
We investigated this question and found that
on several different types of reading tasks, socioeconomic
status and phonological awareness
each accounted for unique variance in
skill.32 Furthermore, in certain cases, we
found that SES actually seemed to modulate
the relationship between phonological awareness
and reading. That is, at the highest levels
of phonological awareness, children were on
average reading well regardless of socioeconomic
background. In contrast, at lower levels
of phonological skill, a disparity emerged
such that higher-SES children continued to
read relatively well, whereas lower-SES children
began to struggle.
Together, these findings imply that the relationship
between socioeconomic background
and reading does not simply reflect differences
in the development of phonological
awareness skills. In contrast, multiple factors
play complex roles in the development of
reading and in predicting whether a child will
acquire this crucial skill easily or with difficulty.
Put simply, disparate causes may lead
to the same cognitive difficulties. Two different
children may have similar problems in
learning to read, but one may have inherently
poor phonological awareness skills, while the
other may be growing up in an environment
with scant access to literacy materials and instruction.
Is it possible then, that a child who
struggles with reading in the context of a lowliteracy
environment might have difficulties
that are fundamentally different from those
of a child who struggles despite access to a
higher-literacy environment? Might these
two children respond differently to different
types of intervention?
This brings us to a key application for neuroimaging.
If similar low levels of performance
in a skill such as reading may have different
causes, then imaging the brain may
help to tease such effects apart, extending
our knowledge beyond the limits of behavioral
data. It is now possible to examine
whether similar behavioral profiles resulting
from different causes could be rooted in different
effects on brain development. It may
be differences in brain development, rather
than differences in behavioral performance,
that ultimately predict an individual child?s
response to intervention. In the next section,
we examine how differences in experience
influence the development of neurocognitive
systems crucial for academic success.
Experience and Brain
Development
Thus far, we have focused on the developmental
course of several core cognitive
processes and their neural underpinnings, as
well as on how cognitive achievement is associated
with socioeconomic background and
perhaps race. The next challenge is to understand
how a child?s experiences?many of
which may reflect his or her socioeconomic,
racial, or ethnic background?may affect the
developing brain. Understanding how experience
influences behavioral and brain development
may make it possible to design educational
curriculums to target the specific
brain regions that underlie cognitive skills
important for academic success.
Experience shapes brain development at
many levels of organization, from molecules
to larger brain systems.33 Variations in such
types of experience as cognitive stimulation
and early life stress lead to functional and
anatomical changes throughout the brain in
both animals and people. Scientists can, for
example, cause broad neural changes in ani-
N e u r o s c i e n c e P e r s p e c t i v e s o n D i s p a r i t i e s i n S c h o o l R e a d i n e s s a n d C o g n i t i v e A c h i e v e m e n t
VOL. 15 / NO. 1 / SPRING 2005 77
mals by manipulating the laboratory environment,
enriching or depriving the animals? experience
in various ways.34 In humans, stress
has garnered much attention as one particular
experience that may affect cognitive and
academic achievement. Stressful life conditions
have been associated with low socioeconomic
status, and differences in emotional
support in the home account for a significant
portion of the variance in children?s verbal,
reading, and math skills, even when maternal
education, family structure, prenatal care, infant
health, nutrition, and mother?s age are
taken into account.35 Such cognitive differences
may be caused in part by biological responses
to stress.
Children raised in chronically stressful or
abusive situations demonstrate increased or
irregular production of stress hormone.36 In
animals, such abnormal levels of stress hormone
lead to adverse brain development,
particularly in the hippocampus.37 Reduced
hippocampal volume has also been found in
human adults in a variety of stress-related
conditions, including post-traumatic stress
disorder and major depression.38 Given the
critical role of the hippocampus in learning
and memory, it is not surprising that changes
in hippocampal activity caused by prolonged
exposure to elevated stress hormone may
lead to deficits in learning.39
Developmental studies of maltreated children
find generalized intellectual and academic
impairments, as measured by IQ or
achievement tests.40 Studies applying more
specific neurocognitive methods suggest that
these children also show deficits in cognitive
control.41 MRI studies of children suffering
from post-traumatic stress disorder caused by
maltreatment have found not only that their
brains are smaller overall than those of children
who have not been maltreated, but also
that their frontal lobe structure is abnormal.
42 These studies, however, cannot draw
causal relationships between maltreatment
and brain changes.
To sort out these findings, we have begun to
examine how one extreme form of chronic
childhood stress?being placed in an orphanage?
affects a child?s developing brain. Researchers
have recognized for some time that
both a child?s age at placement and the duration
of the placement affect the child?s development.
43 We have recruited and collected
preliminary data on fourteen children between
the ages of five and eleven who spent
time in an orphanage. The children were
adopted between the ages of six months and
five years, except for one boy, who was
adopted at age eight. They were placed in the
orphanage between birth and age two, with
the exception of the same boy, who was
placed at age five.
Of the fourteen children, seven have at least
one clinical psychiatric diagnosis. Strikingly,
the older the children were at adoption, the
more likely they are to have symptoms, and
ultimately a diagnosis. The healthiest children
were placed in the orphanage young
and adopted young, and they spent relatively
less time in the orphanage overall.
Most of the children?s general cognitive ability
scores fell within the average range, but
their estimated full-scale IQ scores were negatively
correlated with time spent in the orphanage
(see figure 1). The children who
lived there a shorter time tended to have
higher IQ scores.
To assess cognitive control in these children,
we used the Go?No Go test.44 The performance
of the adopted children on the test differed
from that of twelve age-matched con-
Kimberly G. Noble, Nim To t t e n h a m , a n d B . J . C a s e y
78 THE FUTURE OF CHILDREN
trols in overall accuracy. Performance was
negatively correlated with age of adoption:
children adopted at a younger age tended to
score higher on the test (see figure 2). Thus
stress associated with institutionalization appears
to be linked with decreased cognitive
ability as measured both by general intelligence
tests and by specific measures of cognitive
control. These findings are in line with
those noted earlier, that traumatized children
show abnormal maturation of prefrontal
function.
How do these cognitive changes relate to
brain changes? We examined the effects of institutionalization
on brain development using
magnetic resonance imaging on a subset of
eight of these children. As seen in figure 3,
MRIs of those children showed an association
between total brain volume and estimated IQ,
a trend that has been repeatedly demonstrated
elsewhere.45 The MRIs also showed a
moderate association between the length of
time a child spent in an orphanage and the
child?s prefrontal volume (after overall brain
volume had been taken into account).
Because the hippocampus is implicated in
memory and learning and because it is vulnerable
to stress, we tested for a link between
its volume and the length of time a child lived
in the orphanage. As figure 4 shows, the volume
decreased as a function of time spent in
the institution. (We controlled for current
age and overall brain volume.) These results,
too, are in line with those noted in adults
with post-traumatic stress disorder. Not surprisingly,
we found that previously institutionalized
children perform poorly on learning
and memory tasks. Preliminary findings
from our laboratory showed that these children
were significantly slower than the control
group to learn new stimulus-response associations
and override old ones, an ability
that correlates with hippocampal activity.46
Hippocampal volume was also correlated
with time spent with the adopted family: the
longer a child lived with a stable family, the
greater his or her hippocampal volume. This
finding suggests a powerful effect of the positive
experience of adoption from orphanage
to home.
Although most research on stress and humans
has focused on extreme?and rare?
cases such as institutionalization, milder daily
elevations in stress may have long-term effects
as well. In children of low socioeco-
N e u r o s c i e n c e P e r s p e c t i v e s o n D i s p a r i t i e s i n S c h o o l R e a d i n e s s a n d C o g n i t i v e A c h i e v e m e n t
VOL. 15 / NO. 1 / SPRING 2005 79
80 90 100 110 120
0
10
20
30
40
50
60
l
l
l
l l
l
l
l
l
l
l
l
l
l
Time (months)
Estimated full-scale IQ
0 20 60 80 100
40
50
60
70
80
90
100
Accuracy on Go?No Go Task (percent)
Age (months)
40
l l
l
l
l l l
l
l
l l
l
l
l
Figure 1. Time Spent in Orphanage
and IQ
Figure 2. Cognitive Control and Age
at Adoption
nomic status, for example, Sonia Lupien and
her colleagues found increased levels of salivary
cortisol, which were linked with depression
in their mother.47 This research is potentially
quite relevant to understanding the
biological and neural underpinnings of the
achievement gap between children of different
socioeconomic backgrounds.
As is evident from the effect of adoption in
our study of children placed in an orphanage,
experience need not be negative to shape a
developing brain. On the contrary, positive
differences in experience can quite powerfully
lead to functional reorganization of the
brain. One often-cited example is learning a
second language. As has long been recognized,
the older a person is when exposed to a
second language, the less likely he or she is to
be able to develop true, accent-free fluency.
Recent neuroimaging studies have begun to
elucidate the neurobiological basis for this experience.
48 Typically, the studies present children
with written or spoken words in both
their first and their second languages and examine
differences in brain activity in response.
In bilingual children who learn a second
language before they turn seven, brain
activity in response to the two languages is
similar and takes place in overlapping regions
of the left side of the brain. But in children
who learn a second language later, brain activity
in response to the two languages occurs
in nonoverlapping regions. In particular, the
first language typically elicits the usual leftsided
pattern of activity, whereas the second
often causes a more variable pattern that is
more likely to be localized to the right side.
Although the brain retains plasticity for learning
a second language, the specific pattern of
plasticity appears to depend on the age when
that language is learned, which may also reflect
ultimate fluency.
Finally, discussions of experience-related
plasticity in cognitive ability and brain development
often evoke the issue of genetics.
What is the role of genes in the development
of cognitive abilities? Researchers have long
agreed that both genes and experience influence
cognitive outcomes. For instance, twin
studies have shown that even when genetic
effects are taken into account, violence in the
home is linked with lower IQ.49 Conversely,
both genes and environment affect cognitive
resilience to the effects of low socioeconomic
status.50 Adoption studies have also shown
that the socioeconomic backgrounds of both
biological and adoptive parents are independent
predictors of adopted children?s IQ,
reflecting genetic and experiential influences
on the child, respectively.51
Kimberly G. Noble, Nim To t t e n h a m , a n d B . J . C a s e y
80 THE FUTURE OF CHILDREN
80 100 120
800
1,000
1,200
1,400
1,600
1,800
Brain volume (ml)
Estimated full-scale IQ
l
l
l
l
l
l l l
Figure 3. Brain Volume and IQ
0 20 40 60
0
2
4
6
8
10
Adjusted hippocampal volume (ml)
Time (months)
ll
l l l
l
l
l
Figure 4. Hippocampal Volume and Time
Spent in Orphanage
But the nature-nurture question is more nuanced
than merely being a matter of where
the balance of influence lies. Researchers
now recognize that genes and experience are
not truly independent predictors, but that in
many cases nature is in part moderated by
nurture. Animal research, for example, has
shown that naturally occurring variations in
maternal care can alter the expression of
genes that regulate the response to stress and
that early social attachment relationships can
modify the heritability of aggressive behavior.
52 Human research has drawn similar conclusions.
Of particular relevance to understanding
the gap in school readiness is a
recent study showing that among families of
lower socioeconomic status, variation in IQ is
far more environmental than genetic in origin,
whereas the converse holds in families of
higher socioeconomic status.53 That is, an impoverished
child?s background and experiences
can so heavily influence his or her degree
of achievement that his genetic makeup
is nearly irrelevant in predicting his academic
success. Optimistically, such a powerful role
for experience suggests that intervention may
be particularly successful among disadvantaged
children.
Brain-Targeted Interventions
In this final section, we look ahead to the role
that brain plasticity may play in developing
and testing cognitive interventions in the
three neurocognitive systems on which we
have focused: memory and learning, cognitive
control, and reading. It is premature to
recommend specific interventions on the
basis of brain evidence, but preliminary research
in this nascent field is promising.
Researchers in brain plasticity have as yet
done little work on memory training in humans.
Although animal research has repeatedly
shown that training on memory paradigms
can lead to improved learning and
problem solving that is directly related to hippocampal
plasticity, it is not yet clear whether
similar effects could be observed in children.54
Cognitive control has received somewhat
more attention. Several studies have shown
not only that young children with attention
deficit hyperactivity disorder (ADHD) can
benefit from repeated training on laboratory
tasks known to involve prefrontal function,
but also that training on such tasks can improve
performance on untrained tasks involving
similar regions.55 These studies did not
directly measure brain function, relying instead
on tasks already shown to engage prefrontal
regions. Recently, however, M. R.
Rueda and colleagues showed that four-yearolds
who attended seven sessions of attention
training showed significant improvement on
abstract reasoning skills relative to children
who received a control intervention of watching
videos. Furthermore, during a cognitive
control task administered after their training
was complete, the children showed brain activity
that was more adult-like than that of the
control group.56 These preliminary results
suggest the possibility of designing broader
educational interventions that specifically
target cognitive control, which a recent study
found to be the single best predictor of resilience
among high-risk children, even controlling
for age, gender, negative life events,
chronic strain, abuse, nonverbal IQ, selfesteem,
parental monitoring, and emotional
support.57 Of course, the feasibility of any intervention
program must be assessed outside
the laboratory before being implemented on
a larger scale.
Reading has attracted by far the most attention
from those scientists investigating intervention-
related brain plasticity.58 Many studies
have provided behavioral evidence that
N e u r o s c i e n c e P e r s p e c t i v e s o n D i s p a r i t i e s i n S c h o o l R e a d i n e s s a n d C o g n i t i v e A c h i e v e m e n t
VOL. 15 / NO. 1 / SPRING 2005 81
children with mild to severe reading impairments
can benefit from interventions that explicitly
support phonological awareness and
provide training in the alphabetic decoding
skills necessary to convert print to sound.59
Recent examinations of the neural effects of
such behavioral studies provide a better understanding
of how such programs improve
skills, with the ultimate goal of targeting intervention
to individual children?s needs. Several
investigators have used neuroimaging
techniques to follow brain changes in children
over the course of an intervention. One
investigation found decreased brain activity
in the left superior temporal gyrus region in
eight children with reading difficulties, as
compared with nonimpaired children.60 Following
a two-month intervention involving
eighty hours of phonological processing work
with one of two commercial packages
(Phono-Graphics and Lindamood Phoneme
Sequencing), the reading-impaired children?s
mean standardized reading scores improved
from the 5th percentile to the 50th percentile.
The children also showed increases
in left superior temporal gyrus activity (as
well as a decrease in right-sided activity). The
eight nonimpaired children who did not participate
in the intervention demonstrated stable
brain responses over the same time span.
Importantly, the study included no readingimpaired
control group, making it impossible
to tell whether changes were specific to the
intervention or simply the result of generic
tutoring or even schooling effects. Another
interpretive difficulty was that before the intervention,
the reading-impaired children
showed very low accuracy in performing the
task measured by the brain scanner. The
changes in brain activity following the intervention,
therefore, could have been due not
to a change in brain function per se but
rather to the children?s engagement in a task
to which they had previously not attended.
Similarly, Elise Temple and colleagues measured
changes in functional activity in a group
of reading-impaired children in whom preintervention
functional magnetic resonance
imaging indicated reduced activity in readingrelated
regions relative to children in a control
group.61 After the children in the experimental
group participated in a six-week,
forty-five-hour intervention, including a commercial
computer-based training program
(Fast ForWord Language) and a special
school curriculum for children with dyslexia,
their reading improved significantly. Changes
in their post-test functional MRI results were
widespread, extending to fourteen brain regions,
some of which also changed in the nonimpaired
group. Most of the regions undergoing
change are thought to be typically
involved in reading; several are not. The size
of changes in regions associated with reading
was correlated with improvements in oral language,
but not with reading improvements.
Again, this study is difficult to interpret because
it lacked a reading-impaired control
group randomized to a different intervention.
To make interpretation even more complicated,
in a separate randomized controlled
study, more than 200 children in an urban
school district received Fast ForWord but
made no gains in reading compared with a
control group of reading-impaired children
who did not receive the program.62 This finding
underscores the need for a readingimpaired
control group in imaging studies and
suggests that the strict adherence to an intervention
required in the laboratory setting may
be unrealistic in the classroom.
Finally, a recent study followed a group of
children who received an experimental intervention
consisting of fifty minutes a day of individual
tutoring focused on phonological
awareness and the alphabetic principle and
contrasted it with a ?community interven-
Kimberly G. Noble, Nim To t t e n h a m , a n d B . J . C a s e y
82 THE FUTURE OF CHILDREN
tion? group that received normal schoolbased
remedial reading instruction.63 The
children were tested before and after eight
months of intervention and were also compared
with a control group of nonimpaired
readers. Following the intervention, children
in the experimental group had made significantly
greater gains in reading fluency than
had those in the community intervention
group. They also showed brain activity during
reading that looked remarkably similar to
that of children in the nonimpaired control
group?and they maintained this more typical
pattern of activity for at least one year.
The community intervention group showed
less activity in the typical reading-related
areas than did the other two groups.
Together, these three studies suggest that
brain regions involved with reading in typically
developing readers may prove to be
quite malleable in response to effective therapeutic
interventions. Brain activation patterns
in these regions can change dramatically
over the course of relatively short-lived
interventions. As noted, successful interpretation
of study results requires the rigorous
use of control groups to examine both the behavioral
efficacy and neural specificity of any
intervention effects. In addition, improvements
must be followed over time to verify
that gains persist. Finally, interventions that
succeed in the laboratory must be tested in
real classroom environments before they can
be widely implemented. Although it would
be premature at this time to recommend a
specific program for use, we are becoming
more confident of the efficacy of combined
training in phonological awareness and the
alphabetic principle, as laboratory tests of
that particular combination often show both
improved reading skills and patterns of brain
activity that look more like those seen in typically
developing readers.
But it is not enough for an intervention to improve
reading skills on average. Ultimately,
the goal is to tailor particular interventions
for individual children. If, as we believe, similar
low levels of reading performance?or
any other neurocognitive skill?may result
from different causes, then imaging the brain
may help to tease such effects apart, extending
our knowledge beyond the limits of behavioral
data. We now have the ability to examine
whether similar behavioral profiles
associated with disparate risk factors might
be rooted in different effects on brain development.
In fact, it may be differences in
brain development, rather than in behavioral
performance, that ultimately predict an individual
child?s response to intervention.
Tantalizing preliminary evidence for this suggestion
comes from a study showing that both
socioeconomic status and a particular neuroanatomical
measure (left-right asymmetry
of the planum temporale in the temporal
lobe) independently predicted reading ability.
64 The study suggests that researchers can
predict a child?s reading achievement levels
better by using a combination of information
about the brain and about social background
than by using either type of information
alone. By using both types of information,
they might one day be able to design interventions
that meet an individual child?s needs
in ways that simple behavioral measures
alone cannot. Indeed, by thus honing the
tools of intervention, they may ultimately reduce
the gap in achievement so often observed
for underserved groups.
N e u r o s c i e n c e P e r s p e c t i v e s o n D i s p a r i t i e s i n S c h o o l R e a d i n e s s a n d C o g n i t i v e A c h i e v e m e n t
VOL. 15 / NO. 1 / SPRING 2005 83
Kimberly G. Noble, Nim To t t e n h a m , a n d B . J . C a s e y
84 THE FUTURE OF CHILDREN
Endnotes
1. Martha J. Farah and Kimberly G. Noble, ?Socioeconomic Influences on Brain Development: A Preliminary
Study,? in Developing Individuality in the Human Brain: A Tribute to Michael Posner, edited by E. Awh, U.
Mayr, and S. Keele (Washington: American Psychological Association, forthcoming).
2. B. J. Casey, Jay N. Giedd, and Kathleen M. Thomas, ?Structural and Functional Brain Development and Its
Relation to Cognitive Development,? Biological Psychiatry 54, nos. 1?3 (2000).
3. B. J. Casey and others, ?A Developmental fMRI Study of Prefrontal Activation during Performance of a
Go?No-Go Task,? Journal of Cognitive Neuroscience 9 (1997).
4. Kenneth K. Kwong and others, ?Dynamic Magnetic Resonance Imaging of Human Brain Activity during
Primary Sensory Stimulation,? Proceedings of the National Academy of Sciences, USA 89 (1992); Seji Ogawa
and others, ?Intrinsic Signal Changes Accompanying Sensory Stimulation: Functional Brain Mapping with
Magnetic Resonance Imaging,? Proceedings of the National Academy of Sciences, USA 89, no. 13 (1992).
5. Silvia A. Bunge and others, ?Prefrontal Regions Involved in Keeping Information in and out of Mind,? Brain
124, no. 10 (2001); B. J. Casey and others, ?Activation of Prefrontal Cortex in Children during a Non-Spatial
Working Memory Task with Functional MRI,? Neuroimage 2 (1995); B. J. Casey, Nim Tottenham, and John
Fossella, ?Clinical, Imaging, Lesion and Genetic Approaches toward a Model of Cognitive Control,? Developmental
Psychobiology 40 (2002); B. J. Casey and others, ?A Developmental Functional MRI Study of Prefrontal
Activation during Performance of a Go?No-Go Task,? Journal of Cognitive Neuroscience 9 (1997);
Sarah Durston and others, ?The Effect of Preceding Context on Inhibition: An Event-Related fMRI Study,?
Neuroimage 16, no. 2 (2002); Torkel Klingberg, Hans Forssberg, and Helena Westerberg, ?Increased Brain
Activity in Frontal and Parietal Cortex Underlies the Development of Visuospatial Working Memory Capacity
during Childhood,? Journal of Cognitive Neuroscience 14, no. 1 (2002); Beatriz Luna and others, ?Maturation
of Widely Distributed Brain Function Subserves Cognitive Development,? Neuroimage 13, no. 5
(2001); Kathleen M. Thomas and others, ?A Developmental Functional MRI Study of Spatial Working
Memory,? Neuroimage 10 (1999).
6. Durston and others, ?The Effect of Preceding Context on Inhibition?; Casey and others, ?A Developmental
fMRI Study of Prefrontal Activation during Performance of a Go?No-Go Task? (see note 5).
7. Susan E. Gathercole, ?The Development of Memory,? Journal of Child Psychology and Psychiatry 39, no. 1
(1998).
8. Charles A. Nelson and Sara J. Webb, ?A Cognitive Neuroscience Perspective on Early Memory Development,?
in The Cognitive Neuroscience of Development, edited by Michelle de Haan and Mark H. Johnson
(New York: Psychology Press, 2003).
9. Larry R. Squire, Frank Haist, and Arthur P. Shimamura, ?The Neurology of Memory: Quantitative Assessment
of Retrograde Amnesia in Two Groups of Amnesiac Patients,? Journal of Neuroscience 9 (1989).
10. Jay N. Giedd and others, ?Brain Development during Childhood and Adolescence: A Longitudinal MRI
Study,? Nature Neuroscience 2, no. 10 (1999); Hidetsuna Utsunomiya and others, ?Development of the
Temporal Lobe in Infants and Children: Analysis by MR-Based Volumetry,? American Journal of Neuroradiology
20, no. 4 (1999).
11. Gathercole, ?The Development of Memory? (see note 7).
12. Casey, Tottenham, and Fossella, ?Clinical, Imaging, Lesion and Genetic Approaches toward a Model of
Cognitive Control? (see note 5).
13. Marilyn J. Adams, Learning to Read: Thinking and Learning about Print (MIT Press, 1990); Lynette
Bradley and Peter E. Bryant, ?Categorizing Sounds and Learning to Read?a Causal Connection,? Nature
301 (1983); Richard K. Wagner and Joseph K. Torgesen, ?The Nature of Phonological Processing and Its
Causal Role in the Acquisition of Reading Skills,? Psychological Bulletin 101 (1987).
14. Wagner and Torgesen, ?The Nature of Phonological Processing? (see note 13); G. Wayne MacDonald and
Anne Cornwall, ?The Relationship between Phonological Awareness and Reading and Spelling Achievement
Eleven Years Later,? Journal of Learning Disabilities 28, no. 8 (1995).
15. Bruce M. McCandliss and Kimberly G. Noble, ?The Development of Reading Impairment: A Cognitive
Neuroscience Model,? Mental Retardation and Developmental Disabilities Research Reviews 9 (2003).
16. Barry Horwitz, Judith M. Rumsey, and Brian C. Donohue, ?Functional Connectivity of the Angular Gyrus in
Normal Reading and Dyslexia,? Proceedings of the National Academy of Sciences, USA 95 (1998); Cathy J.
Price and others, ?Segregating Semantic from Phonological Processes during Reading,? Journal of Cognitive
Neuroscience 9, no. 6 (1997); Kenneth Pugh and others, ?Cerebral Organization of Component Processes in
Reading,? Brain 119 (1996); Bennett A. Shaywitz and others, ?Disruption of Posterior Brain Systems for
Reading in Children with Developmental Dyslexia,? Biological Psychiatry 52 (2002); Panagiotis G. Simos
and others, ?Dyslexia-Specific Brain Activation Profile Becomes Normal Following Successful Remedial
Training,? Neurology 58 (2002); Elise Temple and others, ?Disrupted Neural Responses to Phonological and
Orthographic Processing in Dyslexic Children: An fMRI Study,? NeuroReport 12, no. 2 (2001).
17. Giedd and others, ?Brain Development during Childhood and Adolescence? (see note 10); Elizabeth B.
Sowell and others, ?Mapping Sulcal Pattern Assymetry and Local Cortical Surface Gray Matter Distribution
in vivo: Maturation in Perisylvian Cortices,? Cerebral Cortex 12 (2002); Tomas Paus and others, ?Structural
Maturation of Neural Pathways in Children and Adolescents: In Vivo Study,? Science 283 (1999).
18. Bradley L. Schlaggar and others, ?Functional Neuroanatomical Differences between Adults and School-Age
Children in the Processing of Single Words,? Science 296 (2002); Shaywitz and others, ?Disruption of Posterior
Brain Systems for Reading in Children with Developmental Dyslexia? (see note 16).
19. McCandliss and Noble, ?The Development of Reading Impairment: A Cognitive Neuroscience Model? (see
note 15).
20. Enrico Mezzacappa, ?Alerting, Orienting, and Executive Attention: Developmental Properties and Socio-
Demographic Correlates in an Epidemiological Sample of Young, Urban Children,? Child Development
(forthcoming).
21. U.S. Bureau of the Census, Statistical Abstract of the United States (2000).
22. See also Theresa Hawley and Elizabeth R. Disney, ?Crack?s Children: The Consequences of Maternal Cocaine
Abuse,? Social Policy Report 6 (1992); Nancy K. Klein, Maureen Hack, and Naomi Breslau, ?Children
Who Were Very Low Birthweight: Development and Academic Achievement at Nine Years of Age,? Journal
of Developmental and Behavioral Pediatrics 10 (1989); Herbert L. Needleman and others, ?The Long Term
Effects of Low Doses of Lead in Childhood: An Eleven-Year Followup Report,? New England Journal of
Medicine 322 (1990); Robert H. Bradley and others, ?The Home Environments of Children in the United
N e u r o s c i e n c e P e r s p e c t i v e s o n D i s p a r i t i e s i n S c h o o l R e a d i n e s s a n d C o g n i t i v e A c h i e v e m e n t
VOL. 15 / NO. 1 / SPRING 2005 85
States, Part I: Variations by Age, Ethnicity and Poverty Status,? Child Development 72, no. 6 (2001); Jeanne
Brooks-Gunn, Pamela K. Klebanov, and Greg .J. Duncan, ?Ethnic Differences in Children?s Intelligence
Test Scores: Role of Economic Deprivation, Home Environment, and Maternal Characteristics,? Child Development
67 (1996); Vonnie C. McCloyd, ?The Impact of Economic Hardship on Black Families and Children:
Psychological Distress, Parenting, and Socioemotional Development,? Child Development 61 (1990);
W. Steven Barnett, ?Long-Term Cognitive and Academic Effects of Early Childhood Education on Children
in Poverty,? Preventive Medicine 27 (1998).
23. Margaret E. Ensminger and Kate E. Fothergill, ?A Decade of Measuring SES: What It Tells Us and Where
to Go from Here,? in Socioeconomic Status, Parenting and Child Development, edited by Marc H. Bornstein
and Robert H. Bradley (Mahwah, N.J.: Lawrence Erlbaum Associates, 2003). Fong-ruey Liaw and Jeanne
Brooks-Gunn, ?Cumulative Familial Risks and Low-Birthweight Children?s Cognitive and Behavioral Development,?
Journal of Clinical Child Psychology 23, no. 4 (1994); Judith Smith, Jeanne Brooks-Gunn, and
Pamela K. Klebanov, ?Consequences of Living in Poverty for Young Children?s Cognitive and Verbal Ability
and Early School Achievement,? in Consequences of Growing up Poor, edited by Greg Duncan and Jeanne
Brooks-Gunn (New York: Russell Sage, 1997); Jeanne Brooks-Gunn, Guang Guo, and Frank Furstenberg,
?Who Drops out of and Who Continues beyond High School?? Journal of Research on Adolescence 3 (1993);
Nazli Baydar, Jeanne Brooks-Gunn, and Frank Furstenberg, ?Early Warning Signs of Functional Illiteracy:
Predictors in Childhood and Adolescence,? Child Development 64 (1993).
24. Allen W. Gottfried and others, ?Socioeconomic Status in Children?s Development and Family Environment:
Infancy through Adolescence,? in Socioeconomic Status, Parenting and Child Development, edited by Marc
H. Bornstein and Robert H. Bradley (Mahwah, N.J.: Lawrence Erlbaum Associates, 2003).
25. Leon Feinstein, ?Inequality in the Early Cognitive Development of British Children in the 1970 Cohort,?
Economica 70 (2003).
26. Kimberly G. Noble, M. Frank Norman, and Martha J. Farah, ?Neurocognitive Correlates of Socioeconomic
Status in Kindergarten Children,? Developmental Science (forthcoming).
27. Kimberly G. Noble, Martha J. Farah, and Bruce M. McCandliss, ?Normal Neurocognitive Development:
Influence of Socioeconomic Status,? paper presented to the Cognitive Neuroscience Society, San Francisco,
2004.
28. Karl R. White, ?The Relation between Socio-Economic Status and Academic Achievement,? Psychological
Bulletin 91 (1982).
29. Steven A. Hecht and others, ?Explaining Social Class Differences in Growth of Reading Skills from Beginning
Kindergarten through Fourth Grade: The Role of Phonological Awareness, Rate of Access, and Print
Knowledge,? Reading and Writing 12, nos. 1?2 (2000).
30. See note 13.
31. Judith A. Bowey, ?Socioeconomic Status Differences in Preschool Phonological Sensitivity and First-Grade
Reading Achievement,? Journal of Educational Psychology 87, no. 3 (1995); Hecht and others, ?Explaining
Social Class Differences in Growth of Reading Skills from Beginning Kindergarten through Fourth Grade?
(see note 29); Ita S. Raz and Peter Bryant, ?Social Background, Phonological Awareness and Children?s
Reading,? British Journal of Developmental Psychology 8, no. 3 (1990).
Kimberly G. Noble, Nim To t t e n h a m , a n d B . J . C a s e y
86 THE FUTURE OF CHILDREN
32. Kimberly Noble, Martha J. Farah, and Bruce M. McCandliss, ?The Effects of Socioeconomic Status and
Phonological Awareness on Reading Development,? paper presented to the Society for the Scientific Study
of Reading, Amsterdam, The Netherlands, 2004.
33. William T. Greenough, James E. Black , and Christopher S. Wallace, ?Experience and Brain Development,?
Child Development 58, no. 3 (1987); Sonia J. Lupien and others, ?Child?s Stress Hormone Levels Correlate
with Mother?s Socioeconomic Status and Depressive State,? Biological Psychiatry 48 (2000); Bruce S.
McEwen, ?From Molecules to Mind: Stress, Individual Differences, and the Social Environment,? Annals of
the New York Academy of Sciences 935 (2001); Mark R. Rosenzweig and Edward L. Bennett, ?Psychobiology
of Plasticity: Effects of Training and Experience on Brain and Behavior,? Behavioral Brain Research 78
(1996).
34. Rosenzweig and Bennett, ?Psychobiology of Plasticity? (see note 33).
35. The association between stress and low SES is noted by Brooks-Gunn, Klebanov, and Duncan, ?Ethnic Differences
in Children?s Intelligence Test Scores? (see note 22). The effects of emotional support in the home
are documented by Sanders Korenman, Jane E. Miller, and John E. Sjaastad, ?Long-Term Poverty and
Child Development in the United States: Results from the NLSY,? Children and Youth Services Review 17
(1995).
36. Mary Carlson and Felton Earls, ?Psychological and Neuroendocrinological Sequelae of Early Social Deprivation
in Institutionalized Children in Romaina,? Annals of the New York Academy of Sciences 807 (1997);
Megan R. Gunnar, ?Early Adversity and the Development of Stress Reactivity and Regulation,? in The Minnesota
Symposia on Child Psychology, vol. 31: The Effects of Early Adversity on Neurobehavioral Development,
edited by C. A. Nelson (Mahwah, N.J.: Lawrence Erlbaum and Associates, 2000); Michael D. De Bellis
and others, ?Developmental Traumatology, Part I: Biological Stress Systems,? Biological Psychiatry 45
(1999).
37. Elizabeth Gould and others, ?Neurogenesis in the Dentate Gyrus of the Adult Tree Shrew Is Regulated by
Psychosocial Stress and Nmda Receptor Activation,? Journal of Neuroscience 17 (1997); Elizabeth Gould
and others, ?Proliferation of Granule Cell Precursors in the Dentate Gyrus of Adult Monkeys Is Diminished
by Stress,? Brain Research: Developmental Brain Research 103 (1998); Patima Tanapat, Lisa A. Galea, and
Elizabeth Gould, ?Stress Inhibits the Proliferation of Granule Cell Precursors in the Developing Dentate
Gyrus,? Journal of Developmental Neuroscience 16 (1998).
38. J. Douglas Bremner and others, ?MRI-Based Measurement of Hippocampal Volume in Patients with Combat-
Related Posttraumatic Stress Disorder,? American Journal of Psychiatry 152, no. 7 (1995); J. Douglas
Bremner and others, ?Hippocampal Volume Reduction in Major Depression,? American Journal of Psychiatry
157, no. 1 (2000); Yvette I. Sheline and others, ?Hippocampal Atrophy in Recurrent Major Depression,?
Proceedings of the National Academy of Sciences USA 93 (1996); David C. Steffens and others, ?Hippocampal
Volume in Geriatric Depression,? Biological Psychiatry 43 (2000).
39. The role of the hippocampus in learning and memory is discussed by Squire, Haist, and Shimamura, ?The
Neurology of Memory? (see note 9). On the effects of prolonged exposure to stress hormones, see Robert
M. Sapolsky, Stress, the Aging Brain, and the Mechanisms of Neuron Death (MIT Press, 1992).
40. Martha Augoustinos, ?Developmental Effects of Child Abuse: A Number of Recent Findings,? Child Abuse
and Neglect 11 (1987); Normand J. Carrey and others, ?Physiological and Cognitive Correlates of Child
N e u r o s c i e n c e P e r s p e c t i v e s o n D i s p a r i t i e s i n S c h o o l R e a d i n e s s a n d C o g n i t i v e A c h i e v e m e n t
VOL. 15 / NO. 1 / SPRING 2005 87
Abuse,? Journal of the American Academy of Child and Adolescent Psychiatry 34 (1995); John Money,
Charles Annecillo, and John F. Kelly, ?Abuse-Dwarfism Syndrome: After Rescue, Statural and Intellectual
Catchup Growth Correlate,? Journal of Clinical Child Psychology 12, nos. 279?83 (1983); Cynthia Perez and
Cathy S. Widom, ?Childhood Victimization and Longterm Intellectual and Academic Outcomes,? Child
Abuse and Neglect 18 (1994); Robert Pianta, Byron Egeland, and Martha F. Erickson, ?Results of the
Mother-Child Interaction Research Project,? in Child Maltreatment: Theory and Research on the Causes
and Consequences of Child Abuse and Neglect, edited by Dante Cicchetti and Vicki Carlson (Cambridge
University Press, 1989); Penelope K. Trickett, Catherine A. McBride-Chang, and Frank W. Putnam, ?The
Classroom Performance and Behavior of Sexually Abused Girls,? Developmental Psychopathology 6 (1994).
41. Sue R. Beers and Michael D. De Bellis, ?Neuropsychological Function in Children with Maltreatment-
Related Posttraumatic Stress Disorder,? American Journal of Psychiatry 159 (2002); Ali R. Moradi and others,
?Everyday Memory Deficits in Children and Adolescents with PTSD: Performance on the Rivermead
Behavioral Memory Test,? Journal of Child Psychology and Psychiatry 40 (1999).
42. Michael D. De Bellis and others, ?Developmental Traumatology Part II: Brain Development,? Biological
Psychiatry 45 (1999); Victor G. Carrion and others, ?Attenuation of Frontal Asymmetry in Pediatric Posttraumatic
Stress Disorder,? Biological Psychiatry 50 (2001).
43. H. Durfee and K. Wolf, ?Anstaltspflege und Entwicklung im Ersten Lebensjahr,? Zeitschrift fur Kinderforschung
(translated in Spitzer 1940) 42, no. 3 (1933); Lawson G. Lowry, ?Personality Distortion and Early
Institutional Care,? American Journal of Orthopsychiatry 10, no. 3 (1940).
44. Durston and others, ?The Effect of Preceding Context on Inhibition? (see note 5).
45. Ian J. Deary and Peter G. Caryl, ?Neuroscience and Human Intelligence Differences,? Trends in Neuroscience
20 (1997); Larry A. Flashman and others, ?Intelligence and Regional Brain Volumes in Normal Controls,?
Intelligence 25, no. 3 (1997); Bruce F. Pennington and others, ?A Twin MRI Study of Size Variations
in the Human Brain,? Journal of Cognitive Neuroscience 12, no. 1 (2000); Uener Tan and others, ?Magnetic
Resonance Imaging Brain Size/IQ Relations in Turkish University Students,? Intelligence 27, no. 1 (1999);
John C. Wickett and others, ?Relationships between Factors of Intelligence and Brain Volume,? Personality
and Individual Differences 29, no. 6 (2000).
46. Casey, Tottenham, and Fossella, ?Clinical, Imaging, Lesion and Genetic Approaches toward a Model of
Cognitive Control? (see note 5).
47. Lupien and others, ?Child?s Stress Hormone Levels Correlate with Mother?s Socioeconomic Status and Depressive
State? (see note 33).
48. Daphne Bavelier and Helen J. Neville, ?Cross-Modal Plasticity: Where and How?? Nature Reviews Neuroscience
3 (2002).
49. Karestan C. Koenen and others, ?Domestic Violence Is Associated with Environmental Suppression of IQ in
Young Children,? Development and Psychopathology 15 (2003).
50. Julia Kim-Cohen and others, ?Genetic and Environmental Processes in Young Children?s Resilience and
Vulnerability to Socioeconomic Deprivation,? Child Development 75, no. 3 (2004).
51. Christiane Capron and Michel Duyme, ?Assessment of Effects of Socio-Economic Status on IQ in a Full
Cross-Fostering Study,? Nature 340, no. 6234 (1990).
Kimberly G. Noble, Nim To t t e n h a m , a n d B . J . C a s e y
88 THE FUTURE OF CHILDREN
52. Michael J. Meaney, ?Maternal Care, Gene Expression, and the Transmission of Individual Differences in
Stress Reactivity across Generations,? Annual Review of Neuroscience 24 (2001); Stephen J. Suomi, ?Gene-
Environment Interactions and the Neurobiology of Social Conflict,? Annals of the New York Academy of
Sciences 1008 (2003).
53. Eric Turkheimer and others, ?Socioeconomic Status Modifies Heritability of IQ in Young Children,? Psychological
Science 14, no. 6 (2003).
54. Rosenzweig and Bennett, ?Psychobiology of Plasticity? (see note 33).
55. Kimberly A. Kerns, Karen Eso, and Jennifer Thomson, ?Investigation of a Direct Intervention for Improving
Attention in Young Children with ADHD,? Developmental Neuropsychology 16, no. 2 (1999); Klingberg,
Forssberg, and Westerberg, ?Increased Brain Activity in Frontal and Parietal Cortex Underlies the Development
of Visuospatial Working Memory Capacity during Childhood? (see note 5).
56. M. Rosario Rueda and others, ?Training of Attention in 4 Year Old Children,? paper presented to the Cognitive
Neuroscience Society, San Francisco, 2004.
57. John C. Buckner, E. Mezzacappa, and W. R. Beardslee, ?Characteristics of Resilient Youths Living in
Poverty: The Role of Self-Regulatory Processes,? Development and Psychopathology 15, no. 1 (2003).
58. Bruce M. McCandliss and Michael Wolmetz, ?Developmental Psychobiology of Reading Disability,? Biological
Psychiatry (forthcoming).
59. Barbara R. Foorman and others, ?How Letter-Sound Instruction Mediates Progress in First-Grade Reading
and Spelling,? Journal of Educational Psychology 83, no. 4 (1991); Joseph K. Torgesen and others, ?Intensive
Remedial Instruction for Children with Severe Reading Disabilities: Immediate and Long-Term Outcomes
from Two Instructional Approaches,? Journal of Learning Disabilities 34 (2001); Frank R. Vellutino and others,
?Cognitive Profiles of Difficult-to-Remediate and Readily Remediated Poor Readers: Early Intervention
as a Vehicle for Distinguishing between Cognitive and Experiential Deficits as Basic Causes of a Specific
Reading Disability,? Journal of Educational Psychology 88, nos. 601?38 (1996).
60. Simos and others, ?Dyslexia-Specific Brain Activation Profile Becomes Normal Following Successful Remedial
Training? (see note 16).
61. Elise Temple and others, ?Neural Deficits in Children with Dyslexia Ameliorated by Behavioral Remediation:
Evidence from Functional MRI,? Proceedings of the National Academy of Sciences, USA 100, no. 5
(2003).
62. Cecilia E. Rouse and Alan B. Krueger, ?Putting Computerized Instruction to the Test: A Randomized Evaluation
of a ?Scientifically-Based? Reading Program,? Economics of Education Review (forthcoming).
63. Bennett A. Shaywitz and others, ?Development of Left Occipitotemporal Systems for Skilled Reading in
Children after a Phonologically-Based Intervention,? Biological Psychiatry 55 (2004).
64. Mark A. Eckert, Linda J. Lombardino, and Christiana M. Leonard, ?Planar Asymmetry Tips the Phonological
Playground and Environment Raises the Bar,? Child Development 72, no. 4 (2001).
N e u r o s c i e n c e P e r s p e c t i v e s o n D i s p a r i t i e s i n S c h o o l R e a d i n e s s a n d C o g n i t i v e A c h i e v e m e n t
VOL. 15 / NO. 1 / SPRING 2005 89
Low Birth Weight and School Readiness
Nancy E. Reichman
Summary
In the United States black women have for decades been twice as likely as white women to give
birth to babies of low birth weight who are at elevated risk for developmental disabilities. Does
the black-white disparity in low birth weight contribute to the racial disparity in readiness?
The author summarizes the cognitive and behavioral problems that beset many low birth
weight children and notes that not only are the problems greatest for the smallest babies, but
black babies are two to three times as likely as whites to be very small. Nevertheless, the racial
disparities in low birth weight cannot explain much of the aggregate gap in readiness because
the most serious birth weight?related disabilities affect a very small share of children. The author
estimates that low birth weight explains at most 3?4 percent of the racial gap in IQ scores.
The author applauds the post-1980 expansions of Medicaid for increasing rates of prenatal care
use among poor pregnant women but stresses that standard prenatal medical care cannot improve
aggregate birth outcomes substantially. Smoking cessation and nutrition are two prenatal
interventions that show promise. Several early intervention programs have been shown to improve
cognitive skills of low birth weight children. But even the most promising programs can
narrow the readiness gap only a little because their benefits are greatest for heavier low birth
weight children and because low birth weight explains only a small share of the gap.
The author stresses the importance of reducing rates of low birth weight generally and of extending
to all children who need them the interventions that have improved cognitive outcomes
among low birth weight children. But because black infants are more likely to be born at
the lowest birth weights, preventing low birth weight?when researchers learn how to?is
likely to be more effective than early intervention in narrowing birth weight?related racial gaps
in school readiness.
VOL. 15 / NO. 1 / SPRING 2005 91
www.future of children.org
Nancy E. Reichman is associate professor of pediatrics at the Robert Wood Johnson Medical School, University of Medicine and Dentistry
of New Jersey. She gratefully acknowledges eight continuous years of funding from the National Institute of Child Health and Human Development,
without which this article would not have been possible. She also wishes to thank Nigel Paneth for his inspiring chart of survival
rates of very low birth weight infants over time; Elana Broch for locating the 1980 vital statistics report and countless other materials over
many years; Jennifer Borkowski for cheerfully and meticulously extracting, crunching, and compiling data from an assortment of historical
vital statistics reports; Julien Teitler for numerous hours of brainstorming; Elaine Barfield, Kavita Bhanot, and Christopher Ramos for helping
to synthesize portions of the literature; Anne Case and many others for helpful comments on an earlier draft; Rehan Shamim for compiling
and formatting the references; Lenna Nepomnyaschy and Ofira Schwartz-Soicher for reviewing the final draft.
In the United States, black women
have for decades been twice as likely
as white women to give birth to babies
of low birth weight?those
weighing less than 2,500 grams, or
about 5.5 pounds. Not only is low birth
weight a leading cause of infant mortality, but
infants who survive are at elevated risk for
many long-term health conditions and developmental
disabilities that can impair school
readiness. The black-white disparity in low
birth weight is so large and so persistent that
it raises the question of whether it contributes
to racial disparities in children?s cognitive
abilities and in readiness.
This article, which focuses on the effect of
low birth weight on the racial gap in test
scores, consists of six sections. The first provides
a brief overview of low birth weight in
the United States?definition, trends, and associated
rates of survival and child disability.
The second discusses disparities in low birth
weight by race, ethnicity, and nativity, as well
as survival rates by race. The third section, the
heart of the paper, examines the link between
low birth weight and school readiness. It reviews
the cognitive and behavioral problems
that beset many low birth weight children,
noting that the problems are greatest for the
smallest babies and that black babies are
much more likely than white babies to be very
small. It also explores the effect of birth
weight on the black-white gap in readiness
and confirms earlier findings that the racial
disparity in birth weight explains only a few
percentage points of the aggregate gap. The
fourth section looks at the determinants of
low birth weight, focusing on those that vary
by race. The fifth considers past efforts to
tackle the problem of low birth weight
through prevention or through amelioration
of its adverse consequences. It highlights
early intervention programs that have been
shown to improve cognitive outcomes among
low birth weight children and thus close at
least a small portion of the readiness gap. The
final section summarizes the article?s key findings,
highlights important implications, and
offers recommendations.
Low Birth Weight in
the United States
Low birth weight is a widely used and much
studied marker of infant health.1 It is well
measured, reliably recorded, and readily
available from vital statistics files and many
other data sets. Birth weight is often categorized
as very low (less than 1,500 grams, or
about 3.3 pounds), low (less than 2,500
grams), or normal (2,500 grams or more).
Further distinctions include extremely low
(less than 1,000 grams) and moderately low
(1,500?2,499 grams) birth weight. Births can
also be characterized by gestational age: very
preterm (less than 32 weeks), preterm (less
than 37 weeks), and term (37 weeks or
more). These terms and their definitions are
summarized in table 1, along with the corresponding
rates of births in the United States
in 2000. Babies considered small for gestational
age (SGA) or growth retarded are typically
below the 10th percentile in sex-specific
birth weight for gestational age. All low birth
weight babies are preterm or growth retarded
(they can be both), and virtually all
very low birth weight babies are preterm.
Trends
Babies born in the United States are more
likely to be low birth weight than those born
in almost every other developed country.2
Low birth weight is the second leading cause
of infant mortality in the United States after
birth defects, and surviving infants are at elevated
risk for debilitating medical conditions
and learning disorders.3 Figure 1 shows rates
of low birth weight, very low birth weight,
Nancy E. Reichman
92 THE FUTURE OF CHILDREN
and infant mortality (death before age one) in
the United States from 1980 to 2000. Thanks
to increased specialization in delivering maternal
and newborn health care and to advances
in neonatal intensive care technology,
the United States made substantial progress
in reducing the infant mortality rate over this
period, although its gains have lagged behind
those of other developed countries.4 Rates of
low and very low birth weight, meanwhile, increased
slightly, owing partly to the increasing
prevalence of multiple births; the rate of
low birth weight among singleton births has
remained steady, at about 6 percent.5
Low birth weight babies are much more likely
to survive today than they once were. Since
1960, survival rates have increased dramatically
for very low and extremely low birth
weight babies born in the United States (figure
2). Although less than 10 percent of extremely
low birth weight singleton infants
born in 1960 lived to their first birthday, that
figure increased to 27 percent for those born
in 1980 and to 57 percent for those born in
2000.6 And while fewer than half of very low
birth weight (defined here as 1,000?1,499
grams) singleton babies born in 1960 survived,
by 2000 the share surviving had increased to
Low Birth We i g h t a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 93
Table 1. Definitions of Low Birth Weight and Related Outcomes, United States
Term Definition Percent of live births, 2000
Normal birth weight At least 2,500 grams 92.4
Low birth weight (LBW) Less than 2,500 grams 7.6
Moderately low birth weight 1,500?2,499 grams 6.2
Very low birth weight (VLBW) Less than 1,500 grams 1.4
Extremely low birth weight (ELBW) Less than 1,000 grams 0.7
Preterm Less than 37 weeks? gestation 11.6
Very preterm Less than 32 weeks? gestation 1.9
Source: Joyce A. Martin and others, ?Births: Final Data for 2000,? National Vital Statistics Reports 52, no. 10 (Hyattsville, Md.: National
Center for Health Statistics, February 12, 2002).
0
2
4
6
8
10
12
14
Infant Mortality
(per 1,000)
VLBW (%)
LBW (%)
2000 1998 1996 1994 1992 1990 1988 1986 1984 1982 1980
Percent
Figure 1. Low Birth Weight, Very Low Birth Weight, and Infant Mortality Rates, United
States, 1980-2000
Source: Centers for Disease Control and Prevention, Morbidity and Mortality Weekly Report, vol. 51(27): 589?92 (www.cdc.gov/ mmwr/
preview/mmwrhtml/mm5127a1.htm).
more than 90 percent. Likewise, the survival
rates of moderately low birth weight singleton
infants increased from 91 percent in 1980 to
98 percent in 2000.7 The new survivors, however,
are at high risk for health and developmental
problems, as discussed below.
Survival and Disability
The majority of moderately low birth weight infants
thrive, suffering few or no problems. It is
the lightest babies who are most at risk of disabilities,
both cognitive and physical, that can
impair school readiness. Of the many child
health conditions associated with low birth
weight, perhaps the most potentially disabling is
cerebral palsy, a group of disorders characterized
by the inability to control movement and
often accompanied by cognitive impairments.8
Preterm very low birth weight infants are up to
30 percent more likely to develop cerebral palsy
than are babies born at term.9 Other serious
conditions associated with low birth weight or
preterm birth include mental retardation, respiratory
distress syndrome (RDS), bronchopulmonary
dysplasia (BPD), retinopathy of
prematurity (ROP), and deafness. RDS and
BPD can lead to feeding difficulty, recurrent
respiratory infections, asthma, and growth
delay.10 ROP, a disorder caused by abnormal
growth of blood vessels in the eye, can lead to
blindness.11 All these disabilities can impair
learning and inhibit a child?s school readiness.
Almost without exception, the prevalence
of these disabling conditions increases as
birth weight decreases.
A recent review of forty-two studies of infants
born after 1970 found no change between
1976 and 1990 in the prevalence of major
neurodevelopmental disabilities among ex-
Nancy E. Reichman
94 THE FUTURE OF CHILDREN
100
90
80
70
60
50
40
30
20
10
0
1,500?2,499g
1,000?1,499g
2000 1980 1960
91
97 98
93
79
45
9
27
57
Percent
Figure 2. One-Year Survival Rates of Singleton Low Birth Weight Infants,
by Birth Weight, United States, 1960, 1980, and 2000
Sources: Data for 1960 are from U.S. Department of Health, Education, and Welfare, Public Health Service, Office of Health Research, Statistics,
and Technology, ?A Study of Infant Mortality from Linked Records, by Birth Weight, Period of Gestation, and Other Variables, United
States, 1960 Live-Birth Cohort,? (PHS) 79-1055 (Hyattsville, Md.: National Center for Health Statistics, May 1972). Data for 1980 are
from U.S. Department of Health and Human Services, Public Health Service, ?National Infant Mortality Surveillance (NIMS) 1980,? (Atlanta:
Centers for Disease Control, Center for Chronic Disease Prevention and Health Promotion, Division of Reproductive Health, December
1989). Data for 2000 are from National Center for Health Statistics, ?Live Births, Infant Deaths, and Infant Mortality Rates by Plurality,
Birthweight, Race of Mother, and Gestational Age: United States, 2000 Period Data,? table LFWK 46 (www.cdc.gov/nchs/datawh/statab/unpubd/
mortabs.htm#Linked).
Notes: This figure was adapted from a slide provided by Nigel Paneth, M.D., M.P.H., Michigan State University. Only single births are used.
The figures for 1960 were calculated using cutoffs of tremely immature (26 weeks or less) and extremely
small (800 grams or less) survivors.
Throughout that period, cerebral palsy affected
12 percent of extremely immature and
8 percent of extremely small survivors; mental
retardation affected 14 percent of each
group; 8 percent of each group was blind;
and 3 percent of each group was deaf. Overall,
22 percent of extremely immature survivors
and 24 percent of extremely small survivors
had at least one major disability.12
Disparities in Low Birth Weight
by Race, Ethnicity, and Nativity
The black-white disparity in low birth weight
in the United States is glaring and persistent.
In 2000, 13 percent of babies born to black
mothers were low birth weight, compared to
6.5 percent of those born to white mothers.13
(By contrast, rates of low birth weight for the
other racial groups reported by the National
Center for Health Statistics were close to that
of whites: 6.8 percent among American Indians
and 7.3 percent among Asians and Pacific
Islanders.)14 The two-to-one disparity between
blacks and whites has persisted for
more than forty years, exists at most maternal
age ranges, cannot be explained by differences
in rates of multiple births, and cannot be explained
by socioeconomic status alone.15 Even
infants born to college-educated black women
are at much greater risk than infants born to
college-educated white women of being low
birth weight.16 Black mothers were 63 percent
more likely to have preterm deliveries than
white mothers (17.3 percent as against 10.6
percent) in 2000.17 The rates of small-forgestational-
age births among infants born at
term in 1998 were 17.4 percent among blacks
and 9.0 percent among whites.18
Ethnicity
Rates of low birth weight also vary among
women of different ancestral origins. The
rate for women of Hispanic descent was 6.4
percent in 2000, on par with the rate for
whites. But within that broad group, rates
differ widely. In 2000, women of Cuban and
Mexican descent had low birth weight rates
of 6.5 percent and 6.0 percent, respectively,
while Puerto Ricans had a rate of 9.3 percent.
19 The disparity between Puerto Ricans
and Mexicans has baffled researchers because
both groups are at high risk for adverse
outcomes based on their socioeconomic status,
and island-born Puerto Ricans, as U.S.
citizens, have greater access than foreignborn
Mexicans to Medicaid.20 The disparity
may have to do with unmeasured differences
in culture, diet, stress, or lifestyle.21 Researchers
have termed the unexpectedly favorable
rates among Mexican American
women, despite their socioeconomic disadvantages
and comparatively low use of prenatal
care, the epidemiologic or Hispanic paradox.
22 Explaining this paradox could provide
clues about how to blunt the negative effects
of poverty on birth outcomes of other disadvantaged
groups. Blacks of Puerto Rican or
other Hispanic ethnicity have a lower probability
of low birth weight than blacks who are
non-Hispanic, but very few (3 percent) of the
622,598 births to black mothers in 2000 were
to mothers who identified themselves as Hispanic.
23
Several researchers, notably Gosta Rooth,
have questioned the standard 2,500 gram
cutoff for low birth weight, arguing that it
does not account for variation in mean birth
weights across countries that may be due to
differences in, for example, maternal
height.24 That threshold may likewise not be
appropriate for all racial and ethnic groups in
the United States, but the ?natural? underlying
distributions are not known and may
themselves be determined by factors such as
health and socioeconomic status rather than
Low Birth We i g h t a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 95
biological predisposition. Nigel Paneth, in an
excellent summary of this issue, suggests that
there is not enough evidence to dismiss the
glaring racial disparities in low birth weight
in the United States as ?normal.?25
Nativity
In 2000 some 80 percent of U.S. births to
white women and 88 percent of births to
black women were to mothers born in the
United States.26 Many groups of immigrant
mothers, particularly Mexicans, make less
use of prenatal care and other health services
than their U.S.-born ethnic counterparts because
of multiple legal, language, socioeconomic,
and cultural barriers.27 Yet the birth
outcomes of Mexican immigrants are even
more favorable than those of U.S.-born Mexican
mothers. In fact, for virtually every racial
and ethnic group in the United States, immigrants
have better birth outcomes than U.S.-
born mothers.28 Thus, although immigrants
encounter numerous barriers to prenatal
care, they have offsetting health, social, or
lifestyle advantages that promote favorable
birth outcomes.
Several studies have analyzed birth outcomes
of black women by nativity.29 Of particular interest,
Richard David and James Collins
found that African-born black mothers have
rates of low birth weight much closer to those
of U.S.-born white mothers than to those of
U.S.-born black mothers of predominantly
African descent. This suggests that blackwhite
disparities in low birth weight may be
due to social and environmental factors
rather than biological predisposition, although
one cannot rule out the possibility
that the differences are due to selective
migration.
Low Birth Weight, Survival, and Race
Given the large disparity in low birth weight
between blacks and whites and the small disparities
between whites and other racial
groups and between whites and Hispanics, in
the remainder of this article I focus on blackwhite
differences in school readiness. Whenever
possible I focus on the lowest birth
weight infants, because although they compose
small proportions of all births, they suffer
the highest rates of disability and therefore
have the poorest long-term prognosis
for school readiness and academic achievement.
As figure 3 shows, the rate of low birth
weight among blacks (single births) was the
same in 2000 as in 1980; that for whites in-
Nancy E. Reichman
96 THE FUTURE OF CHILDREN
Figure 3. Extremely Low Birth Weight, Very Low Birth Weight, and Low Birth Weight
Rates by Race, United States, 1980 and 2000
Sources: See figure 2.
Notes: Only single births are used. For 1980, race is based on both parents? races from birth certificates; for 2000, on mother?s race from
birth certificate.
0
2
4
6
8
10
12
1,500-2,499g
1,000-1,499g
BLACK WHITE BLACK WHITE
1980 2000
Percent
creased slightly.30 The black-white disparity
occurs across all low birth weight groups but
is even larger for the lowest weight groups.
And while the overall rates of low birth
weight have remained constant, the shares of
births in lowest weight groups have increased,
particularly for blacks. Between
1980 and 2000 the rate of extremely low
birth weight rose almost 50 percent among
blacks and a third among whites, while the
rate of very low birth weight rose about 25
percent among blacks and 15 percent among
whites. These higher rates may reflect increased
obstetric intervention that prevents
fetal loss. Overall reported fetal deaths at 20
or more weeks? gestation declined 12 percent
over 1990?2000 alone; the decreases for
non-Hispanic whites and non-Hispanic
blacks were 10 percent and 5 percent,
respectively.31
The rate of infant mortality (death in the first
year) has fallen steadily for both blacks and
whites over the past twenty-five years. In
1980, 18 out of 1,000 black singleton babies
did not live to their first birthday; by 2000
that figure had fallen to 12 out of 1,000. For
white babies the comparable rates were 9 out
of 1,000 in 1980 and 5 out of 1,000 in 2000.32
As with low birth weight, the two-to-one
black-white disparity in infant mortality has
persisted over time, although the percentage
decline in infant mortality has been greater
among whites than among blacks.
Birthweight-specific survival rates are remarkably
equivalent for black and white singletons.
In the past, black low birth weight
infants had a paradoxical survival advantage,
perhaps owing to differences in fetal health
and differential rates of fetal loss. In 1980, 83
percent of black and 76 percent of white singleton
infants of very low birth weight (here,
1,000?1,499 grams) survived their first year;
for extremely low birth weight infants, the
survival rates were 29 percent for blacks and
27 percent for whites. In 2000, survival rates
for very low birth weight infants were 93 percent
for whites and 94 percent for blacks; and
for extremely low birth weight babies, 58 percent
for whites and 57 percent for blacks.
Even taking into account multiple births, recent
figures show no indication of racial disparities
in birth weight?specific survival or in
birth weight?specific neonatal survival (the
first 28 days of life).33 The lifesaving advantages
of neonatal care thus appear to be
color-blind, at least in the aggregate. (These
figures do not speak to whether there are disparities
in newborn care more generally.)
However, because black infants are much
more likely to fall into the lowest weight
groups, a disproportionate fraction of black
survivors is at high risk for adverse health and
developmental outcomes.
Among survivors born in 2000 (including
multiple births), the share of black infants
who were extremely low birth weight is 1.00
percent, more than three times that for
whites (0.32 percent). The difference is similar
for very low birth weight babies (2.31 percent
for blacks, as against 0.89 percent for
whites).34 Thus among children born in 2000
who survived their first year of life, black
children are more than two and a half times
as likely as white children to have been extremely
or very low birth weight?and therefore
to be at risk of serious cognitive delays
that could affect school readiness and academic
achievement when they enter kindergarten
in 2005.35
Low Birth Weight and School
Readiness
Extensive research confirms that low birth
weight children are at greater risk for cognitive
and school performance problems than
Low Birth We i g h t a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 97
are their normal birth weight peers, and that
the risk for adverse outcomes increases as
birth weight decreases.36 A meta-analysis of
case control studies reported from 1980 to
November 2001 found that the mean IQ for
school-aged children born very preterm is
approximately two-thirds of a standard deviation
below that of controls who were born at
term.37 A population-based study using
linked birth certificate and school records
from Florida found that the risk of specific
school-identified disabilities increases as
birth weight decreases.38 Enrollment in special
education also follows a birth weight gradient,
with the lightest babies being most
likely to be placed in such programs.39 While
all of these findings are based on cohorts
born before the major advances in neonatal
intensive care of the 1990s, research on later
cohorts yields similar results.40
Children born preterm have greater difficulty
completing tasks involving reading,
spelling, and math than their full-term peers,
though math scores are more consistently related
to preterm birth or very low birth
weight than are reading achievement
scores.41 Preterm children tend to have language
difficulties related to grammar and abstraction.
42 They also tend to be more inattentive,
aggressive, and hyperactive, as well
as less able to handle leadership roles than
their full-term peers.43
Some cognitive deficits are the direct result
of medical disorders.44 Compromised motor
skills in many preterm infants, for example,
may lead to learning disabilities and handicaps.
45 Studies of the brains of preterm and
full-term children have identified areas that
correspond to the cognitive deficits observed.
Brain volume tends to be reduced, resulting
in larger ventricles containing more cerebrospinal
fluid, thinning of the corpus callosum
(which indicates less white matter), and a reduction
in gray matter. The sensorimotor cortex,
amygdala, and hippocampus are also
often reduced.46 These anatomical deficiencies
are most likely a result of immaturity,
physiological instabilities, or stressful experiences
as neonates.47
Birth weight may also have indirect effects on
cognitive development through parenting.
The medical, developmental, and behavioral
problems of a very light infant may heighten
parental stress, which may in turn impair the
child?s learning. Research in this area is in its
infancy. According to one recent study, mothers
of very low birth weight infants suffered
more psychological distress than mothers of
term infants at one month, at two years, and
at three years, with the severity of stress positively
related to the child?s developmental
outcomes.48
Collectively, past studies based on hospital or
regional cohorts have found that among extremely
low birth weight infants, 8 to 18 percent
have IQ scores under 70 (a cutoff often
used to define mental retardation), and 25 to
29 percent have IQs in the 70?84 (borderline)
range at school age (generally ages six or
eight to ten). The corresponding figures for
very low birth weight infants (here,
1,000?1,499 grams) are 5 percent and 19 percent;
for moderately low birth weight infants,
5 percent and 17 percent; and for normal
Nancy E. Reichman
98 THE FUTURE OF CHILDREN
Children born preterm
have greater difficulty
completing tasks involving
reading, spelling, and math
than their full-term peers.
birth weight infants, 0 to 4 percent and 4 to
14 percent (the figures for very low and moderately
low birth weight infants are based on
only one study).49
Birth Weight and Socioeconomic Status
Birth weight is but one of many risk factors
for cognitive impairment. One of the most
salient risk factors is low socioeconomic status.
Disentangling the effects of birth weight
from those of the many socioeconomic disadvantages
linked with low birth weight is
difficult. Research to date indicates that very
low birth weight (and?much less so?
moderately low birth weight) does have
independent deleterious effects on early
cognitive outcomes, such as IQ and PIAT
scores.50 But while it might be interesting
from a variety of vantage points to disentangle
the effects of birth weight and socioeconomic
status, the two are so highly correlated
that it may not be relevant for policy
purposes to do so.
Low Birth Weight and Aggregate
Educational Outcomes
Clearly, individual children born low birth
weight can be seriously disadvantaged with
respect to schooling. But because most serious
birth weight?related disabilities tend to
occur at the lowest weight ranges and therefore
affect a very small proportion of children,
low birth weight may not explain much
of the observed variation in educational attainment
at the aggregate level. A recent
study of children born in the 1958 British
birth cohort, for example, found that children
born at low birth weight passed significantly
fewer O-level exams. But being born at low
birth weight, or being born to a mother who
smoked during pregnancy (also a predictor of
poor educational outcomes), explained only
2.5 percent of the variation in O-level
results.51
Low Birth Weight and the Black-White
Gap in Test Scores
Only two studies of which I am aware have
presented data indicating the potential effect
of low birth weight on racial test score gaps.
Yolanda Padilla and her coauthors, in a study
using National Longitudinal Survey of Youth
(NLSY) child data and focusing on the effects
of the Mexican-American birth weight advantage
on early childhood development, found
that low birth weight explains less than 1 percent
of the (unadjusted) black-white gap in
scores on the Peabody Picture Vocabulary
Test-Revised (PPVT-R) among three- and
four-year-olds in the late 1980s and early
1990s.52 Jeanne Brooks-Gunn and her coauthors
presented a similar estimate in a recent
analysis of the contributions of family and test
characteristics to the black-white test score
gap.53 Also using NLSY child data, they found
that low birth weight and gender together explain
less than 2 percent of the unadjusted
racial gap in PPVT-R scores at age five.
My own estimate of the potential impact of
birth weight on the racial gap in one test of
cognitive ability?full-scale IQ score?is similar,
though somewhat higher. My subject is all
black and white infant survivors born in 2000,
including multiples. In contrast to Padilla and
Brooks-Gunn I do not use the NLSY data, because
although that data set has actual test
scores, it may underrepresent the very lightest
babies. Instead I use vital statistics data,
which provide exact race-specific birth weight
distributions for surviving infants in the
United States, though test scores must be imputed.
I assigned an IQ score to each survivor,
based on the infant?s birth weight. I then computed
the racial gap in imputed IQ scores and
divided this figure by the total observed racial
gap in IQ scores, to compute the maximum
proportion of the overall gap that can be explained
by birth weight.54 Using various dis-
Low Birth We i g h t a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 99
tributions of IQ scores based on past research
and a range of assumptions, I found that birth
weight explains a maximum of 3 to 4 percent
of the racial gap in IQ scores, or one-half a
point in IQ.
Determinants of Low Birth Weight
Researchers have identified and analyzed
many social, medical, and behavioral risk factors
for low birth weight, some of which
could contribute to racial disparities in low
birth weight, and ultimately to school readiness.
Many of these risk factors are intricately
intertwined, and for the most part I will not
attempt to establish or disentangle causal
effects.
Socioeconomic Status
Women of low socioeconomic status are at increased
risk for delivering low birth weight
babies, whether socioeconomic status is defined
by income, occupation, or education.55
Education may also have independent effects,
above and beyond income, because
more highly educated mothers may know
more about family planning and healthy behaviors
during pregnancy. In 1998, the rate of
low birth weight among mothers with less
than a high school education was 9 percent,
as against 7.9 percent among high school
graduates, and 6.5 percent among mothers
with at least some college.56 In 2000, 78.6
percent of white women giving birth, and
74.5 percent of black women giving birth,
had twelve or more years of education.57
Black Americans are much more likely than
whites to come from a disadvantaged socioeconomic
background, but that does not fully
explain the racial disparity in low birth
weight.58
Marital Status
Marital status is also a key correlate of birth
weight. In 1992, the rate of low birth weight
babies among unmarried mothers in the
United States was 10.4 percent, as against 5.7
percent among married mothers.59 In 2000,
27.1 percent of low birth weight babies born
to white mothers and 68.5 percent of low
birth weight babies born to black mothers
had unmarried parents.60 The marital status
disparity may reflect either the greater likelihood
of unmarried mothers to be poor or
other characteristics that vary by marital
status.61
Maternal Age
In 2000, 19.7 percent of births to black
women and 10.6 percent to white women in
the United States were to teens. The rate of
low birth weight babies among teen mothers
was 35 percent higher than that among mothers
aged twenty to twenty-nine (9.6 percent
as against 7.1 percent). The rate among the
youngest teens?those fifteen and younger?
was 14.1 percent, higher than in any age
group except forty-five to fifty-four.62 Teen
mothers? birth weight disadvantage has several
explanations. A pregnant teenager who is
still growing may compete for nutrients with
the fetus. Becoming pregnant within two
years after menarche increases the risk for
preterm delivery.63 Many teen pregnancies
are unplanned, unwanted, or discovered late,
and pregnant teens are more likely than older
mothers to be poor, to be undereducated, or
to lack access to resources or services?all, in
themselves, risk factors for low birth
weight.64
In 1992 Arline Geronimus found, surprisingly,
that black teen mothers seem to have a
paradoxical advantage in birth outcomes over
older black mothers. She speculated that this
finding may be due to ?weathering? among
black women?more rapid age-related deterioration
in health than among white women
because of greater cumulative exposure to
Nancy E. Reichman
100 THE FUTURE OF CHILDREN
harsh living conditions. Thus young maternal
age may not be as much a risk factor among
black mothers as it is among whites.65 Unadjusted
national figures for black mothers
from 2000 do not reflect this pattern; low
birth weight rates among black mothers were
lowest among mothers in their twenties.66 If
the national sample were restricted to disadvantaged
black mothers, however, the Geronimus
weathering pattern might become
apparent.
On the other end of the age spectrum,
women who give birth in their late thirties or
older are also at increased risk for having low
birth weight babies. In 2000, 9.7 percent of
births to black women and 13.9 percent of
births to white women in the United States
were to women aged thirty-five and over.67
For these women the risks are biological:
older ova and a greater likelihood of medical
risk factors such as hypertension.68 Older
women also have more unintended pregnancies?
itself a risk factor for low birth
weight?than do women in their twenties
and early thirties.69
One study found that women aged thirty
and older are at greater risk for poor birth
outcomes than teens of the same race,
though offsetting factors such as higher
socioecoomic status mask this risk.70 That
same study, which controlled for such socioeconomic
characteristics as whether the
birth was covered by Medicaid, found evidence
of the Geronimus weathering phenomenon.
Black mothers aged fifteen to
nineteen were at lower risk of delivering low
birth weight babies than were black mothers
in their twenties. Given the complicated relation
between maternal age and low birth
weight, it is difficult to assess the extent to
which black mothers are at increased risk in
this regard.
Medical Conditions
Among the medical risk factors for low birth
weight and preterm birth are prior low birth
weight or preterm delivery, cervical abnormalities,
hypertension, anemia, and bacterial
infections.71 Chronic physical or psychological
stress also increases the risk.72 Among the
risk factors for fetal growth retardation are
previous low birth weight births, infections,
sexually transmitted diseases, poor maternal
hematological status, hypertension-related
complications, renal disease, heart disease,
third trimester bleeding, and sickle cell disease.
73 Nutritional inadequacy can also impair
fetal growth.74
Most, but not all, of these medical risk factors
are more prevalent among blacks than
whites. Most are rare. In 2000, for example,
3.8 percent of black mothers and 2.1 percent
of white mothers were anemic during pregnancy;
1.4 percent of black mothers and 0.7
percent of white mothers had chronic hypertension.
Black mothers had higher rates of
acute or chronic lung disease, genital herpes,
hydramnios or oligohydramnios (too little or
too much amniotic fluid), hemoglobinopathy
(a blood disorder), pregnancy-associated hy-
Low Birth We i g h t a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 101
One study found that women
aged thirty and older are at
greater risk for poor birth
outcomes than teens of the
same race, though offsetting
factors such as higher
socioeconomic status mask
this risk.
pertension, eclampsia, incompetent cervix,
and previous preterm babies or growthretarded
infants. White mothers had higher
rates of cardiac disease, renal disease, Rh
sensitization, and uterine bleeding.75 Bacterial
vaginosis, a mild bacterial infection more
common among black women than white
women, has been linked with preterm delivery
of low birth weight infants.76
Prenatal Substance Use
Maternal cigarette smoking during pregnancy
decreases fetal growth rates and substantially
increases the risks of spontaneous
abortion, preterm delivery, low birth weight,
placental ruptures, placenta praevia, and
perinatal death. Prenatal alcohol and drug
use are also linked with poor birth outcomes,
though the relationships are less clear-cut
and not as dose-response specific as that of
smoking.77 Substance abuse during pregnancy,
particularly of alcohol and illicit drugs,
is notoriously underreported. Based on reported
rates of smoking, black mothers do
not appear to be at increased risk for low
birth weight. In 2000, 9.1 percent of black
mothers and 13.2 percent of white mothers
in the United States reported smoking cigarettes
(at all) during pregnancy. The proportion
of black and white mothers who reported
consuming alcohol at all during
pregnancy according to birth records in 2000
was virtually identical?about 1 percent of
each group.78 However, these rates are
nowhere near the proportion (16.3 percent)
of pregnant women aged eighteen to fortyfour
who reported alcohol consumption in
the past month in the 1995 Behavioral Risk
Factor Surveillance System.79 For this reason,
prenatal alcohol consumption has since
been removed from the U.S Standard Certificate
of Live Birth. In the 2001 National
Household Survey on Drug Abuse, reported
rates of current illicit drug use were similar
among white (4.0 percent) and black (3.7
percent) pregnant women.80
Intergenerational Health
Several studies have found strong associations
between parents? (generally mothers?)
birth weight and the birth weight of their
child.81 A recent study comparing maternal
cousins (children whose mothers are sisters),
and thus filtering out much of the confounding
effect of socioeconomic status, found that
maternal and paternal low birth weights together
explain a much larger share of the
racial disparity in low birth weight than do individual
characteristics and socioeconomic
variables combined.82 This finding suggests
that there is a biological transmission of low
birth weight across generations, which may
contribute to racial differences in low birth
weight. This is an important finding that can
be used to target interventions, but given the
strong association between birth weight and
socioeconomic status, it should not be used
to dismiss racial disparities as immutable.
Promising Directions for Future
Research on Risk Factors
Other risk factors warrant further study and
ultimately may offer strategies for reducing
rates of low birth weight and narrowing racial
disparities in low birth weight and school
readiness. For the most part, research on
these risks is in its infancy, and the associations
being explored should not be interpreted
as causal.
MATERNAL LIFESTYLE. Despite the beneficial
effects of employment on income, mothers
who work in strenuous occupations, including
those that involve prolonged
standing, are at heightened risk for both
preterm delivery and having low birth weight
babies.83 Occupational exposures to toxic
substances and solvents have also been linked
Nancy E. Reichman
102 THE FUTURE OF CHILDREN
to preterm delivery.84 Given that a greater
share of black women than white women (in
2002, 9 percent as against 5 percent) in the
United States work as operators, fabricators,
and laborers, black mothers may be more
likely than white mothers to encounter strenuous
working conditions and toxic
exposures.85
NEIGHBORHOODS. Living in a poor neighborhood
may pose health risks above and beyond
those associated with individual
poverty. Houses and other buildings in poor
neighborhoods tend to be old and in poor
condition; environmental toxins tend to be
high; and access to medical care and other
services tends to be limited.86
One study of Chicago in 1990 found that living
in different neighborhoods accounted for
as much as 30 percent of the difference in
mean birth weight between non-Hispanic
blacks and whites, though it is unclear
whether these ?neighborhood effects? reflect
social, economic, or physical characteristics
of neighborhoods or unobserved individuallevel
risk factors that vary by neighborhood.87
Neighborhood socioeconomic characteristics,
such as census tract?level income, are important
predictors of low birth weight, even after
controlling for many individual-level characteristics.
88 In Chicago, violent crime in neighborhoods
has been found to have a negative
association with birth weight, while a combined
measure of social interaction and community
involvement has a positive association.
89 Many studies have linked low birth
weight to residential environmental exposures,
including air pollution, substances in
drinking water, and industrial chemicals.90
Three-quarters of the residents of highpoverty
neighborhoods in the United States
are minorities, and the number of blacks living
in poor areas increased from 2.4 million
in 1970 to 4.2 million in 1990. Thus black
women are at high risk for delivering low
birth weight babies on the basis of the neighborhoods
in which they live.91
PATERNAL FACTORS. Finally, a growing body
of research suggests that paternal behaviors
and occupational exposures before conception
may affect infant health. Male reproductive
toxicity can have three mechanisms?
nongenetic (seminal fluid), genetic (gene
mutations or chromosomal abnormalities),
and epigenetic (effects on gene expression,
genomic imprinting, or DNA methylation).92
One study linked paternal drinking and low
birth weight, but its finding has not been
replicated.93 Others have found associations
between paternal smoking and low birth
weight, although it is difficult to disentangle
potential direct effects of paternal smoking
from indirect effects through maternal exposure
to secondhand smoke.94 Paternal occupational
exposures are also a risk factor. Excess
rates of preterm delivery, growth
retardation, and low birth weight have been
found in occupations that involve paternal exposure
to pesticides, solvents, and lead.95 In
2002, 28 percent of employed black men, as
against 16 percent of employed white men in
the United States, worked as operators, fabricators,
and laborers, perhaps making black
fathers more likely than white fathers to be
exposed to toxic substances at work.96
Interventions
Child health policymakers and practitioners
have implemented many programs both to
prevent low birth weight and to improve the
life chances of low birth weight babies, especially
in the areas of school readiness and
achievement. To the extent that the programs
succeed, they could help narrow racial gaps
in school readiness by as much as 3 to 4 percent,
as noted.
Low Birth We i g h t a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 103
Preventing Low Birth Weight
Recognizing the close links between low birth
weight and socioeconomic status, policymakers
have emphasized a strategy of expanding prenatal
care eligibility and services for poor pregnant
women. The expansion of Medicaid eligibility
and outreach to pregnant women in the
late 1980s and early 1990s increased access to
prenatal care, improved services, and helped
more women begin care earlier in their pregnancies.
97 Rates of both early and adequate use
of prenatal care increased substantially between
1981 and 1998 for both blacks and
whites, and, except for some groups of young
mothers, racial disparities in the use of prenatal
care decreased.98 In 2000, 85.0 percent of
white mothers and 74.3 percent of black mothers
who gave birth in the United States began
prenatal care in the first trimester of pregnancy;
3.3 percent of white mothers and 6.7
percent of black mothers had late or no prenatal
care.99 Nevertheless, the U.S. rate of low
birth weight, even for singletons, has not declined?
perhaps owing in part to the declining
rate of fetal mortality?and remains higher
than that of most other developed countries.
It is difficult to ascertain the effectiveness of
prenatal care in reducing low birth weight.
Randomized controlled trials?the gold standard
in such research?are rarely feasible because
of ethical concerns about depriving
women of care. In a rare randomized trial,
Lorraine Klerman and colleagues compared
augmented and standard prenatal care provided
to Medicaid-eligible African American
women. The augmented care improved
women?s satisfaction with care and knowledge
about risk conditions but did not reduce
the rate of low birth weight.100
Studies other than randomized controlled trials
face several methodological challenges,
including selection bias. With favorable selection,
women with the best expected outcomes
are the most likely to seek prenatal
care and to do so early, so the estimated effect
of care could be overstated. With adverse
selection, women with the worst expected
outcomes are most likely to seek care
and to do so early, so the estimated effect of
care could be understated.
Research on the effects of expanded Medicaid
eligibility and services on birth weight has
produced mixed findings. Collectively, studies
indicate only modest positive effects,
stronger among blacks than whites.101 One
reason for the inconsistent findings may be
that prenatal care varies widely?in the services
and interventions offered, in the settings
in which it is provided, and in quality. Moreover,
interventions targeted at low-income
families often lose clients by attrition, and
programs are not always implemented as intended.
Two recent studies have found that
legislated changes in providers?one through
hospital desegregation in Mississippi in the
Civil Rights era and another, more recently,
through changes in Medicaid hospital payments
in California?reduced rates of low
birth weight among African American
children.102
Nancy E. Reichman
104 THE FUTURE OF CHILDREN
Nevertheless, the U.S. rate
of low birth weight, even for
singletons, has not declined?
perhaps owing in part to the
declining rate of fetal
mortality?and remains
higher than that of most
other developed countries.
Unquestionably, prenatal medical care can
benefit certain mothers and their babies
enormously. All women, pregnant or not,
should get preventive and regular medical
care. But standard prenatal care cannot be
expected to improve aggregate birth outcomes
because most treatable medical conditions
during pregnancy affect only a small
proportion of women.103 A recent comprehensive
review found no evidence that prenatal
educational or psychosocial services,
home visiting programs, or any medical interventions,
even those to prevent infections,
prevented either preterm birth or fetal
growth retardation.104 Researchers have recently
found that progesterone supplementation
reduces preterm birth among women
who have had a previous preterm birth, but
studies of its effectiveness and safety are still
ongoing.105 One promising way to reduce aggregate
rates of low birth weight is to reduce
smoking.106 Another is through better nutrition.
Three recent studies found that participation
in the Supplemental Nutrition Program
for Women, Infants, and Children
(WIC) raised birth weight.107
The point is not that prenatal care programs
have no positive effects. Rather, variations in
content, implementation, or compliance
make it difficult to pinpoint their effects.
They may improve maternal health by connecting
mothers to the health care system.
They may reduce fetal death. Those that include
family planning and other psychosocial
services that could affect future fertility and
prenatal behaviors could, in turn, improve
maternal or infant health and increase the
use of pediatric care. At the minimum,
women of childbearing age should receive
standard medical care beginning well before
pregnancy, as well as smoking cessation and
nutritional services as needed. But prenatal
care?even enhanced care?will not automatically
offset a lifetime of maternal health
disadvantages.
Improving Cognitive Outcomes
Associated with Low Birth Weight
Practitioners have established many early intervention
programs to enhance the cognitive
development of low birth weight infants and
to improve their school readiness. Many programs
pertaining to low birth weight and
school readiness have been designed as randomized
clinical trials, making them relatively
straightforward to evaluate.
A broad review of such interventions found
modest success overall, with the most effective
programs involving parents as well as
children.108 One such ?two-generation? intervention,
the Infant Health and Development
Program (IHDP), targeted low birth weight
premature infants at eight sites. In the treatment
group, 377 children received two years
of high-quality center-based care at ages two
and three. Family support, including home
visits and parent group meetings, was also
provided. The 608 children in the control
group received none of these services. Both
groups received the same medical care.
Many researchers have examined the readiness-
related effects, both cognitive and behavioral,
of the IHDP. Jeanne Brooks-Gunn
and her coauthors found that the mean IQ of
the intervention group at age three was 93.6,
while that of the control group was 84.2; and
that heavier low birth weight infants benefited
more than lighter infants (those weighing
less than 2,000 grams).109 For both black
and white subsamples, children whose mothers
had a high school education or less gained
more from the intervention than those whose
mothers had attended college, with the latter
showing no significant enhancement in IQ
scores at age three.110 Several studies found
Low Birth We i g h t a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 105
that the intervention improved cognitive
scores at ages twenty-four months and thirtysix
months, and one found lower (more favorable)
behavior problem scores at twenty-four
and thirty-six months.111 Children who had
large gains on IQ score, cognitive skills,
school achievement, and behavior at age
three, however, generally did not sustain the
gains at age eight, although the heavier low
birth weight intervention group still outscored
the control groups on measures of cognition
and school achievement.112 And another
study found that children at age eight
who had attended the program for at least 400
days scored 7 to 10 points higher on IQ tests
than those in the control group. Again, effects
were greater for the heavier low birth weight
infants (about 14 points) than for the lighter
low birth weight infants (about 8 points).113
Combining home visits with hospital-based
intervention also appears to be effective in
enhancing the cognitive function of low birth
weight children. In a randomized controlled
trial of an intervention in Vermont that provided
four home visits and seven hospital sessions,
the experimental low birth weight
group scored higher on several standardized
tests at age seven than did a control group
that received no treatment; differences in
outcomes first became statistically significant
at age three.114 The experimental group also
scored as high as a normal birth weight comparison
group. A recent review of interventions
targeting socially deprived families concluded
that home visits accompanied by early
stimulation in the neonatal unit, as well as by
preschool placement, appeared to improve
the cognitive development of low birth
weight and premature children.115
In sum, early intervention can improve the
cognitive and behavioral development of low
birth weight children. Two-generation programs,
which serve both mothers and children,
and those that combine home visits
with either center-based day care or hospitalbased
therapy appear particularly effective,
with more pronounced gains for heavier low
birth weight children.
Implications and
Recommendations
The message of this article is mixed and cautious.
Although racial disparities in low birth
weight are large and persistent, they explain,
at most, 3 to 4 percent of the racial gap in IQ
scores. Resolving the problem of low birth
weight will thus close only a small portion of
the racial gap in school readiness. The adverse
cognitive outcomes associated with low
birth weight are being addressed successfully
by several types of early intervention programs,
but their benefits are greatest for
heavier children, whereas it is the lightest
children who are at the greatest risk.
Overall, there is both good and bad news
about low birth weight. The encouraging
news is that over the past two decades
n Infant mortality rates among both blacks
and whites have declined.
n Birth weight?specific survival rates of both
black and white infants have increased
dramatically.
n Birth weight?specific survival rates show
no racial disparities. Black and white infants
appear to benefit equally in terms of
survival, at least in the aggregate, from
neonatal care technology.
n Thanks to public health campaigns and the
Medicaid expansions of the 1980s and
1990s, levels of prenatal care use are high
among both blacks and whites.
Nancy E. Reichman
106 THE FUTURE OF CHILDREN
The discouraging news is that
n Black babies continue to be twice as likely
as white babies to die before their first
birthday. Despite declining infant mortality
rates among both blacks and whites,
the infant mortality rate among blacks in
2000 was still higher than that among
whites twenty years earlier. Although the
absolute racial gap in infant mortality has
narrowed somewhat, the proportional gap
has increased by half.
n Rates of low birth weight in the United
States have not declined over the past
twenty years?overall, for blacks or for
whites. This apparently bad news may be
due, at least in part, to declining rates of
fetal death. However, the aggregate rate of
low birth weight in the United States
exceeds that of most other developed
countries.
n Black babies continue to be twice as likely
as white babies to be low birth weight.
Racial disparities are most pronounced at
the lowest birth weight ranges?those associated
with the poorest child health and
developmental outcomes.
I offer several recommendations for improving
maternal and child health generally and
for combating low birth weight as a way to reduce
racial disparities in school readiness.
First, policymakers and practitioners must
focus on maternal health risks well before
conception. It is extremely difficult, if not impossible,
to counteract a lifetime of disadvantage
during the gestational period. The emphasis
must be on women?s health rather than
on prenatal care. Many analysts have made
this same point, but its importance cannot be
overemphasized. Second, researchers must
pay more attention to maternal and paternal
environmental exposures and to the biological
role of fathers, more broadly, in infant
health and child health and development.
Third, reducing rates of low birth weight
would improve cognitive and behavioral outcomes
among the entire population of
school-aged children. At the same time, it
would narrow racial gaps in school readiness,
particularly if it were part of a multipronged,
integrative approach focusing on the many
inputs to school readiness reviewed in this
volume.
Although the 3 to 4 percent potential contribution
of low birth weight to the racial gap in
IQ scores may not seem large, eliminating
one source of the disparity is a step in the
right direction. Moreover, beyond the question
of school achievement, low birth weight
is a problem that must be addressed to meet
the national goals of increasing quality and
years of healthy life and eliminating racial,
ethnic, and socioeconomic disparities in
health.116
Early intervention can and has improved cognitive
and behavioral outcomes among low
birth weight children. Ideally, such interventions
should be available to all children who
could benefit. That said, they appear to be of
greater benefit to heavier low birth weight
children than to lighter ones. Because black
infants are more likely than white infants to
fall into the lowest weight ranges, preventing
low birth weight?when we learn how to do
so?is likely to be more effective than remedial
intervention at narrowing racial gaps in
school readiness.
Low Birth We i g h t a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 107
Endnotes
1. See the special issue of The Future of Children (vol. 5, no. 1, Spring 1995) devoted to low birth weight.
2. UNICEF (United Nations Children?s Fund), The State of the World?s Children (Oxford University Press,
2003).
3. T. J. Matthews, Fay Menacker, and Marian F. MacDorman, ?Infant Mortality Statistics from the 2001 Period:
Linked Birth/Infant Death Data Set,? National Vital Statistics Reports 52, no. 2. (Hyattsville, Md.: National
Center for Health Statistics, September 15, 2003); Maureen Hack, Nancy K. Klein, and H. Gerry
Taylor, ?Long-Term Developmental Outcomes of Low Birth Weight Infants,? The Future of Children 5, no.
1 (1995), 176?96.
4. Jeffrey D. Horbar and Jerold F. Lucey, ?Evaluation of Neonatal Intensive Care Technologies,? The Future
of Children 5, no. 1 (1995): 139?16; Virginia M. Fried and others, Chartbook on Trends in the Health of
Americans (Hyattsville, Md.: National Center for Health Statistics, 2003).
5. Joyce A. Martin and others, ?Births: Final Data for 2000,? National Vital Statistics Reports 52, no. 10 (Hyattsville,
Md.: National Center for Health Statistics, February 12, 2002).
6. Data on singleton births allow for comparisons over time that are not confounded by changes in the prevalence
of multiple births.
7. The survival rate for normal birth weight singleton infants was 99 percent in 1960, 1980, and 2000.
8. For an extensive review of child medical and developmental conditions associated with low birth weight,
see Hack, Klein, and Taylor, ?Long-Term Developmental Outcomes of Low Birth Weight Infants? (see note
3); National Institute of Neurological Disorders and Stroke, ?NINDS Cerebral Palsy Information Page?
(www.ninds.nih.gov/health_and_medical/disorders/cerebral_palsy.htm [July 14, 2004]).
9. March of Dimes Birth Defects Foundation, ?Cerebral Palsy? (www.marchofdimes.com/professionals/
681_1208.asp [April 5, 2004]).
10. University of California, San Francisco (UCSF) Children?s Hospital, ?Bronchopulmonary Dysplasia? (www.
ucsfhealth.org/childrens/health_professionals/manuals/31_ChronicLungDis.pdf [April 5, 2004]); UCSF
Children?s Hospital, ?Respiratory Distress Syndrome? (www.ucsfhealth.org/childrens/health_professionals/
manuals/25_RDS.pdf [April 5, 2004]).
11. National Institutes of Health, National Eye Institute, ?Retinopathy of Prematurity? (www.nei.nih.gov/
health/ rop/#2 [April 6, 2004]).
12. John M. Lorenz and others, ?A Quantitative Review of Mortality and Developmental Disability in Extremely
Premature Newborns,? Archives of Pediatric and Adolescent Medicine 152, no. 5 (1998): 425?35.
13. Martin and others, ?Births? (see note 5).
14. Ibid.
15. In 2000 the 2:1 ratio was present in the following maternal age groups: twenty to twenty-four, twenty-five to
twenty-nine, thirty to thirty-four, thirty-five to thirty-nine, and forty to forty-four. The disparities are narrower
for very young and very old mothers. For mothers younger than fifteen, the black-white low birth
weight ratio is 1.4, for mothers aged fifteen to nineteen it is 1.7, and for mothers forty-five to fifty-four it is
Nancy E. Reichman
108 THE FUTURE OF CHILDREN
1.3. Among singleton births, the rate of low birth weight in 2000 was 4.99 percent among white mothers
and 11.15 percent among black mothers; the corresponding rates in 1980 were 4.90 and 11.46 percent, respectively.
Martin and others, ?Births? (see note 5).
16. Kenneth C. Schoendorf and others, ?Mortality among Infants of Black as Compared with White College
Educated Parents,? New England Journal of Medicine 326, no. 23 (1992): 1522?26.
17. Martin and others, ?Births? (see note 5).
18. Cande V. Ananth and others, ?Small-For-Gestational-Age Births among Black and White Women: Temporal
Trends in the United States,? American Journal of Public Health 93, no. 4 (2003): 577-79.
19. Martin and others, ?Births? (see note 5).
20. W. Parker Frisbie and Seung-eun Song, ?Hispanic Pregnancy Outcomes: Differentials over Time and Current
Risk Factor Effects,? Policy Studies Journal 31, no. 2 (2003), 237.
21. Jose E. Becerra and others, ?Infant Mortality among Hispanics: A Portrait of Heterogeneity,? Journal of the
American Medical Association 265, no. 2 (1991): 217?21; Fernando S. Mendoza and others, ?Selected Measures
of Health Status for Mexican-American, Mainland Puerto Rican, and Cuban-American Children,?
Journal of the American Medical Association 265, no. 2 (1991): 227?32.
22. Kyriakos S. Markides and Jeannine Coreil, ?The Health of Hispanics in the Southwestern United States: An
Epidemiological Paradox,? Public Health Reports 101, no. 3 (1986): 253?65.
23. Nancy E. Reichman and Genevieve M. Kenney, ?The Effects of Parents? Place of Birth and Ethnicity on
Birth Outcomes in New Jersey,? in Keys to Successful Immigration: Implications of the New Jersey Experience,
edited by Thomas J. Espenshade (Washington: Urban Institute Press, 1997): 199?230; Nancy E.
Reichman and Genevieve M. Kenney, ?Prenatal Care, Birth Outcomes and Newborn Hospitalization Costs:
Patterns among Hispanics in New Jersey,? Family Planning Perspectives 30, no. 4 (1998): 182?87; Martin
and others, ?Births? (see note 5).
24. Gosta Rooth, ?Low Birthweight Revised,? Lancet 1, no. 8169 (1980): 639?41.
25. Nigel Paneth, ?The Problem of Low Birth Weight,? The Future of Children 5, no. 1 (1995): 19?34.
26. Martin and others, ?Births? (see note 5).
27. Stephen A. Norton, Genevieve M. Kenney, and Marilyn Rymer Ellwood, ?Medicaid Coverage of Maternity
Care for Aliens in California,? Family Planning Perspectives 28, no. 3 (1996): 108?12; Julia M. Solis and
others, ?Acculturation, Access to Care, and Use of Preventive Services by Hispanics: Findings from
HHANES, 1982?84,? American Journal of Public Health 80, suppl. (1990): 11?19; Fernando M. Trevino
and others, ?Health Insurance Coverage and Utilization of Health Services by Mexican Americans, Mainland
Puerto Ricans, and Cuban Americans,? Journal of the American Medical Association 265, no. 2 (1991):
233?37; Patricia Moore and Joseph T. Hepworth, ?Use of Perinatal and Infant Health Services by Mexican-
American Medicaid Enrollees,? Journal of the American Medical Association 272, no. 14 (1994): 1111?15;
Sylvia Guendelman, ?Mexican Women in the United States,? Lancet 344, no. 8919 (1994): 352.
28. Nancy S. Landale, R. S. Oropesa, and Bridget K. Gorman, ?Immigration and Infant Health: Birth Outcomes
of Immigrant and Native-Born Women,? in Children of Immigrants: Health, Adjustment and Public
Assistance, edited by D. J. Hernandez (Washington: National Academy Press, 1999).
Low Birth We i g h t a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 109
29. Howard J. Cabral and others, ?Foreign-Born and U.S.-Born Black Women: Differences in Health Behaviors
and Birth Outcomes,? American Journal of Public Health 80, no. 1 (1990): 70?71; Richard J. David and
James W. Collins, ?Differing Birth Weight among Infants of U.S.-Born Blacks, African-Born Blacks and
U.S.-Born Whites,? New England Journal of Medicine 337, no. 17 (1997): 1209?14; Jing Fang, Shantha
Madhavan, and Michael H. Alderman, ?Low Birth Weight: Race and Maternal Nativity?Impact of Community
Income,? Pediatrics 103, no. 1 (1999): e5: 1?6; Gopal K. Singh and Stella M. Yu, ?Adverse Pregnancy
Outcomes: Differences between U.S. and Foreign-Born Women in Major U.S. Racial and Ethnic
Groups,? American Journal of Public Health 86, no. 6 (1996): 837?43.
30. Unless indicated otherwise, the analyses of trends in birth weight and survival are restricted to single live
births, due to incomplete data for multiple births in 1980 and to allow for comparisons over time that are
not confounded by changes in the prevalence of multiple births.
31. Wanda Barfield, ?Racial/Ethnic Trends in Fetal Mortality?United States, 1990?2000,? Morbidity and Mortality
Weekly Report 53, no. 24 (2004) (www.cdc.gov/mmwr/preview/mmwrhtml/mm5324a4.htm [June 27,
2004]). These figures are not limited to singletons. The corresponding decreases for American Indians/Alaska
Natives, Asians/Pacific Islanders, and Hispanics were 27 percent, 8 percent, and 16 percent, respectively.
32. Figures for singletons are computed from the same data sources as in figure 2. The corresponding infant
mortality figures, including multiple births, for 1980 are 22 out of 1,000 and 11 out of 1,000 for blacks and
whites, respectively; and for 2000, 14 out of 1,000 and 6 out of 1,000, respectively. Centers for Disease Control
and Prevention, Morbidity and Mortality Weekly Report 51, no. 27 (2002): 589?92 (www.cdc.gov/
mmwr/preview/mmwrhtml/mm5127a1.htm [April 1, 2004]).
33. Data sources on survival rates are the same as in figure 2. Race-specific survival rates were also virtually
identical for moderately low birth weight and normal birth weight in 2000. T. J. Matthews, Fay Menacker,
and Marian F. MacDorman, ?Infant Mortality Statistics from the 2000 Period: Linked Birth/Infant Death
Data Set,? National Vital Statistics Reports 50, no. 12 (Hyattsville, Md.: National Center for Health Statistics,
August 28, 2003). In 1999, 3.1 percent of births to white mothers and 3.3 percent of births to black
mothers were multiples. Black mothers were slightly more likely to have twins, and white mothers were
slightly more likely to have higher-order multiple births. Rebecca B. Russell and others, ?The Changing Epidemiology
of Multiple Births in the United States,? Obstetrics and Gynecology 101, no. 1 (2003): 129?35.
34. See sources for figures 2 and 3.
35. In 2000, 85 percent of deaths up to age five were of children under one year of age. Centers for Disease
Control, ?Deaths by Single Years of Age, Race, and Sex: United States, 2000,? Year 2000 Mortality Statistics,
table 310 (www.cdc.gov/nchs/data/statab/wktbl310.pdf [April 1, 2004]).
36. Marie C. McCormick and others, ?The Health and Development Status of Very Low-Birth-Weight Children
at School Age,? Journal of the American Medical Association 267, no. 16 (1992): 2204?08; Maureen
Hack and others, ?School-Age Outcomes in Children with Birth Weights under 750g,? New England Journal
of Medicine 33, no. 12 (1994): 753?59.
37. Adnan T. Bhutta and others, ?Cognitive and Behavioral Outcomes of School-Aged Children Who Were
Born Preterm: A Meta-Analysis,? Journal of the American Medical Association 288, no. 6 (2002): 728?37.
38. Rachel Nonkin Avchen, Keith G. Scott, and Craig A. Mason, ?Birth Weight and School-Age Disabilities: A
Population-Based Study,? American Journal of Epidemiology 154, no. 10 (2001): 895?901.
Nancy E. Reichman
110 THE FUTURE OF CHILDREN
39. Jennifer Pinto-Martin and others, ?Special Education Services and School Performance in a Regional Cohort
of Low-Birthweight Infants at Age Nine,? Paediatric and Perinatal Epidemiology 18, no. 2
(2004): 120?29.
40. Betty R. Vohr and others, ?Neurodevelopmental and Functional Outcomes of Extremely Low Birth Weight
Infants in the National Institute of Child Health and Human Development Neonatal Research Network,
1993?1994,? Pediatrics 105, no. 6 (2000): 1216?26; Peter Anderson, Lex W. Doyle, and the Victorian Infant
Collaborative Study Group, ?Neurobehavioral Outcomes of School-Age Children Born Extremely Low
Birth Weight or Very Preterm in the 1990s,? Journal of the American Medical Association 289, no. 24
(2003): 3264?72.
41. Anderson and others, ?Neurobehavioral Outcomes of School-Age Children Born Extremely Low Birth
Weight or Very Preterm in the 1990s? (see note 40); Bhutta and others, ?Cognitive and Behavioral Outcomes
of School-Aged Children Who Were Born Preterm? (see note 37); Jennifer R. Bowen, Frances L.
Gibson, and Peter J. Hand, ?Educational Outcome at 8 Years of Children Who Were Born Prematurely: A
Controlled Study,? Journal of Paediatrics and Child Health 38, no. 5 (2002): 438?44; Hack and others,
?School-Age Outcomes in Children with Birth Weights under 750g? (see note 36).
42. Glen P. Aylward, ?Cognitive Function in Preterm Infants: No Simple Answers,? Journal of the American
Medical Association 289, no. 6 (2003): 752?53.
43. Anderson and others, ?Neurobehavioral Outcomes of School-Age Children Born Extremely Low Birth
Weight or Very Preterm in the 1990s? (see note 41); Bhutta and others, ?Cognitive and Behavioral Outcomes
of School-Aged Children Who Were Born Preterm? (see note 37).
44. Anderson and others, ?Neurobehavioral Outcomes of School-Age Children Born Extremely Low Birth
Weight or Very Preterm in the 1990s? (see note 40); Bhutta and others, ?Cognitive and Behavioral Outcomes
of School-Aged Children Who Were Born Preterm? (see note 37).
45. Anderson and others, ?Neurobehavioral Outcomes of School-Age Children Born Extremely Low Birth Weight
or Very Preterm in the 1990s? (see note 40); Jane L. Hutton and others, ?Differential Effects of Preterm Birth
and Small Gestational Age on Cognitive and Motor Development,? Archives of Diseases in Children 76, no. 2
(1997): F75?F81; Susan A. Rose and Judith F. Feldman, ?Memory and Processing Speed in Preterm Children
at Eleven Years: A Comparison with Full-Terms,? Child Development 67, no. 5 (1996): 2005?21.
46. Anderson and others, ?Neurobehavioral Outcomes of School-Age Children Born Extremely Low Birth
Weight or Very Preterm in the 1990s? (see note 40); Bradley S. Peterson and others, ?Regional Brain Volume
Abnormalities and Long-Term Cognitive Outcome in Preterm Infants,? Journal of the American Medical
Association 284, no. 15 (2000): 1939?47.
47. Bhutta and others, ?Cognitive and Behavioral Outcomes of School-Aged Children Who Were Born
Preterm? (see note 37).
48. Lynn T. Singer and others, ?Maternal Psychological Distress and Parenting Stress after the Birth of a Very
Low-Birth-Weight Infant,? Journal of the American Medical Association 281, no. 9 (1999): 799?805.
49. Hack, Klein, and Taylor, ?Long-Term Developmental Outcomes of Low Birth Weight Infants? (see note 3).
A review of the more recent literature indicates that these rates have remained essentially unchanged.
50. There are as yet too few studies to derive reliable estimates of the size of these effects. See, for example,
Thomas D. Matte and others, ?Influence of Variation in Birth Weight within Normal Range and within Sib-
Low Birth We i g h t a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 111
ships on IQ at Age 7 Years: Cohort Study,? British Medical Journal 323, no. 7308 (2001): 310?14; Robert
Kaestner and Hope Corman, ?The Impact of Child Health and Family Inputs on Child Cognitive Development,?
Working Paper 5257 (Cambridge, Mass.: National Bureau of Economic Research, 1995); Jason D.
Boardman and others, ?Low Birth Weight, Social Factors, and Developmental Outcomes among Children
in the United States,? Demography 39, no. 6 (2002): 353?68.
51. Anne Case, Angela Fertig, and Christina Paxson, ?The Lasting Impact of Child Health and Circumstance,?
Journal of Health Economics (forthcoming).
52. Yolanda C. Padilla and others, ?Is the Mexican American ?Epidemiologic Paradox? Advantage of Birth
Maintained through Early Childhood?? Social Forces 80, no. 3 (2002): 1101?23.
53. Jeanne Brooks-Gunn and others, ?The Black-White Test Score Gap in Young Children: Contributions of
Test and Family Characteristics,? Applied Developmental Science 7, no. 4 (2003): 239?52.
54. See Brooks-Gunn and others, ?The Black-White Test Score Gap in Young Children? (note 53), for the magnitude
of the observed total gap (denominator).
55. Dana Hughes and Lisa Simpson, ?The Role of Social Change in Preventing Low Birth Weight,? The Future
of Children 5, no. 1 (1995): 87?102.
56. U.S. Department of Health and Human Services, Healthy People 2010, 2nd ed. (Government Printing Office,
2000).
57. Fried and others, Chartbook on Trends in the Health of Americans (see note 4).
58. Bernadette D. Proctor and Joseph Dalaker, ?Poverty in the United States: 2002,? Current Population Reports,
P60-222 (U.S. Census Bureau, 2003).
59. Stephanie J. Ventura and others, ?The Demography of Out-of-Wedlock Childbearing,? Report to Congress
on Out-of-Wedlock Childbearing (U.S. Department of Health and Human Services, 1995).
60. Martin and others, ?Births? (see note 5).
61. Proctor and Dalaker, ?Poverty in the United States? (see note 58). Poverty, in this case, identifies families
with a female householder and no husband present.
62. Martin and others, ?Births? (see note 5).
63. Allison M. Fraser, John E. Brockert, and Ryk H. Ward, ?Association of Young Maternal Age with Adverse
Reproductive Outcomes,? New England Journal of Medicine 332, no. 17 (1995): 1113?17.
64. Alan Guttmacher Institute, Sex and America?s Teenagers (New York and Washington, 1994); Cheryl D.
Hayes, ed., Risking the Future: Adolescent Sexuality, Pregnancy, and Childbearing, vol. 1 (Washington: National
Academy Press, 1987).
65. Arline T. Geronimus, ?The Weathering Hypothesis and the Health of African-American Women and Infants:
Evidence and Speculations,? Ethnicity and Disease 2, no. 3 (1992): 207-21; Arline T. Geronimus,
?Black/White Differences in the Relationship of Maternal Age to Birthweight: A Population-Based Test of
the Weathering Hypothesis,? Social Science and Medicine 42, no. 4 (1996): 589-97.
66. Martin and others, ?Births? (see note 5).
67. Ibid.
Nancy E. Reichman
112 THE FUTURE OF CHILDREN
68. Robert L. Goldenberg and Lorraine V. Klerman, ?Adolescent Pregnancy?Another Look,? New England
Journal of Medicine 332, no. 17 (1995): 1161?62.
69. Sarah S. Brown and Leon Eisenberg, eds., The Best Intentions: Unintended Pregnancy and the Well-Being
of Children and Families (Washington: National Academy Press, 1995).
70. Nancy E. Reichman and Deanna L. Pagnini, ?Maternal Age and Birth Outcomes: Data from New Jersey,?
Family Planning Perspectives 29, no. 6 (1997): 268?72, 295.
71. Institute of Medicine, Committee to Study the Prevention of Low Birth Weight, Preventing Low Birth
Weight (Washington: National Academy Press, 1985); Charles J. Lockwood, ?Predicting Premature Delivery?
No Easy Task,? New England Journal of Medicine 346, no. 4 (2002): 282?84.
72. Peter W. Nathanielsz, ?The Role of Basic Science in Preventing Low Birth Weight,? The Future of Children
5, no. 1 (1995): 57?70; Greg R. Alexander and Carol C. Korenbrot, ?The Role of Prenatal Care in Preventing
Low Birth Weight,? American Journal of Public Health 5, no. 1 (1995): 103?20. For an up-to-date discussion
of what is known about the process of preterm birth, see Donald R. Mattison and others, eds., ?The
Role of Environmental Hazards in Premature Birth,? Workshop Summary, Roundtable on Environmental
Health Sciences, Research, and Medicine, Board of Health Sciences Policy (Washington: Institute of Medicine
of the National Academy, 2003).
73. Nathanielsz, ?The Role of Basic Science in Preventing Low Birth Weight?; Alexander and Korenbrot, ?The
Role of Prenatal Care in Preventing Low Birth Weight? (for both, see note 72).
74. Virginia Rall Chomitz, Lilian W. Y. Cheung, and Ellice Lieberman, ?The Role of Lifestyle in Preventing
Low Birth Weight,? The Future of Children 5, no. 1 (1995): 121?38.
75. Martin and others, ?Births? (see note 5).
76. Sharon Hillier and others, ?Association between Bacterial Vaginosis and Preterm Delivery of a Low-Birth-
Weight Infant,? New England Journal of Medicine 33, no. 26 (1995): 1737?42. Many studies have replicated
this finding.
77. Chomitz, Cheung, and Lieberman, ?The Role of Lifestyle in Preventing Low Birth Weight? (see note 74).
78. Martin and others, ?Births? (see note 5).
79. J. Durham and others, ?Alcohol Consumption among Pregnant and Childbearing-Aged Women: United
States, 1991 and 1995,? Morbidity and Mortality Weekly Report 46, no. 16 (April 25, 1997) (www.health.
org/ nongovpubs/mmwr [April 1, 2004]).
80. Substance Abuse and Mental Health Services Administration (SAMHSA), ?Drug Abuse in America: 2001,?
excerpted from the 2001 National Household Survey on Drug Abuse, September 5, 2002 (www.
policyalmanac.org/crime/archive/drug_abuse.shtml [January 21, 2004]).
81. Irvin Emanuel and others, ?The Washington State Intergenerational Study of Birth Outcomes: Methodology
and Some Comparisons of Maternal Birthweight and Infant Birthweight and Gestation in Four Ethnic
Groups,? Paediatric and Perinatal Epidemiology 13, no. 3 (1999): 352?69; James W. Collins Jr. and others,
?Low Birth Weight across Generations,? Maternal and Child Health Journal 7, no. 4 (2003): 229?37; Dalton
Conley, Kate W. Strully, and Neil G. Bennett, The Starting Gate: Birth Weight and Life Chances (University
of California Press, 2003); Dalton Conley and Neil G. Bennett, ?Is Biology Destiny? Birth Weight
and Life Chances,? American Sociological Review 65, no. 3 (2000): 458?67.
Low Birth We i g h t a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 113
82. Conley, Strully, and Bennett, The Starting Gate (see note 81).
83. Ellen L. Mozurkewich and others, ?Working Conditions and Adverse Pregnancy Outcome: A Metaanalysis,?
Obstetrics and Gynecology 95, no. 4 (2000): 623?35.
84. Boguslaw Baranski, ?Effects of the Workplace on Fertility and Related Reproductive Outcomes,? Environmental
Health Perspectives Supplements 101, suppl. 2 (1993): 81?90; Sohail Khattak and others, ?Pregnancy
Outcome Following Gestational Exposure to Organic Solvents: A Prospective Controlled Study,?
Journal of the American Medical Association 281, no. 12 (1999): 1106?09; Eunhee Ha and others, ?Parental
Exposure to Organic Solvents and Reduced Birth Weight,? Archives of Environmental Health 57, no. 3
(2002): 204?14.
85. Jesse McKinnon, ?The Black Population in the United States: March 2002,? Current Population Reports,
P20-541 (U.S. Census Bureau, 2003). Corresponding figures for pregnant women alone are not available.
86. Council of Economic Advisers, Changing America: Indicators of Social and Economic Well-Being by Race
and Hispanic Origin (September 1998) (www.access.gpo.gov/eop/ca/pdfs/ca.pdf).
87. Narayan Sastry and Jon M. Hussey, ?An Investigation of Racial and Ethnic Disparities in Birth Weight in
Chicago Neighborhoods,? Demography 40, no. 4 (2003): 701?25.
88. See, for example, Jeffrey D. Morenoff, ?Neighborhood Mechanisms and the Spatial Dynamics of Birth
Weight,? American Journal of Sociology 108, no. 5 (2003): 976-1017.
89. Ibid.
90. J. Felix Rogers and others, ?Association of Very Low Birth Weight with Exposures to Environmental Sulfur
Dioxide and Total Suspended Particulates,? American Journal of Epidemiology 151, no. 6 (2000): 602?13;
Sven E. Rodenbeck, Lee M. Sanderson, and Antonio Rene, ?Maternal Exposure to Trichloroethylene in
Drinking Water and Birth Weight Outcomes,? Archives of Environmental Health 55, no. 3 (2000): 188?94;
Akerke Baibergenova and others, ?Low Birth Weight and Residential Proximity to PCB-Contaminated
Waste Sites,? Environmental Health Perspectives 111, no. 10 (2003): 1352?57.
91. Paul A. Jargowsky, Poverty and Place: Ghettos, Barrios, and the American City (New York: Russell Sage
Foundation, 1997).
92. Jacquetta M. Trasler and Tonia Doerksen, ?Teratogen Update: Paternal Exposures?Reproductive Risks,?
Teratology 60, no. 3 (1999): 161?72.
93. Ruth E. Little and Charles F. Sing, ?Association of Father?s Drinking and Infant?s Birth Weight,? New England
Journal of Medicine 314, no. 25 (1986): 1644?45.
94. Jun Zhang and Jennifer M. Ratcliffe, ?Paternal Smoking and Birthweight in Shanghai,? American Journal of
Public Health 83, no. 2 (1993): 207?10; Fernando D. Martinez, Anne L. Wright, and Lynn M. Taussig, ?The
Effect of Paternal Smoking on the Birthweight of Newborns Whose Mothers Did Not Smoke,? American
Journal of Public Health 84, no. 9 (1994): 1489?91.
95. David A. Savitz and others, ?Male Pesticide Exposure and Pregnancy Outcome,? American Journal of Epidemiology
146, no. 12 (1997): 1025?37; Petter Kristensen and others, ?Perinatal Outcome among Children
of Men Exposed to Lead and Organic Solvents in the Printing Industry,? American Journal of Epidemiology
137, no. 2 (1993): 134?44; Maarja-Liisa Lindbohm, ?Effects of Parental Exposure to Solvents on Pregnancy
Nancy E. Reichman
114 THE FUTURE OF CHILDREN
Outcome,? Journal of Occupational and Environmental Medicine 37, no. 8 (1995): 908?14; Shao Lin and
others, ?Does Paternal Occupational Lead Exposure Increase the Risks of Low Birth Weight or Prematurity??
American Journal of Epidemiology 148, no. 2 (1998): 173?81; Yi Min, Adolfo Correa-Villasenor, and
Patricia A. Stewart, ?Parental Occupational Lead Exposure and Low Birth Weight,? American Journal of
Industrial Medicine 30, no. 5 (1996): 569?78.
96. McKinnon, ?The Black Population in the United States? (see note 85).
97. Janet Currie and Jonathan Gruber, ?Saving Babies: The Efficacy and Cost of Recent Changes in the Medicaid
Eligibility of Pregnant Women,? Journal of Political Economy 104, no. 6 (1996): 1263?96; Lisa Dubay
and others, ?Changes in Prenatal Care Timing and Low Birth Weight by Race and Socioeconomic Status:
Implications for the Medicaid Expansions for Pregnant Women,? Health Services Research 36, no. 2 (2001):
373?98.
98. Greg R. Alexander, Michael D. Kogan, and Sara Nabukera, ?Racial Differences in Prenatal Care Use in the
United States: Are Disparities Decreasing?? American Journal of Public Health 92, no. 12 (2002): 1970?75.
Although timing of care is widely used as a measure of prenatal care use, it captures neither the intensity
nor the quality of care received. The Adequacy of Prenatal Care Utilization (APNCU) Index, which is based
on both the timing of prenatal care and the number of visits, does a better job of capturing the intensity of
care; for a description and comparison with earlier indexes of the adequacy of prenatal care, see Milton
Kotelchuck, ?An Evaluation of the Kessner Adequacy of Prenatal Care Index and a Proposed Adequacy of
Prenatal Care Utilization Index,? American Journal of Public Health 84. no. 9 (1994): 1414?20. Even these
indexes do not capture the quality of care received, however.
99. Martin and others, ?Births? (see note 5).
100. Lorraine V. Klerman and others, ?A Randomized Trial of Augmented Prenatal Care for Multiple-Risk,
Medicaid-Eligible African American Women,? American Journal of Public Health 91, no. 1 (2001): 105?11.
101. Currie and Gruber, ?Saving Babies? (see note 97); Dubay and others, ?Changes in Prenatal Care Timing
and Low Birth Weight by Race and Socioeconomic Status? (see note 97); Arnold M. Epstein and Joseph P.
Newhouse, ?Impact of Medicaid Expansion on Early Prenatal Care and Health Outcomes,? Health Care
Financing Review 19, no. 4 (1998): 85?99; Nancy E. Reichman and Maryanne J. Florio, ?The Effects of Enriched
Prenatal Care Services on Medicaid Birth Outcomes in New Jersey,? Journal of Health Economics
15, no. 4 (1996): 455?76; Theodore Joyce, ?Impact of Augmented Prenatal Care on Birth Outcomes of
Medicaid Recipients in New York City,? Journal of Health Economics 18, no. 1 (1999): 31?67.
102. Douglas Almond and Kenneth Y. Chay, ?The Long-Run and Intergenerational Impact of Poor Infant
Health: Evidence from Cohorts Born during the Civil Rights Era? (National Bureau of Economic Research,
2003) (www.nber.org/~almond/cohorts.pdf [July 25, 2004]); Anna Aizer, Adriana Lleras-Muney, and
Mark Stabile, ?Access to Care, Provider Choice, and Racial Disparities in Infant Mortality,? Working Paper
10445 (Cambridge, Mass.: National Bureau of Economic Research, 2004).
103. Alexander and Korenbrot, ?The Role of Prenatal Care in Preventing Low Birth Weight? (see note 72).
104. Michael C. Lu and others, ?Preventing Low Birth Weight: Is Prenatal Care the Answer?? Journal of Maternal-
Fetal and Neonatal Medicine 13, no. 6 (2003): 362?80.
105. Paul J. Meis and others, ?Prevention of Recurrent Preterm Delivery by 17 Alpha-Hydroxyprogesterone
Caproate,? New England Journal of Medicine 348, no. 24 (2003): 2379?85; American College of Obstetri-
Low Birth We i g h t a n d S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 115
cians and Gynecologists, Committee on Obstetric Practice, ?Use of Progesterone to Reduce Preterm
Birth,? ACOG committee opinion 291, Obstetrics and Gynecology 102, no. 5 (2003):1115?16.
106. Alexander and Korenbrot, ?The Role of Prenatal Care in Preventing Low Birth Weight? (see note 72); Lu
and others, ?Preventing Low Birth Weight? (see note 104).
107. Lori Kowaleski-Jones and Greg J. Duncan, ?Effects of Participation in the WIC Program on Birthweight:
Evidence from the National Longitudinal Survey of Youth,? American Journal of Public Health 92, no. 5
(2002): 799?803; Nancy E. Reichman and Julien O. Teitler, ?Effects of Psychosocial Risk Factors and Prenatal
Interventions on Birth Weight: Evidence from New Jersey?s HealthStart Program,? Perspectives on
Sexual and Reproductive Health 35, no. 3 (2003): 130?37; Marianne P. Bitler and Janet Currie, ?Does WIC
Work? The Effects of WIC on Pregnancy and Birth Outcomes,? Journal of Policy Analysis and Management
(forthcoming).
108. Cecilia M. McCarton, Ina F. Wallace, and Forrester C. Bennett, ?Early Intervention for Low Birth-Weight
Premature Infants: What Can We Achieve?? Annals of Medicine 28, no. 3 (1996): 221?25.
109. Jeanne Brooks-Gunn and others, ?Early Intervention in Low Birth-Weight Premature Infants: Results
through Age 5 Years from the Infant Health and Development Program,? Journal of the American Medical
Association 272, no. 16 (1994): 1257?62.
110. Jeanne Brooks-Gunn and others, ?Enhancing the Cognitive Outcomes of Low Birth Weight, Premature Infants:
For Whom Is the Intervention Most Effective?? Pediatrics 89, no. 6, part 2 (1992): 1209?15.
111. Jeanne Brooks-Gunn and others, ?Enhancing the Development of Low Birth-Weight, Premature Infants:
Changes in Cognition and Behavior over the First Three Years,? Child Development 64, no. 3 (1993): 736-
53. Clancy Blair, Craig T. Ramey, and Michael J. Hardin, ?Early Intervention for Low Birthweight, Premature
Infants: Participation and Intellectual Development,? American Journal of Mental Retardation 99, no.
5 (1995): 542-54.
112. Cecilia M. McCarton and others, ?Results at Age 8 Years of Early Intervention for Low-Birth-Weight Premature
Infants: The Infant Health and Development Program,? Journal of the American Medical Association
277, no. 2 (1997): 126?32.
113. Jennifer L. Hill, Jeanne Brooks-Gunn, and Jane Waldfogel, ?Sustained Effects of High Participation in an
Early Intervention for Low Birth-Weight Premature Infants,? Developmental Psychology 39, no. 4 (2003):
730?44.
114. Thomas M. Achenbach and others, ?Nine-Year Outcome of the Vermont Intervention Program for Low
Birth Weight Infants,? Pediatrics 91, no. 1 (1993): 45?55.
115. Dagmar Lagerberg, ?Secondary Prevention in Child Health: Effects of Psychological Intervention, Particularly
Home Visitation, on Children?s Development and Other Outcome Variables,? Acta Paediatrica 89,
suppl. (2000): 43?52. This review expanded on an earlier review of randomized controlled trials of home
visitation programs by including interventions that offered other forms of support than home visits and
studies published after 1992. See David L. Olds and Harriet Kitzman, ?Review of Research on Home Visiting
for Pregnant Women and Parents of Young Children,? The Future of Children 3, no. 3 (1993): 53?92.
116. U.S. Department of Health and Human Services, Healthy People, 2010 (see note 56).
Nancy E. Reichman
116 THE FUTURE OF CHILDREN
Health Disparities and Gaps
in School Readiness
Janet Currie
Summary
The author documents pervasive racial disparities in the health of American children and analyzes
how and how much those disparities contribute to racial gaps in school readiness. She explores
a broad sample of health problems common to U.S. children, such as attention deficit
hyperactivity disorder, asthma, and lead poisoning, as well as maternal health problems and
health-related behaviors that affect children?s behavioral and cognitive readiness for school.
If a health problem is to affect the readiness gap, it must affect many children, it must be linked
to academic performance or behavior problems, and it must show a racial disparity either in its
prevalence or in its effects. The author focuses not only on the black-white gap in health status
but also on the poor-nonpoor gap because black children tend to be poorer than white children.
The health conditions Currie considers seriously impair cognitive skills and behavior in individual
children. But most explain little of the overall racial gap in school readiness. Still, the cumulative
effect of health differentials summed over all conditions is significant. Currie?s rough calculation
is that racial differences in health conditions and in maternal health and behaviors
together may account for as much as a quarter of the racial gap in school readiness.
Currie scrutinizes several policy steps to lessen racial and socioeconomic disparities in children?s
health and to begin to close the readiness gap. Increasing poor children?s eligibility for
Medicaid and state child health insurance is unlikely to be effective because most poor children
are already eligible for public insurance. The problem is that many are not enrolled. Even increasing
enrollment may not work: socioeconomic disparities in health persist in Canada and
the United Kingdom despite universal public health insurance. The author finds more promise
in strengthening early childhood programs with a built-in health component, like Head Start;
family-based services and home visiting programs; and WIC, the federal nutrition program for
women, infants, and small children. In all three, trained staff can help parents get ongoing care
for their children.
VOL. 15 / NO. 1 / SPRING 2005 117
www.future of children.org
Janet Currie is professor of economics at the University of California at Los Angeles. She is affiliated with the National Bureau of Economic
Research and is a research fellow at the Institute for the Study of Labor (IZA). She would like to thank Christina Paxson and Jack Shonkoff
for their comments, as well as Michelle Hemmat for helpful discussions over the course of the year. The financial support of the Center for
Health and Well-Being is gratefully acknowledged.
Every parent knows that a small
child sick with an earache may
not sit still to listen to a story,
indeed may not listen at all,
until she recovers. For some
chronically ill children, the struggle to
achieve academically may go on throughout
childhood. This article explores some of the
health conditions most common to American
children, notes racial disparities in the health
of children, and asks how much disparities in
children?s health might contribute to the
racial gap in school readiness. Given the
growing recognition that school readiness encompasses
behavior as well as cognitive abilities,
I highlight the effects of health on both
domains.
Health problems can affect a child?s school
readiness both directly and indirectly. Lead
poisoning, for example, directly impairs a
child?s cognition and causes behavior problems.
Poor health can also affect readiness indirectly
by crowding out beneficial activities
and changing the way the family treats a
child. For example, parents who perceive a
child as frail or vulnerable may be overly protective.
They may coddle or inadequately discipline
the child or may discourage him or
her from engaging in activities that could
hone both academic and social skills. Maternal
health conditions and health-related behaviors
may also have consequences for a
child?s school readiness.
Clearly, health conditions can impair school
readiness in individual children. Whether
racial health differences are responsible for a
large fraction of the black-white gap in school
readiness is a more complex question. For
health problems to affect the gap, three conditions
must hold. First, the health problem
must affect many children. Severe illnesses
like childhood cancer are mercifully rare and
thus cannot explain the overall readiness gap
between black and white children. Second,
there must be a link between the health condition
in question and academic performance
or behavior problems. Health disparities that
do not affect children?s academic achievement
or behavior cannot contribute to gaps
in achievement or behavior. Third, there
must be a racial gap either in the prevalence
of the health problem or in its effects.
These same considerations have guided my
choice of which health problems to address.
Because space constraints make it impossible
to discuss the possible contribution of every
health condition, let alone every type of
health behavior, I focus on health conditions
and behaviors that affect many children or
that affect children in some racial groups
much more than in others. I also focus on
health conditions whose connection with
school readiness has been documented by research.
Racial disparities in childhood injuries,
for example, are large, but little research
links these gaps to school readiness.
Finally, I focus on five broad health domains:
mental health conditions, chronic conditions,
environmental threats, nutrition, and maternal
health and behaviors. Within those domains,
the specific topics are attention deficit
hyperactivity disorder (ADHD), asthma, lead
poisoning, anemia and iron deficiency,
breastfeeding, and maternal depression. I
consider maternal health and behaviors because
they may have larger effects on racial
disparities in school readiness than do most
of the children?s health conditions.
Within each area, I highlight studies based on
large samples and good research designs. I
focus on black-white and poor-nonpoor gaps
in health status because most studies of disparities
in health discuss these contrasts.
Poor-nonpoor gaps are relevant because
J a n e t C u r r i e
118 THE FUTURE OF CHILDREN
black children tend to be poorer than nonblack
children. In 2002, for example, 37.5
percent of black children under the age of
five were poor, compared with 15.5 percent
of white children.1
Although some of the specific health conditions
considered here have large effects on
children?s cognitive skills and behavior, most
explain little of the overall racial gap in school
readiness. Still, the total cumulative effect of
health differentials summed over all conditions
is significant. ?Back-of-the-envelope?
calculations indicate that racial differences in
health conditions and in maternal health and
behaviors together may account for as much as
a quarter of the racial gap in school readiness.
Health Conditions and
School Readiness
This section considers several specific types
of health problems including child mental
health problems, chronic physical conditions,
environmental hazards, and poor nutrition.
The impact of maternal health conditions and
behaviors is considered in the next section.
Child Mental Health Problems
According to the 1999 U.S. surgeon general?s
report, approximately one in five children
and adolescents in the United States has
symptoms of mental or behavioral disorders.
Attention deficit hyperactivity disorder, the
most commonly diagnosed chronic mental
health problem among young children, is the
focus of this section. The disorder is characterized
by an inability to pay attention (inattention)
or by hyperactivity, or both.2
Children with ADHD are not school ready,
almost by definition. They have great difficulty
with basic tasks such as sitting still and
listening to instructions. They are likely to be
disruptive and to have trouble getting along
with other children because, for example,
they constantly interrupt and have trouble
taking turns. The disorder is also often linked
with cognitive impairments.
A diagnosis of ADHD has three main criteria.
Six or more symptoms of inattention or of
hyperactivity must persist for at least six
months to a degree that is maladaptive and
inconsistent with the child?s developmental
level. Some of the symptoms must be present
before the child reaches the age of seven.
And impairment from the symptom must be
evident in two or more settings, such as home
and school. This last criterion means that
teachers are often important for the diagnosis
of ADHD.3
Assessing the prevalence of ADHD is complicated.
Most studies of its prevalence are
based on diagnosed cases, but considerable
controversy exists over whether the disorder
is over- (or under-) diagnosed. Data from the
National Institute of Mental Health?s Epidemiology
of Child and Adolescent Mental
Health Disorders (MECA) study of 1,285
youths aged nine through seventeen indicate
that 5.1 percent of the children had ADHD.
A study of 21,065 children aged four to fifteen
recruited from 401 family medical practices
found that 9.2 percent had ?attention
deficit-hyperactivity problems? according to
their clinician, but that the clinicians did not
generally use standard diagnostic criteria.4
According to the hyperactivity subscale of the
Strengths and Difficulties Questionnaire of
the National Health Interview Survey, 4.19
percent of boys and 1.77 percent of girls have
?clinically significant? ADHD symptoms.
Among boys, the prevalence is highest among
blacks, at 5.65 percent, as against 4.33 percent
for whites and 3.06 percent for Hispanics.
Prevalence is also higher (6.52 percent)
H e a l t h D i s p a r i t i e s a n d G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 119
in families with incomes less than $20,000
than in families with higher incomes (3.85
percent). When gender, race, age, income,
and parental education are taken into account,
the effect of income remains statistically
significant, but there is no difference in
prevalence between blacks and whites.5
Although drug therapy improves behavior for
approximately 70 to 80 percent of ADHD
children, the evidence that treatment affects
academic performance is much less conclusive.
6 Treatment differs widely by race and
income. Data from the National Health Interview
Survey indicate that the share of parents
who had ever been told that their child
had ADHD was 7.5 percent for whites, 5.7
percent for blacks, and 3.5 percent for Hispanics.
For poor children the rates were 7.1
percent as against 6.6 percent for nonpoor
children. According to the 1997 Medical Expenditure
Panel, 4.4 percent of whites but
only 1.7 percent of blacks were treated for
ADHD, though the probability of receiving
treatment varied little by income. In a Maryland
study of Medicaid patients, blacks were
less than half as likely to have been prescribed
psychotropic drugs as whites were,
indicating that even among children with
similar insurance coverage, treatment patterns
differ by race.7
In one study, teachers were given profiles of
students and asked whether they had
ADHD. The race and gender assigned to the
profiles were randomly varied. Teachers were
most likely to believe that white males had
ADHD and least likely to think that white females
had the disorder, with black students
falling in between. A study based on random
telephone interviews found that in a sample
of 381 high-risk children, 91 percent of the
white parents and 85 percent of the black
parents believed that their child had a problem.
Fifty-one percent of the white children
had been evaluated for ADHD as against
only 28 percent of the black children. Rates
of treatment were 31 percent for whites and
15 percent for blacks. Following up on children
who were diagnosed but not treated,
the researchers found that blacks were more
likely than whites to cite negative expectations
about the treatment (58 percent versus
34 percent), stigma (47 percent versus 32
percent), and financial constraints (32 percent
versus 15 percent).8
Using survey data that followed a group of
children from the United States and Canada,
Mark Stabile and I show that children with
ADHD not only perform more poorly than
children without the disorder on cognitive
tests, but also are at greater risk of having to
repeat a grade and to enroll in special education,
even after controlling for a wide range
of potential confounders. ADHD affects
cognition and behavior more than other
chronic health conditions, such as asthma, or
poor health generally. Our estimates imply
that children with ADHD score at least a
quarter of a standard deviation lower on
standardized tests of mathematics and reading
than other children. Surprisingly, the effect
of ADHD on cognitive and scholastic
outcomes is not strongly related to income in
either country.9
J a n e t C u r r i e
120 THE FUTURE OF CHILDREN
Teachers were most likely to
believe that white males had
ADHD and least likely to
think that white females had
the disorder, with black
students falling in between.
How much of the racial gap in school readiness
might be accounted for by ADHD? Suppose
that a generic test has a mean of 50 and
a standard deviation of 15 and that black children
tend to score at least a half a standard
deviation (8 points) lower than white children
on this test. The studies discussed above
suggest that ADHD lowers test scores by
about a third of a standard deviation (5
points) and that about 4 percent of whites
have the disorder, compared with 6 percent
of blacks. Hence, if the difference in the
prevalence of ADHD were the only difference
between the black and white children,
one would expect the average test score of a
sample of white children to be 49.8, while the
average test score of a sample of black children
would be 49.10
This estimate, though crude, makes clear that
the mean test scores of blacks and whites are
driven by children who do not have any
health conditions. That being so, any given
health condition would have to have quite a
large effect (or a very different prevalence for
whites and blacks) before it could have much
effect on mean differences in test scores.
Chronic Physical Health Conditions
Poor children are more likely than better-off
children to suffer from a wide array of chronic
health problems, particularly severe conditions
such as mental retardation, heart problems,
poor hearing, and digestive disorders.
Chronic conditions affect school readiness in
various ways. First, illness may simply crowd
out other activities with doctor visits and treatment.
Second, children with chronic conditions
may experience more stress, fatigue, or
pain that can interfere with cognitive development.
Third, drugs used to treat some illnesses
may have unanticipated effects. Fourth, illness
may alter relations between children, parents,
and others in a way harmful to the child?s development.
Fifth, illnesses directly affect the
ability to learn, by altering body chemistry.11
This section focuses on asthma. Not only is
asthma one of the most common chronic
conditions among children, but it is also the
subject of much research focused both on
black-white gaps in prevalence and on the relationship
between asthma and measures of
cognitive achievement and behavior.
Asthma is the leading cause of children?s trips
to the emergency room, of their being hospitalized,
and of their being absent from
school. An ?asthmatic? child is one who has
had an episode of blocked airways or who has
a tendency toward such episodes. Doctors
use different methods to diagnose asthma,
and diagnosis depends on the child?s either
having an episode or being treated for
breathing or wheezing problems. Children
whose asthma is adequately managed should
not have acute attacks. Prevalence surveys
that focus on doctor diagnoses and those that
focus on asthma attacks, therefore, lead to
very different estimates.
According to the 2001 National Health Interview
Survey (NHIS), 13 percent of children
under age eighteen have been diagnosed
with asthma, and 6 percent have had an
asthma attack in the past twelve months.
Prevalence rates in diagnosed asthma are
higher for blacks (15.7 percent) than for
whites (12.2 percent) but lowest for Hispanics
(11.2 percent). Rates are also higher for
poor children (15.8 percent) than nonpoor
children (12 percent). Among black children,
7.7 percent had an attack in the past twelve
months, as against 5.7 percent of whites and
only 4 percent of Hispanics.12
The NHIS further shows that 1.6 percent of
white children under age eighteen, and
H e a l t h D i s p a r i t i e s a n d G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 121
J a n e t C u r r i e
122 THE FUTURE OF CHILDREN
Common Chronic Childhood Conditions
Three common chronic conditions?dental caries, allergies, and ear infections?are potentially
implicated in cognitive and behavior problems in children, but research is not yet far enough along
to make it possible to estimate how large those effects might be.
Dental caries (tooth decay) is the most common childhood chronic condition. Chronic pain from
dental disease can affect both children?s cognitive attainment and their behavior. According to the
Centers for Disease Control, poor children have almost twelve times more restricted-activity days
because of dental problems than do higher-income children, and untreated dental disease can
lead to problems of eating, speaking, and learning. It is, however, difficult to get estimates of the
size of these effects.1
Not only is tooth decay extremely common, but it also affect blacks more than whites, so that if it
does significantly affect children?s learning and behavior, then it could contribute to disparities in
school readiness. White, black, and Hispanic children have about the same number of decayed,
missing, or filled teeth, suggesting that the rates of tooth decay are similar. But among two- to
five-year-old children, 14.4 percent of white children have untreated dental caries, as against
25.1 percent for black children and 34.9 percent for Hispanic children.2
Allergies are also extremely common. According to the 2002 National Health Interview Survey,
10.3 percent of children have hay fever, 12.3 percent have respiratory allergies, and 11.3 percent
have other allergies. (These categories are not mutually exclusive, so the share of children with
any allergy is less than the sum of these percentages.) Assessing the prevalence of allergies is
complicated because of serious reporting problems. For example, the probability that a parent reports
an allergy increases with income and education; it is lower for blacks than for whites even
though asthma, which is often associated with allergies, is much more common among blacks.
Given these problems, and the fact that allergies may range from mild to life threatening, it is difficult
to say how much of the gap in school readiness might be attributable to allergies.3
Ear infections (otitis media) affect most young children at one time or another and are the most
common reason why children visit a doctor. Like dental caries, they can be extremely painful,
though more than 80 percent of infections resolve themselves within three days if untreated.
Among children who have had acute otitis media, almost half have persistent effusion after one
month, a condition that can cause hearing loss. Researchers estimate that at any given time
roughly 5 percent of two- to four-year-old children have hearing loss because of middle ear effusion
lasting three months or longer. And hearing loss can delay language development. But the
prevalence of ear infections does not appear to differ between blacks and whites, which suggests
that otitis media cannot be responsible for gaps in school readiness.4
1. Centers for Disease Control, Preventing Chronic Diseases: Investing Wisely in Health, Preventing Dental Caries (U.S. Department of
Health and Human Services, April 6, 2004).
2. Linda M. Kaste and others, ?Coronal Caries in the Primary and Permanent Dentition of Children and Adolescents Ages 1 to 17 Years:
United States, 1988?1991,? Journal of Dental Research 75 (February 1996): 631?41.
3. Achintya N. Dey and others, Summary Health Statistics for U.S. Children: National Health Interview Survey, 2002, Vital Health Statistics
Series 10, no. 221 (Hyattsville, Md.: National Center for Health Statistics, March 2004).
4. Paddy O?Neill, ?Acute Otitis Media,? British Medical Journal (September 25, 1999); Richard Thrasher and Gregory Allen, Ear, Otitis Media
with Effusion (www.emedicine.com/ENT/topic209.htm [December 13, 2002]).
5.7 percent of black children, had been hospitalized
for asthma between 1998 and 1999.
The disparity in hospitalizations is much
greater than that in the number of attacks,
suggesting that black children?s asthma is either
much more serious or much less likely to
be controlled. This conclusion is supported
by the finding that blacks were more likely
than whites to have their activity limited because
of asthma (32.7 percent compared with
21.4 percent). Similar disparities in morbidity
were noted between poor and nonpoor children
(33.2 percent vs. 20.8 percent), but poor
black children were most likely to have activity
limited because of asthma (49 percent as
against about 20 percent for nonpoor black or
white children or for poor white children).13
Consistent with these observations, several
smaller-scale studies have noted that doctors
are less likely to prescribe inhaled antiinflammatory
drugs for minorities than for
whites. One study using nationally representative
data from the National Health and Nutrition
Examination Survey (NHANES) III
focuses on children with moderate to severe
asthma (defined as having been hospitalized
or having two or more acute attacks or three
or more episodes of wheezing over the past
year) and finds that only 26 percent of these
children were taking maintenance medication.
In this group, children who have Medicaid
insurance and who speak Spanish are
more likely to be inadequately medicated for
asthma. Race is not an independent factor.14
Many research papers suggest, perhaps surprisingly,
that asthma has little effect on cognitive
outcomes or schooling attainment.
Most such studies, however, examine children
whose asthma is well controlled. Indeed,
the purpose of such studies is to see
whether the medication children take to control
their asthma affects their cognitive functioning.
But several studies indicate that children
with asthma are more likely than other
children to have behavior problems, even
when the asthma is controlled. For example,
one study found that asthmatic children
scored between two-thirds to one standard
deviation below the normative value on a test
of impulse control, while another found that
asthma doubled the risk of behavioral problems.
These changes in behavior may reflect
relatively subtle effects of childhood illness
on parenting and family functioning.15
One large population-based study using
NHIS data found that asthma affected school
absences, the probability of having learning
disabilities, and grade repetition. Asthmatic
children in grades one to twelve were absent
from school an average of 7.6 days a year as
against 2.5 days for well children. Nine percent
of the asthmatic children (5 percent of
the well children) had learning disabilities; 18
percent (15 percent of the well children) repeated
a grade.16
In the only study to examine school readiness
explicitly, Jennifer Halterman and her collaborators
examine 1,058 children entering
kindergarten in urban Rochester and find
that asthmatic children had lower scores on a
test of school readiness skills and that their
parents were three times more likely to report
that they needed extra help with learning.
Tests of language, motor, and socioemotional
skills showed no differences. The
negative effects were concentrated among a
group of children whose asthma was severe
enough to limit their activity (suggesting that
it was not adequately controlled), a group
more likely to include boys than girls.17
One difficulty in interpreting all these studies
is that because asthma is most prevalent
among poor and minority children, the ap-
H e a l t h D i s p a r i t i e s a n d G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 123
parent effect of asthma on academic performance
and behavior could reflect omitted
third factors. But several studies of homogeneous
groups of children also find differences
in behavior, suggesting that asthma probably
does have a causal effect at least on behavior
problems and hence on school readiness.
A back-of-the-envelope calculation similar to
that for ADHD can help determine whether
these differences are large enough to affect
the mean test score gap. The studies discussed
above suggest that some 5 percent of
black children, but only 3 percent of white
children, have asthma severe enough to limit
their activity. The major effect of asthma is
on behavior, so I will assume that asthmatic
children score a standard deviation higher on
a behavior-problem index than do nonasthmatic
children and that the index has the
same characteristics as the generic test score
assumed above (that is, mean of 50, standard
deviation of 15, average black-white difference
of 8). Under these assumptions, the average
behavior-problem score among blacks
would be 50.4; that among whites, 50.2.
Again, although asthma has important effects
on individual children, it cannot account for
much of the racial gap in measures of school
readiness.18
Environmental Exposures to
Hazardous Substances
The literature on asthma strongly suggests
that its greater prevalence among impoverished
children could be due in part to characteristics
of their housing. The degree of segregation
by race, ethnicity, and income in
American cities suggests that some groups
are more likely than others to be exposed to
environmental hazards. Moreover, to the extent
that known environmental hazards are
capitalized into housing prices, pollution will
lower rents, making hazardous areas more attractive
to poor people than to rich ones.
Conversely, low land prices in poor neighborhoods
may draw in new hazards. One environmental
hazard whose effect on children?s
health has been studied extensively is lead.
Lead has long been known to be toxic. Blood
lead levels above 45 micrograms per deciliter
(microg/dl) can cause damage to the central
nervous system and even death. For many
years, the Centers for Disease Control set 30
microg/dl as the threshold ?level of concern?
for lead poisoning. But in response to evidence
that levels as low as 10 microg/dl could
affect children?s cognitive functioning and
behavior, the CDC lowered the threshold to
25 microg/dl in 1985 and to 10 microg/dl in
1991. Controversy now centers on whether
even lower levels of lead endanger children,
who are generally at higher risk from lead
than adults. In adults only organic lead compounds
can breach the blood-brain barrier; in
children, both organic and inorganic lead can
penetrate that barrier. And children who
have diets deficient in calcium, iron, and zinc
tend to absorb more lead.19
Before the federal government began to regulate
lead, children were exposed to it in
paints, in drinking water (from lead solder in
pipes), in gasoline, and in canned food. According
to the NHANES surveys, 88.2 percent
of children aged one to five had lead levels
above 10 microg/dl during 1976?80. That
share plummeted to 8.6 percent during
1988?91 and fell further to 2.2 percent during
1999?2000?figures that imply that the
number of children with unsafe lead levels
fell from 13.5 million to less than half a million
over this period.20
Still, lead remains in the soil, in paint in older
homes, and in pipes. Some states still have
lead ?hot spots.? One study reported that 68
J a n e t C u r r i e
124 THE FUTURE OF CHILDREN
percent of children attending a pediatric
clinic in inner-city Philadelphia had unsafe
levels of lead in their blood. Poor and black
children are more likely than others to have
unsafe levels.21
The NHANES data from 1999?2000 and data
from state surveillance systems indicate that
60 percent of one- to five-year-old children
with confirmed elevated blood lead levels between
1997 and 2001 were black, indicating a
much higher prevalence among blacks than
among whites. In 2001, 2 percent of white
children and 8.7 percent of black children had
confirmed high blood lead levels. The condition
affects more boys than girls. In 2001, for
example, 40,000 boys and 33,000 girls were
confirmed to have high levels.22
Although some studies have found that increasing
blood lead levels from 10 to 20 microg/
dl reduces IQ scores by as much as 7
points (where one standard deviation is about
15 points), two reviews of many studies of
blood lead levels conclude that such an increase
would reduce IQ by about 2 points. Elevated
lead levels have also been linked to hyperactivity
and behavior problems, most
famously by Herbert Needleman, who argues
that lead exposure causes criminal behavior. In
his study, a sample of delinquents was four
times more likely to have high bone lead levels
than a group of matched controls. But because
lead exposure is increasingly strongly correlated
with minority status, poverty, and residence
in decaying older neighborhoods, it is
possible that at least some of the observed correlations
between lead levels and negative outcomes
reflect omitted third factors. These estimates
of the effects of low-level lead exposure
should thus be regarded as upper bounds.23
A calculation similar to those made for
ADHD and asthma suggests that differing
exposure to lead might be responsible for 0.2
point of the average eight-point racial gap in
scores assumed above. If racial disparities in
exposure to other environmental hazards
have also grown, exposure to such hazards
could be an increasingly important cause of
disparities in school readiness.24
Nutrition
U.S. food and nutrition programs were created
to ensure that children and other vulnerable
people would get enough to eat.
Only recently have researchers and policymakers
begun to recognize that many if not
most children eat too much of the wrong
things and that obesity is a greater threat to
child health than insufficient calories. In fact,
children at risk of missing meals (those who
are ?food insecure?) are more likely to be
obese than other children, although they are
also more likely to be lacking specific micronutrients.
Similarly, poor children from
birth to age five are twice as likely as betteroff
children to be obese, about a third more
likely to be anemic, and about 20 percent
more likely to be deficient in vitamin A. It is
possible that many micronutrients will be
found to affect cognitive development among
young children. But because most research
H e a l t h D i s p a r i t i e s a n d G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 125
One study reported that
68 percent of children
attending a pediatric clinic
in inner-city Philadelphia
had unsafe levels of lead in
their blood. Poor and black
children are more likely than
others to have unsafe levels.
to date on the effects of nutrition on cognition
has focused on iron-deficiency anemia,
that will be the focus of this section.25
Among its many negative effects on health,
iron deficiency impairs immune function,
cognitive functioning, and energy metabolism.
Clinically, iron deficiency is defined as
having an abnormal value on at least two out
of three laboratory tests of iron status. Anemia,
a more severe condition, is defined as
iron deficiency plus low hemoglobin.
When infants are about four months old, they
begin to deplete the stores of iron with which
they are born. The widespread use of ironfortified
infant formula and cereals has made
anemia much less of a problem in infants
under one year. But toddlers may stop eating
these iron-fortified infant foods before they
begin to gain adequate iron from their diet.
According to the NHANES III, 9 percent of
toddlers are iron deficient, as against 3 percent
of three- to five-year-olds and 2 percent
of six- to eleven-year-olds. Only 3 percent of
toddlers are anemic, and less than 1 percent
of children aged three to eleven are anemic.
The NHANES 1999?2000 yields similar estimates.
These anemia rates are down considerably
from 15?30 percent in the late 1970s
and early 1980s, a decline variously attributed
to iron-fortified foods and the growth of
the Special Supplemental Nutrition Program
for Woman, Infants, and Children (WIC), a
federal program that offers food supplements
to pregnant, lactating, and postpartum mothers,
infants, and children younger than five.26
Iron deficiency is much more common
among poor and black children than among
other children. Twice as many black children
as white children are iron deficient (16 percent
versus 8 percent for toddlers), while poor
children are more than 50 percent more likely
to be deficient than nonpoor children. If iron
deficiency impairs cognitive functioning, it
could well be responsible for part of the test
score disparities between blacks and whites
and between poor and nonpoor children.
Sally Grantham-McGregor and Cornelius
Ani reviewed observational studies that followed
a group of children over time and
found that conditional on measures of social
background, gender, and birth weight, low
hemoglobin levels in children aged two or
younger are strongly linked to poor schooling
achievement, cognitive development, and
motor development in middle childhood.
These studies, however, do not establish a
causal relationship, given the strong association
between iron deficiency and other factors
that could affect development, such as
poverty.27
Grantham-McGregor and Ani also survey
studies of trials in which anemic or irondeficient
children were given iron supplements.
They find that giving anemic children
iron supplements for two to six months improves
cognitive functioning, although not
enough to allow school-age children to catch
up to their non-anemic peers. Five small-
J a n e t C u r r i e
126 THE FUTURE OF CHILDREN
If iron deficiency impairs
cognitive functioning,
it could well be responsible
for part of the test score
disparities between blacks
and whites and between
poor and nonpoor children.
scale studies (four in developing countries)
that investigated the effect of iron supplementation
on iron-deficient non-anemic children
found little evidence of an effect on
cognition, but it is possible that subtle effects
of improving iron status in these samples of
children without anemia might be detectable
in larger samples.
In short, although the higher rates of iron deficiency
among poor and minority children
are a cause for concern, little concrete evidence
links these disparities to gaps in cognitive
outcomes or schooling attainment. Anemia
itself, which has been more definitively
linked to cognitive deficits and poorer
schooling attainment, has become relatively
rare, even among disadvantaged children. Although
anemia may have contributed to the
readiness gap in the past, it is unlikely to be a
major contributor today.
The Importance of Maternal
Health Conditions and Behaviors
In this section I focus on two aspects of maternal
health conditions and behaviors that
significantly affect children?s cognitive and
social functioning and that are also characterized
by large racial disparities. Because many
other maternal health behaviors could be
considered, my purpose here is merely to illustrate
how potentially important maternal
behaviors can be.
Breast Feeding
The first behavior, breast feeding, exhibits
large disparities by race. The American Academy
of Pediatrics recommends that infants
be breast fed exclusively for their first six
months and that cow?s milk not be introduced
until after the first birthday. Some 70 percent
of white infants, but only 40 percent of black
infants, have ever been breast fed. At six
months, 29 percent of white infants, but only
9 percent of black infants, are still being
breast fed.28
Theoretically, breast feeding affects a child?s
cognitive development through three channels.
First, it prevents diseases such as ear infections
and may even prevent asthma. To
the extent that poor physical health impairs
children?s performance, a lack of breast feeding
could thus be implicated. Second, breast
feeding provides nutrients, such as longchain
fatty acids that may affect infants? brain
development, that are not adequately provided
in most infant formula sold in the
United States. Third, breast feeding may promote
maternal-infant bonding that may, in
turn, be beneficial for learning. Many studies
link breast feeding positively with cognitive
skills. Typically they find IQ gains of two to
five points for healthy infants and up to eight
points for low birth weight babies. Once
again, however, given the strong relationship
between breast feeding and various measures
of socioeconomic status, it is unclear whether
the association between breast feeding and
cognition is causal.29
If, however, breast feeding does affect IQ
scores, then the racial differences in prevalence
are large enough to explain a significant
part of the gap in the generic test score that I
have been considering. Suppose, for example,
that breast feeding for six months raises
IQ by five points, or about one-third of a
standard deviation. Then the fact that 29 percent
of white infants, but only 9 percent of
black infants, are breast fed for six months
would generate a one point difference in average
scores (with the assumed black-white
gap being eight points).30
Maternal Depression
Although my emphasis in this article has
been on child health, the mental health of
H e a l t h D i s p a r i t i e s a n d G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 127
the mother may be a key determinant of the
health of the child. The difficulties associated
with poverty or racism, or both, may leave
some mothers more vulnerable to depression,
and depressed mothers may be less able
than healthy mothers to provide a stimulating
and nurturing environment for their children.
The hypothesis that differences in rates
of maternal depression could be associated
with group-level differences in the attainments
of children, however, has not been directly
tested, so it is necessary to go through
each link in the causal chain.
Evidence abounds that poverty is associated
with a higher risk of depression. The poor are
2.3 times more likely to be depressed than
the nonpoor, adjusting for age, gender, ethnicity,
and prior history of depression. This
higher risk may be due both to heightened
stress and to a lack of resources to cope with
that stress. The incidence of pregnancy and
postpartum depression in a sample of poor,
inner-city women is about one-quarter, double
the rate typically found among middleclass
women. In the Infant Health and Development
Study, 28 percent of poor
mothers, as against 17 percent of nonpoor
mothers, were depressed.31
Given that blacks are generally poorer than
whites, one might expect a higher prevalence
of depression among black mothers than
among white mothers. But research findings
are mixed. Some studies have shown higher
rates of depressive symptoms among blacks
than whites, but studies that use the diagnostic
criteria for major depression generally
find little racial difference in incidence. The
National Comorbidity Study and Epidemiological
Catchment Area Studies found that
blacks were less likely than whites to be depressed,
whereas another study found no
racial difference in the incidence of depression
in a sample of poor women. These findings
suggest that although poor mothers may
be at higher risk than others, race does not
play an independent role in explaining the incidence
of maternal depression. It is possible
that both race and socioeconomic status affect
whether, and how effectively, women are
treated for depression, but there is little hard
evidence that race, per se, is a factor.32
Studies of the relationship between maternal
depression and child development can be divided
into several groups. First, observational
studies of the way depressed mothers interact
with their infants find that they are often inconsistent
and ineffective in disciplining their
children, more likely to use force rather than
compromise, and less likely to interact in a
positive way. These problems are more apparent
among impoverished mothers with depression
than among their better-off counterparts.
Second, many studies document a
relationship between maternal depression
and both current and future child behavior
problems, insecure attachment, and cognitive
problems. Maternal depression, they find, can
reduce test scores by about a third of a standard
deviation among preschool children.33
It is not clear that maternal depression causes
these negative outcomes: the link between
the two could also reflect shared genes or a
shared response of the mother and child to
other external causes. It is also unclear how
pervasive or persistent child responses to maternal
depression are. Several studies, for example,
find the effects of postpartum depression
confined to boys.34
With 37.5 percent of black children under
five and 15.5 percent of white children in
that same age group living in poverty, the socioeconomic
gap in the incidence of maternal
depression noted above?28 percent among
J a n e t C u r r i e
128 THE FUTURE OF CHILDREN
the poor, 17 percent among the nonpoor?
means that maternal depression will affect
some 11 percent of black preschool children
but only 3 percent of white preschool children.
These differing exposures to maternal
depression could account for a half a point of
the assumed eight-point gap in our generic
average test score.35
Potential Policy Responses
Potential policy responses considered here
include measures aimed at reducing disparities
in access to health care, early intervention
programs, family services, and WIC
(the Supplemental Nutrition Program for
Women, Infants, and Children).
Reducing Disparities in Access
to Health Care
Disadvantaged children are not only more
likely than better-off children to have particular
health conditions, they are also less likely to
be treated for them. Could differences in access
to care be responsible for differences in
use of care? Although lack of insurance coverage
remains a serious problem for many children,
past expansions of public health insurance
under Medicaid and the State Children?s
Health Insurance Program (SCHIP) mean
that most poor and near-poor children are already
eligible for public health insurance. This
journal devoted its spring 2003 issue to a discussion
of health insurance for children and
concluded that ?programs already in place
have the potential to virtually eliminate uninsurance
among low-income children.?36
Making more children eligible for care is unlikely
to reduce health disparities greatly because
the most disadvantaged children are already
eligible (though reductions in eligibility
in many states could undo recent progress).
More to the point, many eligible children are
not signed up for public health insurance
until they have an urgent medical problem.
Thus they do not get preventive care. A Medicaid-
eligible child suffering an asthma attack
will be treated, but if she is not enrolled, she
may not receive the monitoring and medication
needed to prevent another attack. The
children with the poorest access to specialists
are those in families with incomes between
125 percent and 200 percent of poverty, even
though many are eligible for SCHIP.37
One way to improve access to care among
children eligible for public health insurance
may be to make it easier to sign up for, and to
maintain, Medicaid coverage. When Jeffrey
Grogger and I examined several state efforts
to streamline the Medicaid application
process, such as shortening application forms
and allowing mail-in applications, we found
little evidence that they were effective. By
contrast, Anna Aizer found that paying community
organizations to help families sign up
for public health insurance in California increased
enrollments among Hispanic and
Asian families and reduced preventable hospitalizations.
Because take-up of social programs
is highest when enrollment is automatic,
the best approach to the problem of
eligible, unenrolled children may be to make
H e a l t h D i s p a r i t i e s a n d G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 129
Although lack of insurance
coverage remains a serious
problem for many children,
past expansions of public
health insurance mean that
most poor and near-poor
children are already eligible
for public health insurance.
all children eligible for Medicaid services and
charge premiums on a sliding scale.38
But further expanding public health insurance
is unlikely ever to eliminate all socioeconomic
disparities in health. The famous 1980 Black
report in Great Britain concluded that links
between socioeconomic status and health became
more pronounced following the advent
of national health insurance in 1948?although
it is possible that the socioeconomic
gap would have widened even further in the
absence of the National Health Service. Moreover,
despite universal take-up of national
health insurance in Britain, the rich receive
more services than the poor, conditional on
their health status. Health is also linked to
household income in Canada, even though
Canadians have universal health insurance.39
A final consideration is that health care
providers are not always trained to offer the
services that children and their mothers require.
A recent study found that pediatricians
rarely recognized depressive symptoms in
most mothers, suggesting that increasing access
to these providers would not necessarily
help children whose problems were linked to
maternal depression.40
Early Childhood Intervention Programs
Most early intervention programs include a
significant health component, in the belief
that they cannot address educational needs
without also addressing health problems. Because
many different children?s programs already
address specific health problems (for
example, by screening for lead poisoning or
by focusing on child nutrition), it may seem
irrational to make health a major focus of educationally
oriented early intervention programs.
But to take advantage of existing
health programs, parents must be knowledgeable
and tireless advocates for their children.
And parents who are struggling to put
bread on the table may not have the time or
energy to get all the services their children
need. Hence the potential value of quality infant
and preschool programs that offer ?onestop
shopping? for these services. Staff members
in such programs may be better than
parents at spotting problems and also more
knowledgeable about community resources.
But researchers have not yet systematically
assessed the importance and effectiveness of
the health services component of early intervention
programs.41
Head Start, the federal program serving disadvantaged
three- to five-year-old children,
mandates that children receive the health assessments
and services that they need. A 1984
Abt Associates study, now quite dated, randomly
assigned children in four sites to Head
Start treatments and non?Head Start controls
and evaluated the health services the children
received. The children entering Head Start
had many and serious health problems. They
had an average of 4.6 unfilled cavities; 34 percent
scored below the 10th percentile for fine
and gross motor skills for their age; 63 percent
had a speech or language problem; and
one-third failed the hearing test. Fourteen
percent had active otitis media.42
Although the Abt study found that compliance
with Head Start health performance
standards was imperfect, the Head Start children
were significantly more likely than the
control children to have received medical
screenings and necessary services. It is also
worth stressing that Head Start has detailed
performance standards for health services
and that programs are regularly evaluated
with respect to indicators such as the fraction
of children who have received dental examinations,
hearing and vision screenings, and
immunizations.
J a n e t C u r r i e
130 THE FUTURE OF CHILDREN
Using data from Head Start budgets and
from the National Longitudinal Survey of
Youth, Matthew Neidell and I found that
Head Start programs that spend a larger
share of their budgets on health and education
raise future child test scores more than
do programs that spend higher shares on
other types of programming, such as programs
for parents.43
Given the large socioeconomic disparities in
health in the United States, it may well be
that the health services offered by early intervention
programs play an important role in
improving the cognitive functioning and future
schooling attainments of impoverished
children. The programs do not seem to duplicate
services, but rather to help children get
the services for which they are eligible
through other programs.
Family-Based Services
Offering health services through programs
such as Head Start will not reach all needy
children, both because not all eligible children
enroll and because not all needy children
are eligible. Home visiting programs
and other family-centered programs offer an
alternative model for service delivery. The
most successful of these programs are those
associated with David Olds.44
Olds?s programs, which focus on families at
risk because the mother is young, poor, uneducated,
and unmarried, involve nurse visits
from the prenatal period until the child turns
two. Evaluators have documented many positive
effects on both maternal behavior and
children?s health. As of age two, children in
one study site were much less likely than control
children to have visited a hospital emergency
room for unintentional injuries or ingestion
of poisonous substances, although
this finding was not replicated at other study
sites. As of age fifteen, children of visited
mothers were less likely to have been arrested
or run away from home, had fewer
sexual partners, and smoked and drank less.
These children were also less likely to have
been involved in verified incidents of child
maltreatment. There was little evidence of
effects on cognition at four years of age (except
among children of initially heavy smokers),
though the reduction in delinquent behavior
among teens could be expected to
improve their school achievement. These
studies suggest that locating children at risk
and ensuring that they receive necessary
services would be a useful complement to
other strategies for reducing disparities in
child health.
The Special Supplemental Nutrition
Program for Women, Infants, and Children
The WIC program probably already plays a
large role in remediating health disparities
that could lead to gaps in school readiness. It
has, for example, been credited with the dramatic
decline in the incidence of anemia
among young children between 1975 (shortly
after it was introduced) and 1985. Several
studies indicate that these improvements in
nutrition affect children?s behavior and ability
to learn. Children whose mothers were on
WIC during the prenatal period score higher
than children not on WIC on the Peabody
Picture Vocabulary Test, a good predictor of
future scholastic achievement.45
In any given month in 1998, 58 percent of all
infants were eligible for WIC and roughly 45
percent of all infants received benefits.
Among children aged one to four, 57 percent
were eligible for WIC and 38 percent of eligible
children received benefits. Participation
tends to drop off sharply after a child?s first
birthday, when WIC stops providing valuable
infant formula.46
H e a l t h D i s p a r i t i e s a n d G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 131
The program offers participants coupons that
can be used only to purchase specific commodities
that meet the nutritional needs of
pregnant or nursing women, infants, and children
under five. It is a promising vehicle for
addressing health disparities in other respects
as well. First, WIC agencies have frequent
contact with participants, who typically come
in at least once quarterly to pick up coupons
and get nutritional counseling. Second, the
agencies are required to help participants get
preventive health care by providing services
on-site or through referrals. Third, agencies
teach pregnant women that ?breast is best,?
although they may undermine this message
by providing free infant formula to women
who choose not to breast feed.
Because WIC already serves many children
who receive inadequate health care and because
it is strongly linked to the provision of
health services, it is worth considering
whether WIC could do more to reduce
health disparities. Further promoting breast
feeding would be particularly worthwhile, as
would offering screenings and referrals for
maternal depression. Keeping children in
the program beyond their first year could increase
access to health screenings and reduce
nutritional problems such as low iron
levels.
Discussion and Conclusions
That there are pervasive differences in health
between black and white children in the
United States is beyond doubt. But do these
disparities explain the racial gaps in school
readiness? The evidence assembled here suggests
that although many specific health conditions
impair cognition and behavior in individual
children, it is unlikely that any
particular condition can explain much of the
racial gap. For example, children with
ADHD score a third of a standard deviation
lower on test scores than children without
the disorder. But because ADHD affects relatively
few children and because racial differences
in its prevalence are small, it explains
little of the racial difference in school readiness.
This does not mean that ADHD or
other health conditions are unimportant.
Clearly ADHD often has devastating effects
on the 4 percent of boys and 2 percent of
girls it affects even if it does not explain much
of the racial gap in outcomes.
Moreover, summed over all health conditions,
health differentials could well explain a
sizeable portion of the racial gap. Three of
the conditions evaluated here?ADHD,
asthma, and lead poisoning?could explain
up to 0.6 of a point in the hypothetical 8
point gap used for illustrative purposes. Not
enough evidence is yet available to evaluate
how much other common conditions such as
injuries, ear infection, and dental caries could
contribute. But it would not be far-fetched to
suppose that differences in health conditions
might together explain one point, or an
eighth of the school readiness gap. And maternal
health and behaviors may have even
J a n e t C u r r i e
132 THE FUTURE OF CHILDREN
Because WIC already serves
many children who receive
inadequate health care and
because it is strongly linked
to the provision of health
services, it is worth
considering whether WIC
could do more to reduce
health disparities.
larger effects on racial gaps in school readiness
because they affect more children. After
all, the majority of children are in excellent
health, which means that mean gaps in test
scores are driven largely by children who do
not have health problems.
Simply summing the various estimates in this
paper suggests that as much of a quarter of
the readiness gap between blacks and whites
might be attributable to health conditions or
health behaviors of both mothers and children.
Summing yields an upper estimate, because
some children may be affected by more
than one condition or behavior. But these
findings confirm once again that mind and
body are intimately connected and that at
least some of the persistent gap in school
readiness between black and white children
may reflect differences in their health.
H e a l t h D i s p a r i t i e s a n d G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 133
Endnotes
1. U.S. Bureau of the Census, ?People in Families with Related Children under 18 by Family Structure, Age,
Sex, Iterated by Income-to-Poverty Ratio and Race? (http://ferret.bls.census.gov/macro/032003/pov/
new03_000.htm [2003]).
2. U.S. Department of Health and Human Services, Mental Health: A Report to the Surgeon General (1999).
3. American Psychiatric Association, Diagnostic and Statistical Manual of Mental Disorders, 4th ed. (Washington,
1994).
4. Peter S. Jensen and others, ?Are Stimulants Overprescribed? Treatment of ADHD in Four U.S. Communities,?
Journal of the American Academy of Child and Adolescent Psychiatry 38, no. 7 (July 1999): 797?804;
Richard C. Wasserman and others, ?Identification of Attentional and Hyperactivity Problems in Primary
Care: A Report from Pediatric Research in Office Settings and the Ambulatory Sentinel Practice Network,?
Pediatrics 103, no. 3 (March 1999): e38.
5. Steven Cuffe, Charity Moore, and Robert McKeown, ?ADHD Symptoms in the National Health Interview
Survey: Prevalence, Correlates, and Use of Services and Medication,? poster presented to the Fiftieth Anniversary
Meeting of the American Academy of Child and Adolescent Psychiatry, Miami, October 20, 2003.
6. James M. Swanson and others, ?Effects of Stimulant Medication on Learning in Children with ADHD,?
Journal of Learning Disabilities 24, no. 4 (April 1991): 219?30.
7. Barbara Bloom and others, Summary Health Statistics for U.S. Children: National Health Interview Survey,
2001, Vital and Health Statistics Series 10, number 216 (Hyattsville, Md.: National Center for Health Statistics,
2003); Mark Olfson and others, ?National Trends in the Treatment of Attention Deficit Hyperactivity
Disorder,? American Journal of Psychiatry 160, no. 6 (June 2003): 1071; Julie M. Zito and others,
?Methylphenidate Patterns among Medicaid Youths,? Psychopharmacology Bulletin 33, no. 1 (1997): 143?47.
8. Kelly B. Raymond, The Effect of Race and Gender on the Identification of Children with Attention Deficit
Hyperactivity Disorder (Ann Arbor, Mich.: UMI Company, 1997); Regina Bussing and others, ?Prevalence
of Behavior Problems in U.S. Children with Asthma,? Archives of Pediatric and Adolescent Medicine 149,
no. 5 (May 1995): 565?72.
9. Janet Currie and Mark Stabile, ?Child Mental Health and Human Capital Accumulation: The Case of
ADHD,? Working Paper (University of California at Los Angeles, Department of Economics, August 2004).
10. For whites, the mean score would be [(.96*50 ) + (.04*45)] = 49.8, and for blacks the mean score would be
[(.94*50) + (.06*45)] = 49.7.
11. Anne Case, Darren Lubotsky, and Christine Paxson, ?Economic Status and Health in Childhood: The Origins
of the Gradient,? American Economic Review 92, no. 5 (December 2002): 1308?34; Janet Currie and
Mark Stabile, ?Socioeconomic Status and Health: Why Is the Relationship Stronger for Older Children??
American Economic Review 93, no. 5 (December 2003): 1813?23; Paul W. Newacheck, ?Poverty and Childhood
Chronic Illness,? Archives of Pediatric and Adolescent Medicine 148 (1994): 1143?49.
12. Olfson and others, ?National Trends in the Treatment of Attention Deficit Hyperactivity Disorder? (see note 7).
13. Lara J. Akinbami, Bonnie J. LaFleur, and Kenneth C. Schoendorf, ?Racial and Income Disparities in Childhood
Asthma in the United States,? Ambulatory Pediatrics 2 (2002): 382?87.
J a n e t C u r r i e
134 THE FUTURE OF CHILDREN
14. Edwin D. Boudreaux and others, ?Multicenter Airway Research Collaboration Investigators,? Pediatrics
111, no. 5, part 1 (2003): 615?21; Tracy A. Lieu and others, ?Racial/Ethnic Variation in Asthma Status and
Management Practices among Children in Managed Medicaid,? Pediatrics 109, no. 5 (May 2002): 857?65;
Alexander N. Ortega and others, ?Impact of Site of Care, Race, and Hispanic Ethnicity on Medication Use
for Childhood Asthma,? Pediatrics 109, no. 1 (January 2002); Jill S. Halterman and others, ?School Readiness
among Urban Children with Asthma,? Ambulatory Pediatrics 1, no. 4 (July?August 2001): 201?05.
15. Scott Lindgren and others, ?Does Asthma or Treatment with Theophylline Limit Children?s Academic Performance??
New England Journal of Medicine 327, no. 13 (September 24, 1992): 926?30; Robert D. Annett
and others, ?Neurocognitive Functioning in Children with Mild and Moderate Asthma in the Childhood
Asthma Management Program,? Journal of Allergy and Clinical Immunology 105, no. 4 (April 2000):
717?24; Linda B. Gutstadt and others, ?Determinants of School Performance in Children with Chronic
Asthma,? American Journal of Diseases in Children 143, no. 4 (April 1989): 471?75; Rachel Calam and others,
?Childhood Asthma, Behavior Problems, and Family Functioning,? Journal of Allergy and Clinical Immunology
112, no. 3 (September 2003): 499?504; Arlene M. Butz and others, ?Social Factors Associated
with Behavioral Problems in Children with Asthma,? Clinical Pediatrics 34, no. 11 (November 1995):
581?90.
16. M. G. Fowler, M. G. Davenport, and Rekha Garg, ?School Functioning of U.S. Children with Asthma,? Pediatrics
90, no. 6 (December 1992): 939?44.
17. Halterman and others, ?School Readiness among Urban Children with Asthma? (see note 14).
18. The asthma studies suggest that 15.7 percent of black children have asthma and that 32.7 percent of black
asthmatics are limited by their condition. Among whites, the comparable figures are 12.2 percent and 22.4
percent. Together, these figures imply that approximately 5 percent of black children and 3 percent of white
children are limited by asthma. Hence, the average behavior problems score among whites would be
[(.97*50) + (.03*58)] = 50.2 compared with an average score among 100 black children of [(.95*50) +
(.05*58)] = 50.4 (where for behavior problems a higher score is worse).
19. Philip O?Dowd, ?Controversies Regarding Low Blood Lead Level Harm,? Medicine and Health, Rhode Island
85, no. 11 (November 2002): 345?48; Robert G. Feldman and Roberta F. White, ?Lead Neurotoxicity
and Disorders of Learning,? Journal of Child Neurology 7, no. 4 (October 1992): 354?59.
20. U.S. Centers for Disease Control, Children?s Blood Lead Levels in the United States (www.cdc.gov/nceh/
lead/research/kidsBLL.htm#Tracking BLL [March 12, 2003]).
21. Shoshana T. Melman, Joseph W. Nimeh, and Ran D. Anbar, ?Prevalence of Elevated Blood Lead Levels in
an Inner-City Pediatric Clinic Population,? Environmental Health Perspectives 106, no. 10 (October 1998):
655?57.
22. Pamela A. Meyer and others, ?Centers for Disease Control and Prevention Surveillance for Elevated Blood
Lead Levels among Children: United States, 1997?2001,? Morbidity and Mortality Weekly Reports Surveillance
Summary 52, no. 10 (September 2003): 1?21.
23. Stuart J. Pocock, Marjorie A. Smith, and Peter A. Baghurst, ?Environmental Lead and Children?s Intelligence:
A Systematic Review of the Epidemiological Evidence,? British Medical Journal 309, no. 6963 (November
5, 1994): 1189?97; Richard L. Canfield and others, ?Low-Level Lead Exposure, Executive Functioning,
and Learning in Early Childhood,? Neuropsychology, Development, and Cognition, Section C Child
H e a l t h D i s p a r i t i e s a n d G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 135
Neuropsychology 9, no. 1 (March 2003): 35?53; Herbert S. Needleman and others, ?Bone Lead Levels in
Adjudicated Delinquents: A Case Control Study,? Neurotoxicology and Teratology 24, no. 6 (November?
December 2002): 711?17.
24. The prevalence of high lead exposure is 8.7 percent among blacks and 2 percent among whites. If high lead
exposure were responsible for a five point decline in IQ scores, and this decline translated into roughly a
third of a standard deviation fall in our generic test score, then we could make the following calculation: the
mean score for blacks would be [(.91*50) + (.9*48)] = 49.8, while the mean score for whites would be
[(.98*50) + (.2*48)] ~ 50.0.
25. Janet Currie, ?U.S. Food and Nutrition Programs,? Means-Tested Transfer Programs in the United States,
edited by Robert Moffitt (University of Chicago Press for NBER, 2003); Jayanta Bhattacharya, Janet Currie,
and Stephen Haider, ?Food Insecurity or Poverty? Measuring Need-Related Dietary Adequacy,? Working
Paper 9003 (Cambridge, Mass.: National Bureau of Economic Research, June 2002).
26. Anne C. Looker and others, ?Prevalence of Iron Deficiency in the United States,? Journal of the American
Medical Association 277, no. 12 (March 26, 1997): 973; Anne C. Looker, Mary E. Cogswell, and Elaine W.
Gunter, ?Iron Deficiency?United States, 1999?2000,? Morbidity and Mortality Weekly Report 51(40) (October
11, 2002): 897?99; Bettylou Sherry, Zuguo Mei, and Ray Yip, ?Continuation of the Decline in Prevalence
of Anemia in Low-Income Infants and Children in Five States,? Pediatrics 107, no. 4 (April 2001):
677?82.
27. Sally Grantham-McGregor and Cornelius Ani, ?A Review of Studies on the Effect of Iron Deficiency on
Cognitive Development in Children,? Journal of Nutrition 131, no. 2S-2 (February 2001): 649S?66S.
28. Ruowei Li and others, ?Prevalence of Breastfeeding in the United States: The 2001 National Immunization
Survey,? Pediatrics 111, no. 5 Supplement (May 2003); Renata Forste, Jessica Weiss, and Emily Lippincott,
?The Decision to Breastfeed in the United States: Does Race Matter?? Pediatrics 108, no. 2 (August, 2001):
291?96.
29. Jacqueline H. Wolf, ?Low Breastfeeding Rates and Public Health in the United States,? American Journal of
Public Health 93, no. 12 (December 2003); Daniel L. Drane and Jeri A. Logemann, ?A Critical Evaluation
of the Evidence on the Association between Type of Infant Feeding and Cognitive Development,? Pediatric
Perinatal Epidemiology 14, no. 4 (October 2000): 349?56; Anjali Jain, John Concato, and John M. Leventhal,
?How Good Is the Evidence Linking Breastfeeding and Intelligence?? Pediatrics 109, no. 6 (June
2002): 1044?53.
30. The average score for white infants would be (.29*50) + (.71*45) = 46.45 and the average score for black infants
would be (.09*50) + (.91*45) = 45.45.
31. Martha L. Bruce, David T. Takeuchi, and Philip J. Leaf, ?Poverty and Psychiatric Status,? Archives of General
Psychiatry 48 (1991): 470?74; Stevan E. Hobfell and others, ?Depression Prevalence and Incidence
among Inner-City Pregnant and Postpartum Women,? Journal of Consulting and Clinical Psychology 63, no.
3 (1995): 445?53; Fong-ruey Liaw and Jeanne Brooks-Gunn, ?Cumulative Familial Risks and Low-Birthweight
Children?s Cognitive and Behavioral Development,? Journal of Clinical Child Psychology 23 (1994):
360?72.
32. Dan L. Tweed and others, ?Racial Congruity as a Contextual Correlate of Mental Disorder,? American Journal
of Orthopsychiatry 60 (1990): 392?402; Ronald Kessler and others, ?Lifetime and 12-Month Prevalence
J a n e t C u r r i e
136 THE FUTURE OF CHILDREN
of DSM-III-R Psychiatric Disorders in the United States,? Archives of General Psychiatry 51 (1994): 8?19;
Hobfell and others, ?Depression Prevalence and Incidence among Inner-City Pregnant and Postpartum
Women? (see note 31); Bruce L. Rollman and others, ?Race, Quality of Depression Care, and Recovery
from Major Depression in a Primary Care Setting,? General Hospital Psychiatry 24 (2002): 381?90.
33. Carolyn Zahn-Waxler and others, ?Antecedents of Problem Behaviors in Children of Depressed Mothers,?
Development and Psychopathology 2 (1990): 271?91; Grazyna Kochanska and others, ?Resolution of Control
Episodes between Well and Affectively Ill Mothers and Their Young Child,? Journal of Abnormal Child
Psychology 15 (1987): 441?56; Stephen M. Petterson and Alison B. Albers, ?Effects of Poverty and Maternal
Depression on Early Child Development,? Child Development 72, no. 6 (November?December 2001):
1794?813; Cheryl T. Beck, ?Maternal Depression and Child Behavior Problems: A Meta-Analysis,? Journal
of Advanced Nursing 29, no. 3 (1999): 623?29; Carla Martins and Elizabeth A. Gaffan, ?Effects of Early Maternal
Depression on Patterns of Infant-Mother Attachment: A Meta-Analytic Investigation,? Journal of
Child Psychology and Psychiatry 41, no. 6 (2000): 737?46; Stephen Cogill and others, ?Impact of Postnatal
Depression on Cognitive Development in Young Children,? British Medical Journal 292 (1986): 1165?67;
Lynne Murray and others, ?The Impact of Postnatal Depression and Associated Adversity on Early Mother-
Infant Interactions and Later Infant Outcomes,? Child Development 67 (1996): 2512?26.
34. Deborah Sharp and others, ?The Impact of Postnatal Depression on Boys? Intellectual Development,? Journal
of Child Psychology and Psychiatry 36 (1995): 1315?37; Murray and others, ?The Impact of Postnatal
Depression and Associated Adversity on Early Mother-Infant Interactions and Later Infant Outcomes? (see
note 37); Sophie Kurstjens and Dieter Wolke, ?Effects of Maternal Depression on Cognitive Development
of Children over the First 7 Years of Life,? Journal of Child Psychology and Psychiatry 42, no. 5 (2001):
623?36; Petterson and Albers, ?Effects of Poverty and Maternal Depression on Early Child Development?
(see note 33).
35. The average score for white children would be (.03*45) + (.97*50) = 49.9 and the average score for black
children would be (.11*45) + (.89*50) = 49.45.
36. Eugene Lewit, Courtney Bennett, and Richard Behrman, ?Health Insurance for Children: Analysis and
Recommendations,? The Future of Children 13, no 1 (Spring 2003): 1?4.
37. Karen Kuhlthau and others, ?Correlates of Use of Specialty Care,? Pediatrics 113, no. 3, part 1 (March
2004): e249-55.
38. Janet Currie and Jeffrey Grogger, ?Medicaid Expansions and Welfare Contractions: Offsetting Effects on
Prenatal Care and Infant Health,? Journal of Health Economics 21 (March 2002): 313?35; Anna Aizer, ?Low
Take-up in Medicaid: Does Outreach Matter and for Whom?? American Economic Review, Papers and Proceedings
(May 2003): pp. 238?41; Janet Currie, ?The Take-up of Social Benefits,? Working Paper 10488
(Cambridge, Mass.: National Bureau of Economic Research, May 2004).
39. Anna Dixon and others, ?Is the NHS Equitable? A Review of the Evidence,? Health and Social Care Discussion
Paper 11 (London School of Economics, 2003); Lori J. Curtis and others, ?The Role of Permanent Income
and Family Structure in the Determination of Child Health in Canada,? Health Economics 10 (4)
(June 2001): 287?302.
40. Amy M. Heneghan and others, ?Do Pediatricians Recognize Mothers with Depressive Symptoms,? Pediatrics
106, no. 6 (December 2000): 1367?73.
H e a l t h D i s p a r i t i e s a n d G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 137
41. Janet Currie, ?Early Childhood Intervention Programs: What Do We Know?? Journal of Economic Perspectives
15, no. 2 (Spring 2001): 213?38.
42. Linda Fosburg and others, ?The Effects of Head Start Health Services: Report of the Head Start Health
Evaluation,? AAI 84-13 (Cambridge, Mass.: Abt Associates Inc., March 15, 1984).
43. Janet Currie and Matthew Neidell, ?Getting Inside the ?Black Box? of Head Start Quality: What Matters and
What Doesn?t,? Working Paper 10091 (Cambridge, Mass.: National Bureau of Economic Research, November
2003).
44. David L. Olds and others, ?Prenatal and Infancy Home Visitation by Nurses: Recent Findings,? The Future
of Children 9, no. 1 (Spring/Summer 1999): 44?65.
45. Ray Yip and others, ?Declining Prevalence of Anemia among Low-Income Children in the United States,?
Pediatrics 258, no. 12 (1987): 1619?23; Lori Kowaleski-Jones and Greg Duncan, ?Effects of Participation in
the WIC Food Assistance Program on Children?s Health and Development: Evidence from NLSY Children,?
Discussion Paper 1207-00 (Madison, Wis.: Institute for Research on Poverty, 2000). For an extensive
review of the WIC literature, see Janet Currie, ?U.S. Food and Nutrition Programs,? in Means Tested Transfer
Programs in the United States, edited by Robert Moffitt (University of Chicago Press for NBER, 2003).
46. Marianne Bitler, Janet Currie, and John Karl Scholz, ?WIC Eligibility and Participation,? Journal of Human
Resources 38 (2003): 1139?79.
J a n e t C u r r i e
138 THE FUTURE OF CHILDREN
The Contribution of Parenting to Ethnic
and Racial Gaps in School Readiness
Jeanne Brooks-Gunn and Lisa B. Markman
Summary
The authors describe various parenting behaviors, such as nurturance, discipline, teaching, and
language use, and explain how researchers measure them. They note racial and ethnic variations
in several behaviors. Most striking are differences in language use. Black and Hispanic
mothers talk less with their young children than do white mothers and are less likely to read to
them daily. They also note some differences in harshness.
When researchers measuring school readiness gaps control for parenting differences, the racial
and ethnic gaps narrow by 25?50 percent. And it is possible to alter parenting behavior to improve
readiness. The authors examine programs that serve poor families?and thus disproportionately
serve minority families?and find that home- and center-based programs with a parenting
component improve parental nurturance and discipline. Programs that target families
with children with behavior problems improve parents? skills in dealing with such children. And
certain family literacy programs improve parents? skills in talking with their children. Several
interventions have significantly reduced gaps in the parenting behavior of black and white
mothers.
Not all improvements in parenting translate to improved school readiness. Home-based programs
affect the mother but do not appear to affect the child, at least in the short term. But
center-based programs with a parenting component enhance both parenting and school readiness.
And some family literacy programs also improve readiness.
Because these successful interventions serve a greater share of minority than nonminority families
and have more positive effects for blacks than for whites, they offer promise for closing the
ethnic and racial gaps in school readiness.
VOL. 15 / NO. 1 / SPRING 2005 139
www.future of children.org
Jeanne Brooks-Gunn is Virginia and Leonard Marx Professor of Child Development and Education at Teachers College and the College of
Physicians and Surgeons, Columbia University. Lisa B. Markman is associate director of the Education Research Section, Woodrow Wilson
School of Public and International Affairs, Princeton University. They thank Eleanor Maccoby, the NICHD Research Network on Child and
Family Wellbeing, and graduate fellows at the National Center for Children and Families at Teachers College, Columbia University.
Everyone knows? that parenting
powerfully influences a child?s
well-being. And volumes of research
confirm that intuitive
link.1 Could parenting behavior
also play a role in the ethnic and racial
gaps in school readiness found by social scientists?
Just as there are stark differences in
the economic, educational, and demographic
conditions in the homes of white children
and of black and Hispanic children, as other
articles in this issue report, there may also be
racial and ethnic variation in parenting behaviors.
If so, such differences may contribute
to the gaps in achievement and readiness
that show up when children reach
elementary school. To explore these possibilities,
we first describe parenting behaviors, as
well as the ways in which researchers often
assess parenting. Then, we ask to what extent
parents matter. That question may come as a
surprise, because parenting is so universally
regarded as important. But social scientists
have raised questions about the extent to
which parents matter (does their behavior
matter at all, and if so does it matter a little or
a lot?), and we pursue them here.
Next we turn our attention to possible racial
and ethnic differences in parenting behavior.
When we find ethnic or racial differences in
parenting?and we do?we provide examples
of how specific parenting behavior translates
into specific child behavior. We also consider
the issue of equivalence in parenting measures
across racial and ethnic groups. Then we
investigate possible programmatic approaches
to altering parenting behavior and
ask how effective parenting programs are. Finally,
we explore both how much parenting
programs can enhance the school readiness
of minority children and how much they can
close the ethnic and racial gaps in school
readiness.
What Is Parenting?
Parenting encompasses the literally hundreds
of activities that parents engage in either with
or for their children. Often, researchers divide
parenting into categories of behavior. In
this article we use seven: nurturance, discipline,
teaching, language, monitoring, management,
and materials.2
Nurturance
Nurturing behavior involves ways of expressing
love, affection, and care. High nurturing
behaviors include expressing warmth, being
responsive to a child?s needs, and being sensitive
to changes in a child?s behavior. Low nurturing
behaviors include detachment, intrusiveness,
and negative regard.3
Researchers measure nurturance by observing
a parent interacting with her child (parents
are not particularly good or accurate reporters
of their own warmth, detachment, or
intrusiveness). They observe naturally occurring
interactions during a two- or three-hour
home visit or during structured tasks that can
be set up at home, at a preschool center, or at
a pediatric clinic. For home visits, the Home
Observation for the Measurement of the Environment
(HOME) Inventory, for example,
asks the observer to record whether she saw
certain behaviors, such as a parent spontaneously
praising a child?s qualities twice; caressing,
kissing, or cuddling a child; or using a
term of endearment.4 The structured tasks
range from free play with a specific set of
toys to problem solving with unique materials
(for example, getting a toy from a box using a
rake or another utensil) to copying a puzzle
or design. Often, researchers videotape the
interactions so that they can code them later.
Sometimes they code very detailed behaviors
(marking the presence or absence of up to
fifteen parent and child behaviors every five
to ten seconds); other systems involve more
Jeanne Brooks-Gunn and Lisa B. Markman
140 THE FUTURE OF CHILDREN
global coding of a number of constructs, such
as sensitivity to a child?s cues, expressed
warmth, intrusiveness, and detachment.
Training of coders is intensive (often for as
long as six weeks) to ensure their reliability.
In semistructured videotaped free play sessions,
the observer gives parents and young
children toys to play with, leaving instructions
deliberately vague. In several studies,
she places three toys in separate bags, so that
the mother uses one toy at a time, and observes
the parent and child for ten to fifteen
minutes.5 The observer rates the session after
repeatedly viewing videotapes of behaviors,
including detachment (low involvement with
and lack of attention to the child), intrusiveness
(over-control and over-involvement in
the child?s play), negative regard (anger, rejection),
sensitivity (extent to which the parent
perceives the child?s signals and responds
appropriately), and positive regard (demonstration
of love, respect, and admiration).6
Sometimes, these behaviors are treated separately,
because they measure different aspects
of nurturance. At other times, they are
clustered together to identify different
groups of parents. For example, we have
identified several groups of parents?we
term them sensitive, directive, uninvolved,
and harsh?based on the coding of behaviors
in the three-bag free play.7
Discipline
Discipline involves parents? responses to
child behaviors that they consider appropriate
or inappropriate, depending on the child?s
age and gender and on parental beliefs, upbringing,
and culture.8 Observers sometimes
measure discipline from what they see during
the course of a home visit. They would describe
discipline as harsh or punitive if the
parent spanked, slapped, or yelled at the
child during the visit.9 Because parents may
be less likely to spank a child with an observer
in the home, observers often ask parents
about frequency of spanking. They also
ask about their use of other discipline strategies,
such as time out, explanations, and taking
away toys or food. In a few studies, they
give parents a scenario. For example, they
ask what a mother would do if her child had a
temper tantrum in the market; or, if her child
had had a tantrum, what she did in response.
Sometimes they calculate a severity-ofpunishment
score or a use-of-reason score.10
Teaching
Teaching typically includes didactic strategies
for conveying information or skills to the
child. Observers set up interaction situations
such as putting together a puzzle that is
slightly difficult for the child; drawing a complex
figure; learning a skill such as tying a
shoe or buttoning a coat; or sorting building
blocks by shape or color, and then observe
teaching behaviors. Often, they rate the
strategies in terms of quality of assistance.
For example, when helping her child with a
puzzle, a mother might do any of the following:
take over and put most of the pieces in
the puzzle; wait until the child runs into difficulty
and then take over; not assist the child
at all; provide cues or prompts (?What would
happen if you turned that puzzle piece
around??) to help the child find the right
place for a piece; provide an overall strategy
(?Can you find all the pieces that go on the
edges of the puzzle??). Observers would code
the latter two examples as high in quality of
assistance.11
The HOME Inventory includes items related
to teaching?does the parent encourage the
child to learn colors, songs, or numbers or to
read a few words?that can be used to create a
scale called Provision of Learning or Learning
T h e C o n t r i b u t i o n o f P a r e n t i n g t o E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 141
Stimulation.12 These reports are based on
parental report, rather than direct observation.
Language
Researchers have extensively studied language
use between parents and young children.
The most comprehensive studies have
transcribed hundreds of hours of motherchild
conversations.13 From those transcriptions,
observers can code the sheer amount
of language heard by and directed to the
child, as well as the number of different
words, length of sentences, questions asked,
elaborations on the child?s speech, and events
discussed. Observers also frequently elicit
parent language by having parents read to
their children.14 Parents vary in how often
they ask the child questions, expand on what
is in the story, and see whether the child understands
the meaning of a word.15 They also
vary in how much they engage in what
Katherine Snow has called nonimmediate
talk, or going beyond the information given
in the story, and in their style of reading.16
The HOME Inventory includes several items
indicative of reading: child has access to at
least ten children?s books; at least ten books
are visible in the home; family buys and reads
a daily newspaper; child has three or more
books of his or her own.17 These items are
tapping something different from frequency
of book reading or style of reading as measured
through direct observation. The underlying
premise is that children who are exposed
to more reading materials live in
households where reading, both adult reading
and parent-child shared reading, is more
common.
Materials
The term materials refers to the cognitively
and linguistically stimulating materials provided
to the child in the home. This category
can overlap with language and with teaching.
For example, some scholars categorize number
of books in the home, number of children?s
books, and number of magazine subscriptions
as materials rather than as
language because they do not know whether
parents use them to foster reading. Other
items included here are toys and books for
learning the alphabet and numbers, educational
toys, musical instruments, push-pull
toys, drawing materials, and the like. The extensiveness
of material items in the home is
associated with family income, which is not
surprising, given that most are purchased.18
Monitoring
Monitoring is what might be called ?keeping
track.? With young children, monitoring
refers to parental watchfulness. For example,
if a child is playing in a room alone, a parent
might periodically check to see what she is
doing or call out to her; if a child is watching
television alone, a parent might keep track of
what program he is watching and change the
channel if it seems inappropriate. Studies
using time-use diaries of children?s days try to
distinguish between time when the parent is
directly interacting with the child and time
when the parent is in the home and responsible
for the child even though the two are
Jeanne Brooks-Gunn and Lisa B. Markman
142 THE FUTURE OF CHILDREN
When parents read to their
children, they vary in how
often they ask the child
questions, expand on what is
in the story, and see whether
the child understands the
meaning of a word.
doing different things.19 Occasionally the distinction
is difficult to make, such as when a
child is watching television and the mother is
in the room, sometimes watching and talking
about a program with the child and sometimes
doing housework. With older children,
monitoring involves knowing what the child
is doing and with whom he is doing it when
he is outside the home.
Management
Management encompasses scheduling events,
completing scheduled events, and the rhythm
of the household. Most studies of young children
either do not measure management at
all or assess it with only one or two short
questions, even though management tasks
consume huge amounts of parenting time.
Most national studies do ask about two
health-related areas: getting the recommended
number of well-child visits and getting
immunizations on time. Sometimes studies
note the appearance of the child (dirty,
not dressed, clothes do not fit) as a possible
indicator of child neglect. Studies do not always
assess taking children to scheduled activities
outside the home (even though time
diary studies suggest that fathers spend the
greatest proportion of their weekend time
with their preschoolers in such activities), but
often do assess taking children to the park
and to visit relatives.20
Researchers sometimes tap the rhythm of the
household, typically through questions about
the regularity of bedtime, bedtime routines
(reading, singing, praying), how many meals
the family eats together, the breakfast routine
(whether breakfast is eaten at all, whether
the television is on).21
Does Parenting Matter?
Despite all the studies reporting links between
parenting and child well-being, we still
need to question whether parenting matters.
22 Our premise is as follows. Even
though the literature is voluminous, it also
has its limits, all of which comes down to the
same problem: we do not know, in most
cases, whether the so-called effects of parenting
are caused by parental behavior or by
something else that may complicate the
causal link. We consider four different factors:
family social, educational, and economic
conditions; genetic similarities between parent
and child; child characteristics; and other
unmeasured characteristics (which we believe
might be operating but have not measured,
or do not know how to measure well).
Although all four factors influence links between
parenting and child well-being, they
do not account completely for these links.
(Another line of evidence supporting the
premise that parents matter, reviewed later in
this article, has to do with the potential of
intervention programs to alter parenting.)
Parenting and Correlated Family
Conditions
First, we know that parents differ in their social,
economic, and educational backgrounds.
And we know that variations in parenting are
associated with such characteristics. The link
between parental talking and child vocabulary
is one example.23 Parents who talk a lot
to their children, ask questions, use many different
words, and discuss events are also
more likely to be highly educated, to have
high incomes, and to have few children, as
well as to have children with large vocabularies.
And these latter characteristics are themselves
associated with child vocabulary. Thus
in reality parental education might account
for the link between parental talking and
child language. If parents who talk a lot are
more likely to be highly educated, we need to
adjust for parental education to be sure the
link between parental language use and child
T h e C o n t r i b u t i o n o f P a r e n t i n g t o E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 143
vocabulary is not inflated. It is relatively easy
to measure parental education and make statistical
adjustments to see if the link between
parental talking and child vocabulary still exists,
just as it is for other characteristics like
family structure, income, parity (number of
children), age, and the like. Studies that
make such adjustments find that the link exists
independent of parental education.
At the same time, the purpose of such studies
is often to show how parental education, for
example, influences children?s language.
Clearly, in that case, the parental education
effect would have to translate into a specific
parenting behavior, such as talking to the
child. So we often consider parenting (in this
case parental talking) to be a pathway
through which parental education influences
child language. That suggests two types of intervention
strategies. One is indirect: to try to
increase maternal education in the hope that
more schooling would cause a mother to talk
more to her child. The other is direct: to try
to increase her talking with the child. The latter
would target behavior directed toward the
child (talking), rather than a more general
characteristic of the adult (more education).
The assumption is that it is possible to pinpoint
the specific parenting behaviors that
contribute to a specific child outcome. High
levels of parental warmth, in the absence of
much parental talking, for example, would
not be expected to increase child vocabulary.
Neither would parental monitoring, unless it
involved lots of talking.
Parenting and Correlated Genetic
Characteristics
Second, perhaps the most widely heralded
causal issue is that parents and children are
genetically related, which can, in part, account
for links between parenting and child
well-being. To continue with our earlier example,
parents who talk a lot and have a large
vocabulary are likely to have children who
are predisposed toward language. That is,
language facility is partly heritable.24 Even in
the absence of parenting behavior, parent
language test scores would be linked with
child language scores.25 How can we tell to
what extent the link is due to environment
(here, language expressed to the child) and to
what extent to genetics (here, the biological
relationship between parent and child)?
Studies informed by behavioral genetics are
useful here.26 Two examples, one from studies
of adopted children and the other from
work with identical twins (monozygotic twins,
whose genetic material is identical, so that
any differences between them must be environmental),
demonstrate that parenting influences
child well-being, over and above genetic
relatedness of parents and children.
Studies of adopted children show striking increases
in cognitive abilities when the children
leave institutional care to be placed with
adoptive parents.27 Children in such studies,
however, move from extremely deprived environments
without consistent caregivers (orphanages)
to stable, two-parent, largely middle-
class homes. The studies speak to the
powerful effect of having parents versus not
having parents, but say little about the effects
of varying levels of parenting behavior.
One study does address normal variation in
parenting. Michel Duyme and colleagues
identified a small sample of adopted children
(fewer than seventy) from a review of more
than 5,000 adoption cases in France.28 They
selected all children between the ages of four
and six who had been placed in prescreened
adoptive homes, removed from their birth
parents because of abuse or neglect, and put
in foster care before their adoption. The children
were given cognitive tests before their
Jeanne Brooks-Gunn and Lisa B. Markman
144 THE FUTURE OF CHILDREN
adoption and again between the ages of
eleven and eighteen. Overall, the children
showed striking gains in IQ test scores from
early childhood to adolescence, from a mean
score of 77 to 91 (14 points or almost one
standard deviation on a test with a mean of
100 and a standard deviation of 15). The authors
classified the adoptive households as
low, middle, or high socioeconomic status
(SES), based on paternal occupation. The
gains were largest for those placed in high
SES families (19 points) and smallest for
those in the low SES families (8 points).29
The assumption is that the high SES families
were providing more language, more teaching,
and more materials, all of which facilitated
the children?s cognitive growth.
A study of children exposed to cocaine prenatally
also illustrates the power of change in
parenting.30 The study recruited more than
400 mothers following delivery. All the new
mothers were considered at high risk for cocaine
use; about half had biological indications
of cocaine use when they and their infants
were tested at delivery. When the
children were four years old, researchers
gave them an IQ test, observed them in their
homes, and gave their caregivers a vocabulary
test. In the group of children who had
been exposed to cocaine before birth, only 55
percent were living with their biological
mothers at the follow-up, as against 95 percent
of those in the group that had not been
exposed. The cocaine-exposed preschoolers
living with their mothers or with a relative
had significantly lower IQ scores than their
counterparts who were living with an adoptive
or foster care mother?even though (not
surprisingly) the latter group had been exposed
to more cocaine than those who were
not removed from the mother. Furthermore,
the IQ scores of the exposed children living
with an adoptive or foster mother were comparable
to those of the children who were not
exposed to cocaine prenatally. For example,
the share of the cocaine-exposed children living
with their mothers who had IQ scores
under 70 (the mild mental retardation range)
was 25 percent, as against 10 percent for the
exposed children who lived with nonrelatives
and 16 percent for the nonexposed children.
As might be expected, the homes of the
groups differed; cocaine-exposed children
living with adoptive or foster mothers had
more stimulating environments, and their
mothers had higher vocabulary scores, than
the cocaine-exposed children living with their
biological mothers or relatives.31
The second class of studies does not rely on
change in parents (from orphanage to family
or from biological to adoptive parent). Instead,
it uses genetic similarity to delve into
parental effects. In a sample of 500 five-yearold
identical twins, mothers were asked to
talk about each of their children. Mothers
tended to describe one twin in more negative
terms than the other. When the children
were in elementary school, their teachers
were asked to rate their behavior.32 Teachers
reported that the twin for whom the mother
had more negative feelings had more behavior
problems than the other twin.33 Because
the children had identical genetic endowments,
it is highly likely that maternal behavior
accounted for the differences in behavior
problems between the twins.
Parenting and Correlated Child
Characteristics
A third causal issue is that parenting behavior
may be in part contingent on the behavior of
the child. That is, not only does parenting affect
child behavior, but also children can influence
parents.34 We provide two examples,
the first having to do with reading, the second
with behavior problems.
T h e C o n t r i b u t i o n o f P a r e n t i n g t o E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 145
Children of mothers who read to them frequently
have large vocabularies, as countless
studies have shown.35 In an evaluation of the
Early Head Start Program, Helen Raikes and
her colleagues have found the expected links
between shared book reading and child vocabulary
in more than 1,000 children seen at
age fourteen months, twenty-four months,
and thirty-six months, even after adjusting for
differences in mothers? verbal abilities.36
(The adjustment is necessary because mothers
with higher verbal abilities are likely to
enjoy reading more than other mothers,
which could influence their shared book
reading with the child, and because language
ability is partly heritable.) Of more interest is
their exploration of the pathways through
which language at age thirty-six months was
influenced. More shared reading at fourteen
months was linked with higher vocabulary
scores at twenty-four months, which affected
the amount of reading at twenty-four and
thirty-six months. Thus, mothers whose children
knew and used more words were reading
more to these children as they developed,
over and above their reading levels at fourteen
months.
One of the best-known examples of child-toparent
effects is an intervention geared to-
Jeanne Brooks-Gunn and Lisa B. Markman
146 THE FUTURE OF CHILDREN
ward children with conduct disorders and
their parents.37 Half the children participated
in a family program, which was effective in
that the children displayed less aggression
after the intervention. But the positive impact
on the children was primarily due to
changes in parenting behavior. That is, the
parents in the intervention group stopped reacting
negatively to their children?s aggressive
behavior by learning other techniques
for dealing with outbursts. In contrast, the
parents in the control group did not alter
their responses to their children?s outbursts,
and therefore the children?s problem behavior
showed no change.
The point here is that child characteristics
can influence parenting. But the existence of
differences among children themselves does
not totally account for parenting effects on
children.
Parenting and Unmeasured Correlates
The final complicating causal issue involves
possible correlates of parenting that have not
been measured. Even studies that adjust for
family conditions and child characteristics
may fail to measure other sources of variation
in parenting and children?s school readiness,
perhaps because of limits of cost or time or
the lack of a reliable indicator.
One characteristic often associated with parenting
and child outcomes is parents? mental
health. Mothers who are diagnosed with clinical
depression or as having high levels of depressive
symptoms engage in less nurturance
and more punitive discipline, as has been
demonstrated countless times for preschoolers
as well as older children.38 And these mothers?
preschool children have more behavioral problems
and (sometimes but not always) lower
cognitive test scores.39 But even when analysts
adjust for maternal depression, parenting still
More shared reading at
fourteen months was linked
with higher vocabulary
scores at twenty-four months,
which affected the amount
of reading at twenty-four
and thirty-six months.
contributes to these indicators of child wellbeing.
Indeed, maternal depression, as well as
other measures of mental health (anxiety, irritability),
is thought to act on children through
its effect on parenting behavior.40 Vonnie
McLoyd, Rand Conger, and their colleagues
have proposed a family stress model that traces
the pathways from low income, financial instability,
and material stress through parental
mental health to parenting to child outcomes.
41 Jean Yeung and her colleagues have
shown that this pathway is stronger for behavioral
problems than for cognitive and language
test scores in young children.42
Even if a study measures many potential correlates,
it is impossible to be sure that it includes
all that are relevant. So scholars use a
variety of statistical techniques to minimize
the likelihood that results are due to something
besides parenting.43 But the most convincing
evidence is gleaned from experiments
where families enter a treatment or a
control group through random assignment.
We present evidence from experiments designed
to test the efficacy of parenting programs
later in the article.
Do Ethnic and Racial Differences
in Parenting Exist?
In this section we first ask whether measures
of parenting are equivalent across ethnic and
racial gaps. Next we consider whether there
are ethnic and racial differences in the seven
dimensions of parenting described earlier
And, finding some, we compare their size
with that of the ethnic and racial gaps in
school readiness. For several domains of parenting,
we find the sizes are similar. Using
evidence of congruence in the strength and
direction of links between parenting and
school readiness for black, white, and Hispanic
children, we ask whether the meaning
of parenting behavior varies from one ethnic
or racial group to another. Although the dimensions
of parenting seem to be equivalent
across groups, the levels of particular behaviors
do, in some instances, vary. At the same
time, there are more similarities than differences
in links between children?s school
readiness and parenting across racial and ethnic
groups; when differences appear, they
seem to be clustered in negative parenting
behaviors.
Equivalence of Parenting Measures
across Ethnic and Racial Groups
Any discussion of parenting gives rise to arguments
about whether parenting behaviors
are the same from one group to another and
whether measures of parenting have the
same meaning from one group to another.
Three considerations are relevant: first,
whether parenting behaviors are universal or
specific to time and place; second, how representative
the parenting behaviors typically
measured and developed using middle-class
white samples are of other groups; and third,
whether a particular society ?privileges? certain
parenting behaviors.
Regarding the first point, many aspects of parenting
described in this article are exhibited
by parents in many societies.44 That is, all parents
have ways of nurturing, teaching, disciplining,
monitoring, and managing their young
children. All provide a linguistic environment
as well as a material environment. But the expression
of these parenting activities sometimes
differs, and the emphasis among behaviors
sometimes varies. In eastern Africa, for
example, parents devote much time to working
with and encouraging their toddlers to develop
their motor skills. Not surprisingly, their
children?s motor skills are more advanced than
those of U.S. children.45 Parents in Western
societies often value language and vocabulary
skills (given their links to doing well in school),
T h e C o n t r i b u t i o n o f P a r e n t i n g t o E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 147
so the language output of Western children is
often greater than that of children in other societies.
46 The point, however, is that parents
across societies engage in teaching activities
(as in the case of motor skills) and language activities
(as in the case of vocabulary). The difference
is in the level of a particular behavior,
not its existence.47
On the second point, the parenting behaviors
measured in most studies are said to be representative
of middle-class families in the
United States.48 We agree with this proposition,
given the samples from which most parenting
measures were derived. Consequently,
some parenting behaviors are probably not
measured, or not measured well. And these
may be behaviors that are more prevalent in
black and Hispanic groups than in white
groups. For example, some groups, such as recent
Hispanic immigrants, may value compliance
of toddlers more than do other groups.49
Are we measuring compliance, and the
parental behaviors that foster it, accurately?
Another example of imperfect measurement
of parenting surfaced from our research
group?s work with a widely used coding
scheme developed by Diana Baumrind,
which distinguishes between authoritative
parenting (warm, firm control) and authoritarian
parenting (negative, harsh control).50
Studies have found that black mothers are
more authoritarian and less authoritative
than white mothers, just as lower SES mothers
are more authoritarian than higher SES
mothers.51 However, black graduate students
in our laboratory felt that these codes did not
represent what they had seen in black families.
So we did an exploratory analysis using a
sample of about 700 black and white mothers
of toddlers, attempting to identify clusters of
mothers based on our videotaped ratings on
both domains. We identified not two but four
groups of mothers?those who were high in
warm, firm control and low in negative, harsh
control (the classic authoritative behavior);
those who were high in negative, harsh control
and low in warm, firm control (the classic
authoritarian behavior); those who were relatively
high in both (what we termed ?tough
love?); and those who were low in both (what
we termed ?detached?). More blacks than
whites were in the tough love group. The
classic authoritarian group was composed
primarily of teenage mothers, both black and
white, while the tough love group comprised
mostly older black mothers with at least a
high school education. Interestingly, children
of mothers in the tough love group had
higher IQ and vocabulary scores than children
in the classic authoritarian or the detached
group, suggesting that previous coding
schemes had confounded two groups of
black mothers by labeling them authoritarian?
and assuming that their parenting had
negative consequences for school readiness.
52 A further example of how difficult it
can be to measure parenting relates to findings
that spanking has less negative consequences
for black than white children.53
Spanking may be more normative for the
black children, and it may occur in the context
of warm parenting?that is, tough love.
As to the third point, perhaps the best evidence
of the validity of a particular parenting
behavior is how well it predicts school readiness.
And given our focus on racial and ethnic
differences, whether parenting predicts
school readiness equally well in different
groups is critical. In general, the parenting
behaviors described in this article are related
to school readiness in U.S. society at this time.
They do not necessarily represent all parenting
behaviors, or particular behaviors valued
by certain groups, or behaviors that promote
outcomes other than school readiness. In this
Jeanne Brooks-Gunn and Lisa B. Markman
148 THE FUTURE OF CHILDREN
one-fifth of a standard deviation or less, or 3
points or less, using our reference test.55 Hispanic
mothers have scores comparable to
whites in most cases.56
Another positive indicator of nurturance is
the sensitivity of the mother, as expressed in
mother-child free play or problem-solving situations.
Black mothers are rated as having
somewhat lower levels of sensitivity?about
one-fifth of a standard deviation?as coded
from fifteen-minute videotaped sessions.57
Measures on the more negative end of the
nurturance continuum are also gleaned from
mother-child interchanges recorded on the
videotapes, which have documented racial
differences in negative regard, intrusiveness,
and detachment, with black mothers scoring
slightly higher than white mothers. The blackwhite
differences are around one-fifth to twofifths
of a standard deviation (3 to 6 points).58
Discipline also varies by racial and ethnic
group. Black mothers are somewhat more
likely to spank their children than are white
mothers.59 White mothers are more likely to
use reasoning as a discipline technique,
though the effects are modest, about one-fifth
or less of a standard deviation (1 to 3 points).60
Perhaps the most striking differences are for
language.61 Transcriptions of naturally occur-
T h e C o n t r i b u t i o n o f P a r e n t i n g t o E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 149
sense, we are privileging more Western, middle-
class parenting behaviors. If we are correct
that these are the behaviors that contribute
to school readiness, and if these are
the behaviors that parenting interventions target
because of their links to school readiness,
then this privileging seems appropriate. It
does not mean that these parenting behaviors
are ?good? while others are not.
Ethnic and Racial Differences
in Parenting
There are ethnic and racial differences in
parenting during early childhood. Evidence
is available on five of the seven parenting dimensions:
nurturance, discipline, teaching,
language, and materials. In all cases, when
differences occur, black mothers have lower
scores on parenting measures than do white
mothers. Similar differences often exist between
Hispanic and white mothers as well,
although the research base for this comparison
is much smaller. In general, the effect
sizes for the ethnic and racial differences
range from one-fifth to three-fifths of a standard
deviation?similar to but slightly
smaller than school readiness measures,
which are roughly two-fifths to four-fifths of a
standard deviation.54 These parenting differences
would translate into 3 to 9 points on a
test that had a mean of 100 and a standard
deviation of 15 (as many tests of vocabulary
and intelligence have). All references to test
points in the rest of the article refer to a test
with such characteristics. School readiness
measures on such a test show racial gaps of 6
to 12 points, depending on the aspect of
readiness being measured.
Evidence for racial and ethnic gaps in nurturance
comes from several sources. On the
first, the observational HOME Warmth
Scale, black mothers sometimes have lower
scores, although the differences are modest:
Another positive indicator of
nurturance is the sensitivity
of the mother, as expressed in
mother-child free play or
problem-solving situations
ring mother-child conversations suggest that
children?s exposure to language and conversation
varies widely across social class groups,
as demonstrated in a sample of forty-two
children from three different social class
groupings.62 As such differences accumulate
over the first years of life, the children in
families with a high socioeconomic background
have engaged in literally thousands
more conversations than children from lower
socioeconomic backgrounds. Even when they
begin speaking (around their first birthday),
higher SES children have larger vocabularies
than the children from middle and low SES
families. By their second birthday, the children
in the middle SES group have pulled
away from those in the low SES group. And
these differences accelerate over time. So by
age three, vocabularies of the children in the
low SES group are half the size of those in
the high SES group and two-thirds the size of
those in the middle SES group. Given the
racial composition of the SES groups in this
study (the majority of black families were in
the low SES group), black-white differences
were equally large.
Scholars have posited differences in family
?speech cultures,? which are associated, in
part, with social class and race.63 The educated
middle- to upper-middle-class ?speech
culture? provides more language, more varied
language, more language topics, more
questions, and more conversation, all of
which are linked with large vocabularies in
toddlers and preschoolers. Repeated and varied,
these parental speech patterns predict
how fast young children learn words.64 Little
research has focused on whether the variations,
if controlled, would reduce the racial or
ethnic gap in school readiness.65
Analysts have also examined shared book
reading as a vehicle for language input.66
Large national or multisite studies often ask
about the frequency of reading.67 From 40 to
55 percent of mothers report reading to their
toddler every day.68 Black mothers are about
two-thirds as likely as white mothers to do so;
Hispanic mothers, about half as likely.69 Ethnic
and racial differences in frequency of
reading exist in population-based as well as
low-income samples. Black and Hispanic
children also come from homes with fewer
reading materials (books, children?s books,
magazines, newspapers) than do white children.
70 The size of such differences is between
one-fifth and three-fifths of a standard
deviation.
Materials in the home also vary by ethnicity
and racial group. Not only do black and Hispanic
families have fewer reading materials
in their homes, but typically they also have
fewer educationally relevant materials of
other types (as indexed by the HOME Learning
Scale). Racial differences on the Learning
Scale are large, from two-fifths to three-fifths
of a standard deviation, or 6 to 9 points on
our reference test.71
Reduction in Racial Gaps in School
Readiness as a Function of Parenting
The racial differences in parenting do account
for a portion of the racial gap in school
readiness. In general, researchers who have
conducted such analyses report that a 12 to
15 point gap between black and white children
is reduced by 3 to 9 points when parenting
is considered.
Most national studies that follow a group of
the same children over time use the Learning
Scale as a measure of parenting.72 This particular
measure of parenting is often posited to
be one of the pathways through which parental
income, education, marital status, and age affect
children (just as language input and
Jeanne Brooks-Gunn and Lisa B. Markman
150 THE FUTURE OF CHILDREN
shared book reading are pathways through
which family social class influences school
readiness). Taking this measure into account
narrows the racial gap in such early childhood
outcomes by one-third to one-half.73
Do Parenting Interventions Work?
Is it possible to enhance parenting through
intervention programs? And if so, do some of
the beneficial effects on children of early
childhood intervention programs operate
through their effects on parenting? We consider
evidence for each question. In general,
programs focused on parenting can alter behavior,
as has been demonstrated in several
well-designed evaluations of experimental
programs (those in which families are randomly
assigned to treatment and control
groups). And some?but not all?of the benefits
that accrue to children seem to operate
through changes in parenting behavior.
Effects of Parenting Interventions
Interventions for parents of young children
fall into four categories: home-based (often
termed home-visiting) programs, centerbased
early childhood education programs
with a parenting component (often termed
center plus programs), family literacy programs,
and programs targeting child behavior
problems by changing parental behavior (the
latter are reviewed in a separate section).74
We focus on programs initiated in the first
four years of a child?s life.75
HOME-VISITING AND CENTER PLUS. Almost
all parenting programs target families in
which parents are poor, have little education,
are young, or are unwed. The programs are
not universal. Some have operated in multiple
sites (which assures that they can be
transferred to other settings and that staff can
be trained to deliver services and curriculum).
Overall, programs have served more
black and urban families than white, Hispanic,
or rural families, so we have more evidence
of program efficacy for the former
than the latter.
Program evaluations have focused mostly on
nurturance, discipline, language, and materials.
They have gathered little information
about teaching and virtually none about monitoring
and management (with the exception
of health practices, which are not reviewed
here).76 Several programs also target parental
mental health.77 Fewer programs have effects
on maternal depression than on nurturance,
language, and materials, suggesting that it
might be easier to alter parenting behavior
than parental emotional state, at least using
parenting interventions, rather than more focused
treatment of depressive symptoms.78
Nurturance has received much attention, because
one of the goals of many home- and center-
based programs with a parenting component
is to enhance sensitivity and reduce
negativity (the same is not true of family literacy
programs). Home-visiting programs are
more likely to affect nurturance than other
parenting behaviors. For example, eleven of
thirteen home-visiting evaluations that reportedly
observed mother-child interactions found
positive benefits.79 (One meta-analysis suggests
that home-visiting programs are better at
reducing parental insensitivity than at changing
other aspects of the mother-child attachment
relationship.)80 Center-based programs
with a parenting component, including Early
Head Start, also report enhancing sensitivity
and reducing negativity.81
Discipline has not been measured as frequently.
When it has, both home-based and
center-based programs with a parenting component
have shown decreases in spanking
and, in several cases, an increase in the use of
T h e C o n t r i b u t i o n o f P a r e n t i n g t o E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 151
reasoning.82 Again, this aspect of parenting is
not the focus of family literacy programs.
Teaching is often a part of intervention programs.
One curriculum, LearningGames, has
been used in the Infant Health and Development
Program, the Abecedarian Program,
and Project Care.83 The object is to present
age-appropriate activities for the child and
the parent to do together, and to provide the
parent role modeling and instruction in how
to approach them. Center-based programs
with a parenting component have reported
improving parents? ability to assist in problem-
solving activities.84 Much less is known
about home-visiting programs in this regard.
Home-based and center-based programs do
not often target maternal language, at least
not directly. We know almost nothing about
whether they increase maternal language
output. Because one determinant of a child?s
increased vocabulary is the mother?s vocabulary,
such a goal might be sensible.
A few literacy programs have tried to change
how parents read with their children, with an
implied goal of using more, and more varied
language. Grover Whitehurst and his colleagues
developed a program of dialogic
reading that trained mothers and teachers to
read with an emphasis on asking children
questions, providing feedback to their re-
Jeanne Brooks-Gunn and Lisa B. Markman
152 THE FUTURE OF CHILDREN
sponses, initiating conversations that went
beyond the book?s content, and delving into
children?s understanding of concepts.85 The
adult training was successful, and children in
the treatment group had higher language
scores than those in the control group. Several
programs with a focus on literacy are
now being evaluated.86
Many home-based and center-based programs
have used the HOME Learning Scale
to assess the parenting dimension that we call
materials. About half of the center-based programs
with a parenting component report
higher scores on this scale after treatment;
fewer home-based programs report such effects.
87 Even Start, a national literacy program,
reported its most consistent treatment
effect on reading materials in the home.88
In conclusion, home- and center-based programs
with a parenting component have their
largest and most consistent effects on nurturance.
They have some effects on discipline
and, in some instances, on materials. Little
evidence exists, for or against, regarding effects
on language. Indeed, language is most
likely to be changed by family literacy programs
that focus directly on shared book
reading and other language settings.
PARENT BEHAVIOR TRAINING PROGRAMS.
Yet another type of parenting program aims
to alter the behavior of parents whose children
exhibit problem behavior. Typically,
children who are disruptive and aggressive
and who act out in the preschool and early
school years are likely to have high rates of
delinquency and school drop-out during adolescence.
89 In the early school years, they are
likely to spend little time engaging in classroom
tasks and are often disliked by their
peers and teachers.90 To address these children?s
needs, researchers and clinicians have
Language is most likely to be
changed by family literacy
programs that focus directly
on shared book reading and
other language settings.
developed several types of programs, focusing
variously on parents, teachers, the child?s
social skills in the classroom, or individual
counseling.91
One parent training program, developed by
Carolyn Webster-Stratton and her colleagues,
crafts group discussions around videotaped
vignettes of typical discipline situations in the
home, often showing several ways to handle a
particular situation.92 This program has been
found to reduce parents? negative discipline
and nurturance behaviors and increase positive
parenting behaviors in mothers. Webster-
Stratton?s Incredible Years Curriculum is
often targeted to families in Head Start. As a
result, it benefits poor families. When the
parent program was expanded to include a
teacher component, it reduced negative behavior
and increased more supportive behavior
in parents, and it enabled teachers to use
more positive management techniques in
their Head Start classrooms. Children in the
program have lower rates of acting out and
aggressive behaviors and are more engaged
in their classrooms than are children in a control
group.93 Webster-Stratton?s programs
have effects on children of between one-half
and two-thirds of a standard deviation, or 7 to
10 points on our reference test.
A few other programs offer a range of services,
beginning with low-intensity services
for all parents in a classroom and moving to
more intensive services for parents whose
children have moderate behavior problems
and even more training and counseling for
families whose children have severe behavior
problems.94 Most of these programs, however,
have focused on kindergartners and first
graders.
Our point is that parent training programs for
children with moderate or severe behavior
problems have been proven successful. Programs
that include both parents and preschool
teachers seem to be the most successful
of all.
Parenting Impacts and Their Effects
on Children
Do the interventions? positive effects on parenting
make any difference in children?s cognitive
performance and school readiness?
Two types of evidence are relevant, the first
having to do with whether the programs
have effects on the children and the second
with whether any of the children?s benefits
are due to the effects of the programs on
parenting.
The answer to the first question depends on
the type of intervention. Few home-visiting
programs have altered children?s school
readiness.95 That being so, the positive parenting
effects for home-based programs
could not be translated into child effects. In
our view, most home-visiting programs are
not intensive enough, and home visitors are
not trained or supervised enough, to be likely
to enhance school readiness.96
In contrast, the center-based early childhood
education programs with a parenting component
have improved vocabulary, reading
achievement, math achievement, and IQ,
with some effects continuing through adolescence
in some studies.97 Although these programs
have few effects on socioemotional development
in preschool, two have lowered
juvenile delinquency and teenage pregnancy
rates.
Second, when programs affect both parents
and children, does the enhanced parenting
affect the child outcomes? This question is
important, especially for center-based programs
with a parenting component, because
T h e C o n t r i b u t i o n o f P a r e n t i n g t o E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 153
these programs could operate through the
parent or through the center services received
directly by the child. In the Infant
Health and Development Program, the positive
effects on the HOME Inventory accounted
for a portion of the IQ benefit at age
three.98 In the Early Head Start Program
Demonstration, about two-fifths to one-half
of the treatment effect on child cognitive test
scores operated through the program?s effect
on parenting behavior.99 Center-based programs
with a parenting component appear effective
at enhancing parenting and school
readiness, with some of the effect on the latter
operating through the former. These programs
are, in our opinion, a good bet for increasing
child well-being.
The Whitehurst literacy program noted
above also had positive child effects. Other
family literacy programs should similarly
yield benefits, with the effects assumed to
operate through parental language use. Although
we have fewer data on which to base
our opinion, we believe that these programs
also show promise for improving parenting
and school readiness. The parent behavior
training programs also have shown effects on
children when targeted to families whose
children have been identified as having problem
behavior.
Can Parenting Interventions Close
the Ethnic and Racial Gaps in
School Readiness?
If parenting interventions are to narrow ethnic
and racial school readiness gaps, they
must meet one of several conditions. First,
effective interventions should be offered to
proportionately more minority than nonminority
families. This could be achieved if
such programs were offered to families with
characteristics?for example, poorly educated
mother, unwed mother, or mother with
poor mental health?that are more often
found in minority than in white families.
Second, even if programs were not provided
to more minority than nonminority families,
they could still reduce the racial gaps if they
were more beneficial to black than white
parents. Third, even if parenting programs
were not more effective for black and Hispanic
than white parents, they could still
narrow ethnic and racial differences if they
were more beneficial to mothers with certain
characteristics, such as being young or
poorly educated, that are more prevalent
among black and Hispanic mothers than
white mothers.
Evidence on the first condition is scanty; estimates
of the shares of black, Hispanic, and
white families receiving parenting programs
do not exist. But more is known about the
second and third conditions. Parenting programs
sometimes do have more beneficial effects
for blacks than for whites and, to a
lesser extent, for younger than for older
mothers. That being the case, parenting programs,
if implemented, could reduce the
racial gap in school readiness.
Who Receives Parenting Programs?
Parenting interventions are almost always
targeted to specific groups, typically parents
who are poor, poorly educated, young and
unwed, live in impoverished communities, or
have mental health problems.100 As such,
they are likely to serve a greater share of minority
than nonminority families?a ratio of
three to one (or higher)?given the differential
prevalence of such conditions.101 No estimates
exist of the number of families with
young children served by parenting programs,
but two home-visiting programs that
focus on parenting?the Nurse Home Visitation
Model and the Healthy Start Model?
have been initiated countrywide.102
Jeanne Brooks-Gunn and Lisa B. Markman
154 THE FUTURE OF CHILDREN
In their article in this volume, Katherine
Magnuson and Jane Waldfogel note that 30
percent of all U.S. children under the age of
six are in some form of center-based child
care and education. Breaking that figure
down, they find 30 percent of white children,
22 percent of Hispanic children, and 40 percent
of black children in center-based care.
Some but not all child care programs also provide
parenting classes or home visiting; publicly
funded programs, such as Head Start, are
most likely to do so.103 Proportionately more
black than white children attend Head Start;
if these programs are effective in altering parenting
behavior, then Head Start could reduce
the racial gap in school readiness. Too
few studies have examined its efficacy vis-?is
parenting outcomes to make an inference
about the probability of Head Start as a path
to reducing racial gaps, but the program does
seem to have positive effects on children.104
Differing Program Effects on Black and
White Parents
If parenting interventions benefit black and
Hispanic parents more than white parents,
they could reduce gaps in school readiness.
Few demonstration programs have examined
this question, in large part because most parenting
programs operate in one community
or neighborhood, so that racial and ethnic
variation in participants is quite limited. But
two multisite demonstrations report larger
effects on black than white mothers in some,
but not all, aspects of parenting.
Through the Infant Health and Development
Program (IHDP), an eight-site randomized
control trial, about 1,000 families with low
birth weight children born in 1985 were offered
parenting-focused home-visit and center-
based child care from birth through the
child?s third year of life. The program assessed
HOME Inventory, mother-child free
play, and problem-solving videotaped interactions,
maternal mental health, and spanking.
105 According to analyses conducted for
this article, black mothers benefited more
from the program than did white mothers
when their children were age three (that is, at
the end of the intervention). Observers noted
more learning and less punitive discipline in
the homes of black mothers in the intervention
than those of black mothers in the control
group; effect sizes were about one-fifth
to one-quarter of a standard deviation, or 3 to
4 points on our reference test. We found no
corresponding treatment differences for the
white mothers.106 In both cases, the scores of
black mothers in the treatment group were
higher than those of their counterparts in the
control group and were comparable to those
of the white mothers in both the treatment
and the control groups.
Researchers report similar findings in the
Early Head Start Demonstration (EHS), a
randomized seventeen-site evaluation of
home- and center-based early childhood intervention
for pregnant women and young
children, conducted from the late 1990s into
2000.107 Black mothers in the intervention
group had more positive and fewer negative
parenting behaviors than did black mothers
in the control group; the effect sizes ranged
from one-fifth to one-half of a standard deviation
(3 to 7.5 points on our reference test).
T h e C o n t r i b u t i o n o f P a r e n t i n g t o E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 155
If parenting interventions
benefit black and Hispanic
parents more than white
parents, they could reduce
gaps in school readiness.
Researchers found these effects in eight parenting
behaviors measured at the end of the
intervention, when the children were three
years old. Hispanic mothers also benefited
from Early Head Start, although not as much
as black mothers and not in as many parenting
behaviors. The program had almost no
effect on the white mothers. The EHS intervention
raised the parenting scores of the
black mothers to levels similar to those of
the white mothers, mirroring the IHDP
results.
Differing Program Impacts by Maternal
Age, Education, and Mental Health
Programs could also reduce the ethnic and
racial gaps if they benefited mothers who
were poorer, younger, or single more than
other mothers, because these characteristics
are more likely among black and Hispanic
mothers than among white mothers. At least
three lines of evidence exist, the first relating
to maternal education, the second to maternal
age, and the third to maternal mental
health. We believe that it would be possible
to reduce racial gaps in school readiness if
the results described below could be replicated
in large-scale programs.
First, early childhood education programs
seem to have more benefits for children of
mothers with a high school education (or
less) than they do for children whose mothers
have some postsecondary schooling.108
Less information is available on whether
such programs affect parenting. In the
IHDP, even though children of less educated
mothers benefited more, their mothers did
not. The Early Head Start Demonstration
had somewhat greater effects on the parenting
behavior of the less educated than on
that of the more educated mothers, as well as
on child engagement and persistence in
mother-child play sessions.109 Effect sizes
range from one-fifth to one-quarter of a standard
deviation (3 to 5 points on our reference
test). At the same time, only EHS
mothers with more than a high school education
showed significant increases in reading
at bedtime and reductions in spanking.
These mixed findings signal caution in accepting
this pathway?larger effects for less
educated mothers?to reducing the racial
gap in school readiness.
Second, young and first-time parents might
also benefit more from parenting interventions
than older, more experienced parents.
And, indeed, whenever benefits of treatment
differ by parental age, they favor the younger,
typically teenage and unwed mother.110 Results
are stronger for the Nurse Home Visitation
Model than for EHS.
Third, although evidence is limited, parenting
interventions do appear to have greater
effects for mothers with low psychological resources.
Of the seventeen sites in the EHS
demonstration, eight asked mothers about
depressive symptoms before the intervention
began; those with more symptoms were more
likely than those with fewer symptoms to see
symptoms reduced during the intervention.
111 In IHDP, by contrast, all intervention
mothers experienced reduced depression
symptoms.112 Early Head Start had somewhat
greater effects on mothers? parenting
behaviors for those with initially high depressive
symptoms.113
David Olds and colleagues have reported
that their Nurse Home Visitation Model had
more positive effects on mothers with low
psychological resources (a measure comprising
mental health, sense of mastery, and
intelligence obtained before the intervention)
than on those with high psychological
resources.114
Jeanne Brooks-Gunn and Lisa B. Markman
156 THE FUTURE OF CHILDREN
Conclusion
Parenting influences young children in many
different ways. The frequency of certain parenting
behaviors, those often linked with
school readiness, are lower for black and Hispanic
mothers than for white mothers,
though adjustment for differences in family
conditions attenuates these differences to an
extent. These racial and ethnic differences in
parenting in large part parallel racial and ethnic
differences in school readiness. When
such parenting differences are controlled,
the gaps in school readiness drop 25 percent
to 50 percent.
It is possible to alter the parenting behavior
of black and Hispanic mothers. In several instances,
interventions have reduced the gaps
in the parenting behavior of black and white
mothers. In these cases, black children also
benefited more than white children from the
intervention. These successful programs have
been high-quality and center-based with a
parenting component (typically through
home visiting). Exclusively home-based programs
have not yielded comparable findings;
they affect the mother but not the child and
therefore (with a few exceptions) cannot narrow
ethnic and racial gaps in school readiness.
We cannot say from existing evidence
whether all center-based programs should
have a parenting component. There is little
evidence documenting the effects of parenting
components in publicly funded programs
such as Head Start. In addition, because virtually
all programs for children under age
four involve the parent, it is not known
whether a center-based program without a
parenting component is as effective as one
with such a component. The rise of the prekindergarten
programs may provide some insight,
because many such programs do not
target the parent in any significant way.
Whether such programs will show similar impacts
on children without parental involvement
remains to be seen. The exciting findings
of the new family literacy programs and
the parent behavior training programs also
provide possible avenues for targeted parenting
programs.
T h e C o n t r i b u t i o n o f P a r e n t i n g t o E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 157
Endnotes
1. Marc H. Bornstein, ed., Handbook of Parenting: Children and Parenting (Mahwah, N.J.: Lawrence Erlbaum
Associates, 2002); Eleanor Maccoby and John A. Martin, ?Socialization in the Context of the Family:
Parent-Child Interaction,? in Handbook of Child Psychology, vol. 4, Socialization, Personality and Social
Development, 4th ed., edited by E. Mavis Hetherington (New York: Wiley, 1983): p. 1.
2. There are other ways of categorizing parenting behaviors. See Lisa J. Berlin, Christy Brady-Smith, and
Jeanne Brooks-Gunn, ?Links between Childbearing Age and Observed Maternal Behaviors with 14-
Month-Olds in the Early Head Start Research and Evaluation Project,? Infant Mental Health Journal 23,
no. 1 (2002): 104?29; Bornstein, Handbook of Parenting (see note 1); Tama Leventhal and others, ?The
Homelife Interview for the Project on Human Development in Chicago Neighborhoods: Assessment of
Parenting and Home Environment for 3-15 Year Olds,? Parenting: Science and Practice 4 (2004); Maccoby
and Martin, ?Socialization in the Context of the Family? (see note 1); Eleanor Maccoby, ?The Role of
Parents in the Socialization of Children: An Historical Overview,? Developmental Psychology 28 (1992):
1006; Lawrence Steinberg and Ann S. Morris, ?Adolescent Development,?Annual Review of Psychology
52 (2000): 83?110.
3. Low nurturance is sometimes linked more closely with aggression or low self-control than with cognitive
and language skills, while the opposite is true for high nurturance behaviors. Jean Ispa and others, ?Maternal
Intrusiveness, Maternal Warmth, Mother-Toddler Relationship Outcomes: Variations across Low-
Income Ethnic and Acculturation Groups,? Child Development (forthcoming); Rebecca Ryan, Ann Martin
and Jeanne Brooks-Gunn, ?Is One Good Enough Parent Good Enough? Patterns of Father and Mother
Parenting and Their Combined Associations with Concurrent Child Outcomes at 24 and 36 Months,? Parenting:
Science and Practice (forthcoming).
4. Robert Bradley and Bettye Caldwell, Home Observation for Measurement of the Environment (University
of Arkansas, 1984); Robert Bradley, ?Home Environment and Parenting,? in Handbook of Parenting, vol.
2, Biology and Ecology of Parenting, edited by Marc Bornstein (Hillsdale, N.J.: Lawrence Erlbaum Associates,
1995), p. 235; Jeffrey B. Bingenheimer and others, ?Measurement Equivalence for Two Dimensions
of Children?s Home Environment,? Journal of Family Psychology (forthcoming); Leventhal and others,
?The Homelife Interview? (see note 2); Miriam Linver, Jeanne Brooks-Gunn, and Natasha Cabrera,
?The Home Observation for Measurement of the Environment (HOME) Inventory: The Derivation of
Conceptually Designed Subscales,? Parenting: Science and Practice (2004). The HOME Inventory, developed
in the 1980s, originally had more than fifty exemplars (items either observed in the home or reported
upon by the mother) of home conditions or parental behaviors that, if absent, might put the child at risk
for less than optimal development. The items were crafted to discriminate among those homes with quite
adverse circumstances; HOME does not differentiate particularly well among homes and families within
the wide range of acceptable to excellent circumstances (and was not designed to do so); see Robert
Bradley, ?Chaos, Culture, and Covariance Structures: A Dynamic Systems View of Children?s Experiences
at Home,? Parenting: Science and Practice 4 (2004). Many variants of HOME have been developed; the
variants have different numbers of items; forms have been developed for different age groups (the first
HOME focusing on early childhood); some forms have been adapted to be more similar across age groups
than earlier forms; and some forms separate scales with only observation and only parental report items;
see Linver and others, ?The Home Observation for Measurement of the Environment? (see above in this
note). Ethnic and racial differences in the coherence of scales have been examined as well; see Robert H.
Jeanne Brooks-Gunn and Lisa B. Markman
158 THE FUTURE OF CHILDREN
Bradley and others, ?The Home Environment of Children in the United States, Part I: Variations by Age,
Ethnicity, and Poverty Status,? Child Development 72 (2001): 1844; Bradley, ?Home Environment and
Parenting,? and Bingenheimer and others, ?Measurement Equivalence? (see Bradley and Bingenheimer
both above in this note). The HOME Inventory is the parenting measure that is used in most of the national,
longitudinal studies in the United States, Canada, and Australia.
5. NICHD Early Child Care Research Network, ?Child Care and Mother-Child Interaction in the First
Three Years of Life,? Developmental Psychology 35 (1999): 1399; Jeanne Brooks-Gunn and others, ?Depending
on the Kindness of Strangers: Current National Data Initiatives and Developmental Research,?
Child Development 71, no. 1 (2000): 257.
6. Berlin, Brady-Smith, and Brooks-Gunn, ?Links between Childbearing Age? (see note 2); NICHD Early
Child Care Network, ?Child Care and Mother-Child Interaction? (see note 5); Ryan, Martin, and Brooks-
Gunn, ?Is One Good Enough Parent Good Enough?? (see note 3).
7. Christy Brady-Smith and Jeanne Brooks-Gunn, analyses prepared for this article using data from the Early
Head Start Demonstration, 2004 (available from the National Center for Children and Families, Teachers
College, Columbia University); Ryan, Martin, and Brooks-Gunn, ?Is One Good Enough Parent Good
Enough?? (see note 3).
8. Kirby Deater-Deckard and others, ?Physical Discipline among African American and European American
Mothers: Links to Children?s Externalizing Behaviors,? Developmental Psychology 32, no. 6 (1996): 1065;
Sara Harkness and Charles Super, ?Culture and Parenting,? in Handbook of Parenting, vol. 2, Biology and
Ecology of Parenting, edited by Marc Bornstein (Mahwah, N.J.: Lawrence Erlbaum Associates, 1995),
p. 211.
9. Robert Bradley, ?Environment and Parenting,? Handbook of Parenting, 2nd ed., edited by Marc Bornstein
(Hillsdale, N.J.: Lawrence Erlbaum Associates, 2002), p. 281; Judith Smith and Jeanne Brooks-Gunn,
?Correlates and Consequences of Harsh Discipline for Young Children,? Archives of Pediatric and Adolescent
Medicine 151 (1997): 777; Leventhal and others, ?The Homelife Interview? (see note 2); Allison
Fuligni, Wen Jui Han, and Jeanne Brooks-Gunn, ?The Infant-Toddler HOME in the Second and Third
Years of Life,? Parenting: Science and Practice 4 (2004).
10. John Love and others, Making a Difference in the Lives of Infants and Toddlers and Their Families: The
Impacts of Early Head Start (U.S. Department of Health and Human Services, 2002).
11. Donna Spiker, Joan Ferguson, and Jeanne Brooks-Gunn, ?Enhancing Maternal Interactive Behavior and
Child Social Competence in Low Birth Weight, Premature Infants,? Child Development 64 (1993): 754;
Alan Sroufe, Byron Egeland, and Terri Kreutzer, ?The Fate of Early Experience following Developmental
Change: Longitudinal Approaches to Individual Adaptation in Childhood,? Child Development 61, no. 5
(1990): 1363; Lindsay Chase-Lansdale, Jeanne Brooks-Gunn, and Elise Zamsky, ?Young African-American
Multigenerational Families in Poverty: Quality of Mothering and Grandmothering,? Child Development
65, no. 2 (1994): 373.
12. Pamela K. Klebanov, Jeanne Brooks-Gunn, and Marie McCormick, ?Does Neighborhood and Family
Poverty Affect Mothers? Parenting, Mental Health and Social Support?? Journal of Marriage and the Family
56, no. 2 (1994): 455; Linver, Brooks-Gunn, and Cabrera, ?The Home Observation for Measurement?
(see note 4).
T h e C o n t r i b u t i o n o f P a r e n t i n g t o E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 159
13. Eve Clark, The Lexicon in Acquisition (Cambridge University Press, 1993); Betty Hart and Todd Risley,
Meaningful Differences in the Everyday Experience of Young American Children (Baltimore: Brookes,
1995); Janellen Huttenlocher and others, ?Early Vocabulary Growth: Relation to Language Input and
Gender,? Developmental Psychology 27 (1991): 236; Zehava Weizman and Catherine Snow, ?Lexical Input
as Related to Children?s Vocabulary Acquisition: Effects of Sophisticated Exposure and Support for Meaning,?
Developmental Psychology 37 (2001): 265.
14. Pia Rebello Britto and Jeanne Brooks-Gunn, ?Beyond Shared Book Reading: Dimensions of Home Process,?
New Directions for Child Development 92 (2001):73; Anat Ninio, ?Joint Book Reading as a Multiple Vocabulary
Acquisition Device,? Developmental Psychology 19 (1983): 445; Catherine Snow and Anat Ninio, ?The
Contracts of Literacy: What Children Learn from Learning to Read Books,? in Emergent Literacy: Writing
and Reading, edited by William Teale and Elizabeth Sulzby (Norwood, N.J.: Ablex, 1986), p. 116.
15. See note 13.
16. Catherine Snow, ?The Theoretical Basis for Relationships between Language and Literacy Development,?
Journal of Research in Childhood Education 6 (1991): 5; Anat Ninio and Catherine Snow, Pragmatic Development
(Boulder, Colo.: Westview, 1996); Catherine Haden, Elaine Reese, and Robyn Fivush, ?Mothers?
Extratextual Comments during Storybook Reading: Stylistic Differences over Time and across Texts,?
Discourse Processes 21, no. 2 (1996): 135; Pia Rebello-Britto, Allison Fuligni, and Jeanne Brooks Gunn,
?An Open Book? Effects of Home-Based Approaches on Children?s Literacy Development,? in Handbook
of Early Literacy Research, vol. 2, edited by David Dickinson and Susan Neuman (New York: Guilford,
forthcoming).
17. Linver, Brooks-Gunn, and Cabrera, ?The Home Observation for Measurement? (see note 4); Bradley and
Caldwell, Home Observation (see note 4).
18. Pamela K. Klebanov and others, ?The Contribution of Neighborhood and Family Income to Developmental
Test Scores over the First Three Years of Life,? Child Development 69 (1998): 1420?36; Susan Mayer,
What Money Can?t Buy: Family Income and Children?s Life Chances (Harvard University Press, 1997).
The HOME Inventory allows items such as pans, household objects, or cereal boxes to be counted when
they are used as musical instruments, for counting and classification, or for alphabet learning. It is not
clear how often such items are counted in any given study, however.
19. Allison S. Fuligni and Jeanne Brooks-Gunn. ?Measuring Mother and Father Shared Caregiving: An Analysis
Using the Panel Study of Income Dynamics-Child Development Supplement,? in Conceptualizing and
Measuring Father Involvement, edited by R. Day and M. Lamb (Mahwah, N. J.: Erlbaum, 2004).
20. Ibid.; Klebanov and others, ?The Contribution of Neighborhood and Family Income? (see note 18).
21. W. T. Boyce and others, ?The Family Routines Inventory: Theoretical Origins,? Social Science and Medicine
17 (1983): 193; Love and others, Making a Difference (see note 10).
22. Bornstein, Handbook of Parenting (see note 1); W. Andrew Collins and others, ?Contemporary Research
on Parenting: The Case for Nature and Nurture,? American Psychologist 55, no. 2 (2001): 218; Maccoby
and Martin, ?Socialization in the Context of the Family? (see note 1).
23. Betty Hart and Todd Risley, The Social World of Children Learning to Talk (Baltimore: Paul Brookes Publishing,
1999).
Jeanne Brooks-Gunn and Lisa B. Markman
160 THE FUTURE OF CHILDREN
24. Robert Plomin, ?Genetic and Environmental Mediation of the Relationship between Language and Nonverbal
Impairment in 4-Year-Old Twins,? Journal of Speech, Language, and Hearing Research 46, no. 6
(2003): 1271.
25. Thomas Bouchard Jr., ?Genetic Influence on Human Psychological Traits: A Survey,? Current Directions
in Psychological Science 13, no. 4 (2004): 148.
26. Michael Rutter and others, ?Testing Hypotheses on Specific Environmental Causal Effects on Behavior,?
Psychological Bulletin 127 (2001): 291; Michael Rutter, ?Nature, Nurture, and Development: From Evangelism
through Science toward Policy and Practice,? Child Development 73 (2002): 1.
27. M. Schiff and others, ?Intellectual Status of Working-Class Children Adopted Early into Upper-Middle-
Class Families,? Science 200 (1978):1503?04; Marie Skodak and Harold Skeels, ?A Final Follow-Up Study
of One Hundred Adopted Children,? Journal of Genetic Psychology 75 (1949): 85.
28. Michel Duyme, Annick-Camille Dumaret, and Stanislaw Tomkiewicz, ?How Can We Boost IQs of ?Dull?
Children? A Late Adoption Study,? Proceedings of the National Academy of Sciences, USA 96 (1999): 8790.
29. It is important to realize that test scores within these groups of children show some stability; correlations
between test scores in early childhood and adolescence were around .30. This demonstrates that even
when stability is found, meaning that the rank ordering of children is somewhat similar across age, it is
possible to increase mean scores (see Dickens, this volume).
30. Lynn Singer and others, ?Cognitive Outcomes of Preschool Children with Prenatal Cocaine Exposure,?
Journal of the American Medical Association 291, no. 20 (2004): 2448.
31. Earlier studies have used different classifications of living arrangements, often combining relative, adoptive,
and foster care. Children in relative care are often in the same household as the mother (that is, the grandmother
has custody of the child). One of these studies has a similar finding to that of Singer and others reported
here; see Toosje Thyssen Van Beveren, Bertis Little, and Melanie Spence, ?Effects of Prenatal Cocaine
Exposure and Postnatal Environment on Child Development,? American Journal of Human Biology 12
(2000): 417. Another does not; see Gideon Koren and others, ?Long-Term Neurodevelopmental Risks in
Children Exposed in Utero to Cocaine. The Toronto Adoption Study,? in Cocaine: Effects on the Developing
Brain, edited by Barry Kosofsky and others (New York: New York Academy of Sciences, 1998), p. 306.
32. The mothers? ratings could not be used, because they had already talked about their emotional feelings
about each twin.
33. Avshalom Caspi and others, ?Maternal Expressed Emotion Predicts Children?s Antisocial Behavior Problems:
Using MZ-Twin Differences to Identify Environmental Effects on Behavioral Development,? Developmental
Psychology 40 (2004): 149.
34. Collins and others, ?Contemporary Research on Parenting? (see note 22); Gerald Patterson, Barbara De-
Baryshe, and Elizabeth Ramsey, ?A Developmental Perspective on Antisocial Behavior,? American Psychologist
44, no. 2 (1989): 329; Sandra Scarr, ?Developmental Theories for the 1990s, Development and
Individual Difference,? Child Development 63 (1992): 1.
35. See note 13.
36. Helen Raikes and others, ?Mother-Child Bookreading in Low-Income Families: Correlates and Outcomes
during the First Three Years of Life,? unpublished, 2004. Early Head Start is a federal program offered by
T h e C o n t r i b u t i o n o f P a r e n t i n g t o E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 161
the Administration on Children, Youth, and Families for pregnant woman and their children from birth to
age three. Initiated in 1995 with 68 programs nationwide, as of 2004 it now serves 700 programs. The evaluation
was a randomized control trial in seventeen sites, with about 3,000 families assigned to receive either
Early Head Start services or not; Love and others, Making a Difference (see note 10).
37. Patterson, DeBaryshe, and Ramsey, ?A Developmental Perspective on Antisocial Behavior? (see note 34).
38. For reviews, see Bornstein, Handbook of Parenting (see note 1), and G. Downey and James Coyne, ?Children
of Depressed Parents: An Integrative Review,? Psychological Bulletin 108 (1990), p. 50.
39. Miriam Linver, Jeanne Brooks-Gunn, and Dafna Kohen, ?Family Processes as Pathways from Income to
Young Children?s Development,? Developmental Psychology 38, no. 5 (2002): 719; Jean Yeung and others,
?How Money Matters for Young Children?s Development: Parental Investment and Family Processes,?
Child Development 73, no. 6 (2002): 1861.
40. Glen H. Elder Jr., Children of the Great Depression: Social Change in Life Experience (Boulder, Colo.:
Westview Press, 1999).
41. Vonnie McLoyd, ?Socioeconomic Disadvantage and Child Development,? American Psychologist 53, no. 2
(1998): 185; Rand Conger, Katherine Conger, and Glen Elder, ?Family Economic Hardships and Adolescent
Adjustment: Mediating and Moderating Processes,? in Consequences of Growing Up Poor, edited by
Greg Duncan and Jeanne Brooks-Gunn (New York: Russell Sage Foundation, 1997), p. 288.
42. Yeung and others, ?How Money Matters? (see note 39).
43. Among these techniques are fixed-effects and longitudinal models, sibling models, and instrumental variable
approaches.
44. Harkness and Super, ?Culture and Parenting? (see note 8); Cynthia Garcia-Coll and others, ?Ethnic and
Minority Parenting,? in Handbook of Parenting, vol. 2: Biology and Ecology of Parenting, edited by Marc
Bornstein (Hillsdale, N.J.: Lawrence Erlbaum Associates, Inc., 1995).
45. Harkness and Super, ?Culture and Parenting? (see note 8).
46. Patricia M. Greenfield and others, ?Cultural Pathways through Universal Development,? Annual Review
of Psychology 54 (2003): 461.
47. Bornstein, Handbook of Parenting (see note 1).
48. Patricia Greenfield, ?Cultural Change and Human Development,? in Development and Cultural Change:
Reciprocal Processes, edited by Elliot Turiel (San Francisco: Wiley, 1999), p. 37; Garcia-Coll and others,
?An Integrative Model? (see note 69).
49. Gail A.Wasserman and others, ?Psychosocial Attributes and Life Experiences of Disadvantaged Minority
Mothers: Age and Ethnic Variations,? Child Development 61 (1990): 566.; Gontran Lamberty and Cynthia
Garcia-Coll, editors, Puerto Rican Women and Children: Issues in Health, Growth, and Development
(New York: Guilford, 1994).
50. Maccoby and Martin, ?Socialization in the Context of the Family? (see note 1).
51. Diana Baumrind, ?An Exploratory Study of Socialization Effects on Black Children: Some Black-White
Comparisons,? Child Development 43 (1972): 261.
Jeanne Brooks-Gunn and Lisa B. Markman
162 THE FUTURE OF CHILDREN
52. Jeanne Brooks-Gunn and Lindsay Chase-Lansdale, ?Adolescent Parenthood,? in Handbook of Parenting,
vol. 3, Status and Social Conditions of Parenting, edited by Marc Bornstein (Mahwah, N.J.: Lawrence Erlbaum
Associates, 1995), p. 113.
53. Deater-Deckard and others, ?A Genetic Study of the Family Environment? (see note 8); Vonnie McLoyd
and Julia Smith, ?Physical Discipline and Behavior Problems in African American, European American,
and Hispanic Children: Emotional Support as a Moderator,? Journal of Marriage and Family 64 (2002): 40.
54. These effect sizes are reduced when characteristics such as maternal age, education, marital status, and income
are controlled in regression analyses. These reductions range from 20 percent to 50 percent, depending
on the parenting measure and the sample (that is, the reductions are much less in low-income
samples, such as the Early Head Start Demonstration). Brady-Smith and Brooks-Gunn, analyses (see note
7); Pamela Klebanov and Jeanne Brooks-Gunn, analyses prepared for this article using data from the Infant
Health and Development Program, 2004 (available from the National Center for Children and Families,
Teachers College, Columbia University); Klebanov and others, ?The Contribution of Neighborhood
and Family Income? (see note 18); Klebanov, Brooks-Gunn, and McCormick, ?Does Neighborhood and
Family Poverty Affect Mothers? Parenting, Mental Health and Social Support?? (see note 12); Meredith
Phillips and others, ?Family Background, Parenting Practices, and the Black-White Test Score Gap,? in
The Black-White Test Score Gap, edited by Christopher Jencks and Meredith Phillips (Brookings, 1998),
p. 103; Raikes and others, ?Mother-Child Bookreading? (see note 36).
55. Jeanne Brooks-Gunn, Pamela Klebanov, and Fong-Ruey Liaw, ?The Learning, Physical, and Emotional
Environment of the Home in the Context of Poverty: The Infant Health and Development Program,?
Children and Youth Services Review 17, no. 1/2 (1995): 251; Klebanov, Brooks-Gunn, and McCormick,
?Does Neighborhood and Family Poverty Affect Mothers? Parenting, Mental Health and Social Support??
(see note 12); Linver, Brooks-Gunn, and Cabrera, ?The Home Observation for Measurement ? (see note
4); Phillips and others, ?Family Background,? (see note 54); Raikes and others, ?Mother-Child Bookreading?
(see note 54).
56. Bingenheimer and others, ?Measurement Equivalence? (see note 4); Robert Bradley and others, ?Early
Indications of Resilience and Their Relation to Experiences in the Home Environments of Low Birthweight,
Premature Children Living in Poverty,? Child Development, 65, no. 2 (1994): 346; Guang Guo and
Kathleen Harris, ?The Mechanisms Mediating the Effects of Poverty on Children?s Intellectual Development,?
Demography 37 (2000): 431; Linver, Brooks-Gunn, and Cabrera, ?The Home Observation for
Measurement? (see note 4).
57. Klebanov and Brooks-Gunn, analyses (see note 54); Brady-Smith and Brooks-Gunn, analyses (see note 7).
58. Berlin, Brady-Smith, and Brooks-Gunn, ?Links between Childbearing Age and Observed Maternal Behaviors?
(see note 2); Brady-Smith and Brooks-Gunn, analyses (see note 7).
59. Smith and Brooks-Gunn, ?Correlates and Consequences of Harsh Discipline? (see note 9); Bradley and
others, ?The Home Environment? (see note 4).
60. Love and others, Making a Difference (see note 10).
61. Less has been done vis-?vis racial differences in teaching than in language. The limited evidence suggests
that black-white differences exist, using measures such as quality of assistance in a teaching task; see
Spiker, Ferguson, and Brooks-Gunn, ?Enhancing Maternal Interactive Behavior? (see note 11).
T h e C o n t r i b u t i o n o f P a r e n t i n g t o E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 163
62. A very small sample of black and white families was followed, including thirteen high SES children (whose
parents were primarily professors, with one being black), twenty-three lower-middle-class children (from
working-class families, with ten being black), and six children on welfare (all of whom were black). Consequently,
race and social class are totally confounded at the upper and lower ends of the SES distribution.
See Hart and Risley, Meaningful Differences in the Everyday Experience (see note 13); Hart and Risley,
The Social World (see note 23).
63. Clark, The Lexicon in Acquisition (see note 13); Hart and Risley, The Social World (see note 23); Shirley
Brice Heath, Ways with Words (Cambridge University Press, 1983).
64. Clark, The Lexicon in Acquisition (see note 13); David Dickinson and Patton Tabors, eds., Beginning Literacy
with Language: Young Children Learning at Home and School. (Baltimore: Paul H. Brookes Publishing,
2001); Janellen Huttenlocher and others, ?Early Vocabulary Growth? (see note 13).
65. Such studies do not exist because the cost of taping and transcribing mother-child conversations is prohibitive
for large-scale studies. Thus, our knowledge of maternal language input and child language output is
gleaned from studies that are unable to look directly at reductions in racial gaps.
66. Elaine Reese and Adell Cox, ?Quality of Adult Book Reading Affects Children?s Emergent Literacy,?
Developmental Psychology 35, no. 1 (1999): 20; Ninio and Snow, Pragmatic Development (see note 16).
67. Lisa McCabe and Jeanne Brooks-Gunn, ?Pre- and Perinatal Home Visitation Interventions,? in Early
Child Development in the 21st Century: Profiles of Current Research Initiatives, edited by Jeanne Brooks-
Gunn and others (Teachers College Press, Columbia University, 2003), p. 145.
68. Pia Rebello Britto, Allison S. Fuligni, and Jeanne Brooks-Gunn, ?Reading, Rhymes, and Routines: American
Parents and Their Young Children,? in Childrearing in America: Challenges Facing Parents with
Young Children, edited by Neal Halfon and others (Cambridge University Press, 2002), p. 117; Raikes and
others, ?Mother-Child Bookreading? (see note 36).
69. The differences between white and Hispanic mothers are not explained by the fact that many Hispanic
mothers speak Spanish, and fewer Spanish than English children?s books are available in the United
States; see Cynthia Garcia?Coll and others, ?An Integrative Model for the Study of Developmental Competencies
in Minority Children,? Child Development 67 (1996): 1891. In the Early Head Start Demonstration,
both English-speaking and Spanish-speaking Hispanic mothers were less likely to read to their
two- and three-year-olds than were white mothers; Raikes and others, ?Mother-Child Bookreading? (see
note 36).
70. Fuligni, Han, and Brooks-Gunn, ?The Infant-Toddler HOME? (see note 9); Phillips and others, ?Family
Background? (see note 54); Raikes and others, ?Mother-Child Bookreading (see note 36).
71. Brooks-Gunn, Klebanov, and Liaw, ?The Learning, Physical, and Emotional Environment? (see note 55);
Klebanov, Brooks-Gunn, and McCormick, ?Does Neighborhood? (see note 12); Guo and Harris, ?The
Mechanisms Mediating? (see note 56); Phillips and others, ?Family Background ? (see note 54).
72. Jeanne Brooks-Gunn and others, ?Depending on the Kindness of Strangers: Current National Data Initiatives
and Developmental Research,? Child Development 71 (2000): 257.
73. Guo and Harris, ?The Mechanisms Mediating the Effects of Poverty? (see note 56); Mayer, What Money
Can?t Buy (see note 18); Phillips and others, ?Family Background? (see note 54).
Jeanne Brooks-Gunn and Lisa B. Markman
164 THE FUTURE OF CHILDREN
74. For reviews, see Jeanne Brooks-Gunn, Lisa Berlin, and Allison Sidle Fuligni, ?Early Childhood Intervention
Programs: What about the Family?? in Handbook of Early Childhood Intervention, 2nd edition, edited
by Jack P. Shonkoff and Samuel J. Meisel (New York: Cambridge University Press, 2000); Jeanne
Brooks-Gunn, Alison Fuligni, and Lisa Berlin, Early Child Development in the 21st Century: Profiles of
Current Research Initiatives (Teachers College Press, 2003); Jeanne Brooks-Gunn, ?Intervention and Policy
as Change Agents for Young Children,? in Human Development across Lives and Generations: The Potential
for Change, edited by P. Lindsay Chase-Lansdale, Kathleen Kiernan, and Ruth Friedman (Cambridge
University Press, 2004).
75. This article does not review the program impacts on children; see W. Steven Barnett, ?Long-Term Effects
of Early Childhood Programs on Cognitive and School Outcomes,? The Future of Children 5, no.3 (1995):
25?50; April Benesich, Jeanne Brooks-Gunn, and Beatrice Clewell, ?How Do Mothers Benefit from Early
Intervention Programs?? Journal of Applied Developmental Psychology 13, no. 3 (1992): 311; Janet Currie,
?Early Childhood Education Programs,? Journal of Economic Perspectives 15, no. 2 (2001): 213; Lynn
Karoly and others, Investing in Our Children: What We Know and Don?t Know about the Cost and Benefit
of Early Childhood Interventions (Santa Monica, Calif.: RAND, 1998); Magnuson and Waldfogel, this
volume.
76. See Brooks-Gunn, Berlin, and Fuligni, ?Early Childhood Intervention Programs? (see note 74).
77. Ibid.
78. Differential effects of parenting interventions for mothers who are and are not depressed are discussed in
the section on differential impacts by maternal characteristics. Our premise is that programs might want to
target services to families with mental health issues, because program effects might be largest for this
group.
79. Brooks-Gunn, Berlin, and Fuligni, ?Early Childhood Intervention Programs (see note 74).
80. Marinus Van IJzendoorn, Femmie Juffer, and Marja Duyvesteyn, ?Breaking the Intergenerational Cycle
of Insecure Attachment: A Review of the Effects of Attachment Based Interventions on Maternal Sensitivity
and Infant Security,? Journal of Child Psychology and Psychiatry and Allied Disciplines 36, no. 2
(1995): 225.
81. Seven out of eight programs reviewed by Brooks-Gunn, Berlin, and Fuligni, ?Early Childhood Intervention
Programs? (see note 74), report such effects, as does Early Head Start; John Love and others, Making
a Difference (see note 10). Positive impacts are much more likely to be found from coding of mother-child
interchanges than from using the Warmth Scale from HOME.
82. See Brooks-Gunn, Berlin, and Fuligni, ?Early Childhood Intervention Programs? (see note 74); John Love
and others, Making a Difference in the Lives of Infants and Toddlers and Their Families: The Impacts of
Early Head Start (U.S. Department of Health and Human Services, 2002).
83. Joseph Sperling and Isabell Lewis, Partners for Learning (Lewisville, N.C.: Kaplan, 1994).
84. John Love and others, Making a Difference (see note 10); Spiker, Ferguson, and Brooks-Gunn, ?Enhancing
Maternal Interactive Behavior? (see note 11).
85. Grover Whitehurst and others, ?Outcomes of an Emergent Literacy Intervention in Head Start,? Journal
of Educational Psychology 86 (1994): 542.
T h e C o n t r i b u t i o n o f P a r e n t i n g t o E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 165
86. David Dickenson and Susan Neuman, editors, Handbook of Early Literacy Research, vol. II (New York:
Guilford, forthcoming).
87. Brooks-Gunn, Berlin, and Fuligni, ?Early Childhood Intervention Programs? (see note 74).
88. Robert St. Pierre and Janet Swartz, ?The Even Start Family Literacy Program,? in Two Generation Programs
for Families in Poverty: A New Intervention Strategy, edited by Shelia Smith (Westport, Conn.:
Ablex Publishing, 1995), p. 37.
89. Daniel Nagin and Richard Tremblay, ?Parental and Early Childhood Predictors of Persistent Physical
Agresssion in Boys from Kindergarten to High School,? Archives of General Psychiatry 58 (2001): 389.
90. Bridget K. Hamre and Robert Pianta, ?Early Teacher-Child Relationships and the Trajectory of Children?s
School Outcomes through Eighth Grade,? Child Development 72 (2001): 625.
91. Allan E. Kazdin, ?Treatment of Antisocial Behavior in Children: Current Status and Future Directions,?
Psychological Bulletin 102 (1987): 187.
92. Carolyn Webster-Stratton, Jamilia Reid, and Mary Hammond, ?Preventing Conduct Problems, Promoting
Social Competence: A Parent and Child Training Partnership in Head Start,? Journal of Child Clinical
Psychology 30 (2001): 283; Carolyn Webster-Stratton and Ted Taylor, ?Nipping Early Risk Factors in the
Bud: Preventing Substance Abuse, Delinquency, and Violence in Adolescence through Interventions Targeted
at Young Children (0?8 years),? Prevention Science 2 (2001): 165; Carolyn Webster-Stratton, ?Preventing
Conduct Problems in Head Start Children: Strengthening Parenting Competences,? Journal of
Consulting and Clinical Psychology 66 (1998): 715.
93. Webster-Stratton, Reid, and Hammond, ?Preventing Conduct Problems? (see note 92).
94. Matthew R. Sanders, Karen M. T. Turner, and Carol Markie-Dadds, ?The Development and Dissemination
of the Triple P?Positive Parenting Program: A Multilevel, Evidence-Based System of Parenting and
Family Support,? Prevention Science 3 (2002): 173?89; Conduct Problems Prevention Research Group,
?Initial Impact of the Fast Track Prevention Trial for Conduct Problems I: The High Risk Sample,? Journal
of Consulting and Clinical Psychology 67 (1999): 631?47; Conduct Problems Prevention Research
Group, ?Initial Impact of the Fast Track Prevention Trial for Conduct Problems II: Classroom Effects,?
Journal of Consulting and Clinical Psychology 67 (1999): 648?57.
95. Barnett, ?Long-Term Effects of Early Childhood Programs? (see note 75); Deanne Gomby and others,
?Long-Term Outcomes of Early Childhood Programs: Analysis and Recommendations,? The Future of
Children 5, no. 3 (1995): 6. A notable exception is the Nurse Home Visitation Program, developed by Olds
and his colleagues; impacts on young children?s social-emotional well-being have been reported, as have
some impacts on adolescent outcomes. See David Olds and others, ?Effects of Nurse Home Visiting on Maternal
Life-Course and Child Development: Age-Six Follow-Up of a Randomized Trial,? Pediatrics (2004).
96. Researchers have conducted two comparative analyses of home-visiting programs. See Jean Layzer and
others, ?National Evaluation of Family Support Programs Final Report, vol. A, The Meta?Analysis,? submitted
to the Administration for Children, Youth, and Families (Cambridge, Mass.: Abt Associates, 2001)
(ERIC no. ED462186); and Monica Sweet and Mark Appelbaum, ?Is Home Visiting an Effective Strategy?:
A Meta-Analytic Review of Home Visiting Programs for Families with Young Children,? Child Development
75 (2004): 1435. Most recently Sweet and Applebaum examined sixty programs, both experi-
Jeanne Brooks-Gunn and Lisa B. Markman
166 THE FUTURE OF CHILDREN
mental and nonexperimental. Programs varied in length, target population, services, child age, and type of
home visitor (professional or para-professional), making it difficult to say much about specific components.
Almost all focused on groups of families at risk for poor child outcomes. Virtually all programs listed
parent education (98 percent) and child development (85 percent) as goals. The authors examined the efficacy
of the programs on ten outcomes, including parenting behavior, child cognitive outcomes, and child
emotional outcomes. The weighted effect sizes were significant for all three, but were much smaller
(about two-thirds smaller) for the experimental than the nonexperimental programs. There was some evidence
that cognitive effects were positive when programs lasted longer and included more home visits.
We speculate that home-visiting programs will be most likely to affect child outcomes if they have a schedule
similar to that of the Nurse Home Visitation Program and if they ensure that families receive the recommended
?dose? of visits (most families get fewer than half the visits planned by the program); see
Deanne Gomby and others, ?Long-Term Outcomes of Early Childhood Programs? (see note 95). We believe?
although evidence, either pro or con, is not available?that programs such as Whitehurst?s dialogic
reading program might be effective as part of a home-visiting program.
97. William Barnett, ?Early Childhood Education,? in School Reform Proposals: The Research Evidence, edited
by Alex Molner (Greenwich, Conn.: Information Age Publishing, 2002), p. 1; Lynn Karoly and others,
Investing in Our Children (see note 75); Craig Ramey and others, ?Early Educational Interventions for
High-Risk Children: How Center-Based Treatment Can Augment and Improve Parenting Effectiveness,?
in Parenting and the Child?s World: Influences on Academic, Intellectual, and Social-Emotional Development,
edited by Sharon Landesman Ramey and others (Mahwah, N.J.: Lawrence Erlbaum Associates,
2002), p. 125.
98. Robert Bradley and others, ?Impact of the Infant Health and Development Program (IHDP) on the Home
Environment of Infants with Low Birth Weight,? Journal of Educational Psychology 86 (1994): 531.
99. Love and others, Making a Difference (see note 10).
100. Other countries have, or have had, more universal parenting programs. A series of home visits after the
birth of a child are provided to all new mothers in several countries. See Shelia Kamerman, ?Early Childhood
Intervention Policies: An International Perspective,? in Handbook of Early Childhood Intervention,
edited by Samuel J. Meisels and others (Cambridge University Press, 2000), p. 613; Gomby and others,
?Long-Term Outcomes of Early Childhood Programs? (see note 95).
101. See Duncan and Magnuson, this volume, and Currie, this volume.
102. Olds and others, ?Effects of Nurse Home Visiting? (see note 95).
103. McCabe and Brooks-Gunn, ?Pre- and Perinatal Home Visitation? (see note 67).
104. See Magnuson and Waldfogel, this volume.
105. Infant Health and Development Program, ?Enhancing the Outcomes of Low-Birth-Weight, Premature
Infants: A Multisite, Randomized Trial,? Journal of the American Medical Association 263 (1990): 3035;
Pamela K. Klebanov, Jeanne Brooks-Gunn, and M. C. McCormick, ?Maternal Coping Strategies and
Emotional Distress: Results of an Early Intervention Program for Low Birth Weight Young Children,? Developmental
Psychology 37, 5 (2001): 654; Spiker, Ferguson, and Brooks-Gunn, ?Enhancing Maternal Interactive
Behavior? (see note 11); Smith and Brooks-Gunn, ?Correlates and Consequences of Harsh Discipline?
(see note 9).
T h e C o n t r i b u t i o n o f P a r e n t i n g t o E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 167
106. Klebanov and Brooks-Gunn, analyses (see note 54).
107. Love and others, Making a Difference (see note 10).
108. Fong-Ruey Liaw and Jeanne Brooks-Gunn, ?Patterns of Low Birth Weight Children?s Cognitive Development
and Their Determinants,? Developmental Psychology 29, no. 6 (1993): 1024; Jeanne Brooks-Gunn
and others, ?Enhancing the Cognitive Outcomes of Low Birth Weight, Premature Infants: For Whom Is
the Intervention Most Effective?? Pediatrics 89, no. 8 (1992): 1209; Love and others, Making a Difference
(see note 10).
109. Love and others, Making a Difference (see note 10). These results adjust for characteristics including maternal
race, age, parity, income, and marital status.
110. David Olds and others, ?Effects of Prenatal and Infancy Nurse Home Visitation on Surveillance of Child
Maltreatment,? Pediatrics 95 (1995): 365; Love and others, Making a Difference (see note 10).
111. Love and others, Making a Difference (see note 10).
112. Pamela Klebanov, Jeanne Brooks-Gunn, and Marie McCormick, ?Maternal Coping Strategies and Emotional
Distress: Results of an Early Intervention Program for Low Birth Weight Children,? Developmental
Psychology 37, no. 5 (2001): 654.
113. Love and others, Making a Difference (see note 10).
114. Harriet Kitzman and others, ?Effects of Prenatal and Infancy Home Visitation by Nurses on Pregnancy
Outcomes, Childhood Injuries, and Repeated Childbearing: A Randomized Controlled Trial,? Journal of
the American Medical Association 278 (1997): 644; David L. Olds and others, ?Home Visiting by Paraprofessionals
and by Nurses: A Randomized, Controlled Trial,? Pediatrics (2002): 486-96.
Jeanne Brooks-Gunn and Lisa B. Markman
168 THE FUTURE OF CHILDREN
Early Childhood Care and Education:
Effects on Ethnic and Racial Gaps
in School Readiness
Katherine A. Magnuson and Jane Waldfogel
Summary
The authors examine black, white, and Hispanic children?s differing experiences in early childhood
care and education and explore links between these experiences and racial and ethnic
gaps in school readiness.
Children who attend center care or preschool programs enter school more ready to learn, but
both the share of children enrolled in these programs and the quality of care they receive differ
by race and ethnicity. Black children are more likely to attend preschool than white children,
but may experience lower-quality care. Hispanic children are much less likely than white children
to attend preschool. The types of preschool that children attend also differ. Both black
and Hispanic children are more likely than white children to attend Head Start.
Public funding of early childhood care and education, particularly Head Start, is already reducing
ethnic and racial gaps in preschool attendance. The authors consider whether further increases
in enrollment and improvements in quality would reduce school readiness gaps. They
conclude that incremental changes in enrollment or quality will do little to narrow gaps. But
substantial increases in Hispanic and black children?s enrollment in preschool, alone or in combination
with increases in preschool quality, have the potential to decrease school readiness
gaps. Boosting enrollment of Hispanic children may be especially beneficial given their current
low rates of enrollment.
Policies that target low-income families (who are more likely to be black or Hispanic) also look
promising. For example, making preschool enrollment universal for three- and four- year-old
children in poverty and increasing the quality of care could close up to 20 percent of the blackwhite
school readiness gap and up to 36 percent of the Hispanic-white gap.
VOL. 15 / NO. 1 / SPRING 2005 169
www.future of children.org
Katherine Magnuson is assistant professor of social work at the University of Wisconsin?Madison. Jane Waldfogel is professor of social work
and public affairs at Columbia University. They are grateful for funding support from The Future of Children, the Russell Sage Foundation,
and the John D. and Catherine T. MacArthur Foundation. They received helpful comments on an earlier draft from Steve Barnett, Janet Currie,
Elisabeth Donahue, Rebecca Maynard, and the journal editors and benefited from many helpful discussions with Marcia Meyers, Dan
Rosenbaum, and Chris Ruhm. They also appreciate assistance from David Blau with the SIPP data.
For children growing up in the
United States, early childhood
care and education have become
an increasingly common
experience. Almost every child
entering kindergarten today has been in care
of some form, and a growing share of kindergartners
has attended preschool or received
center care. On average, preschool and center
care develop young children?s early academic
skills through enriching activities and
sometimes direct instruction.1 Yet the type
and quality of the care that children receive
varies widely. Hispanic children, for example,
are less likely, and black children are more
likely, than white children to be enrolled in a
preschool or in center care.
Do children?s differing experiences of early
childhood care and education affect racial
and ethnic gaps in school readiness? If so, do
they widen the gaps or narrow them? In this
article, we review research on the effects of
child care and education on young children?s
school readiness and look at racial and ethnic
differences both in who receives early childhood
care and education and in the amount
and quality of care.2 All three types of evidence
are important: for early childhood care
and education to influence racial and ethnic
gaps in school readiness, the enrollment, intensity,
or effects of these programs must differ
by race or ethnicity.
Early care and education might widen racial
and ethnic gaps if children from racial and
ethnic minority groups are less likely to be
enrolled in beneficial programs, spend less
time in them, attend lower-quality programs,
or benefit less from them. Conversely, preschool
experiences might narrow racial and
ethnic gaps if children from minority groups
are more likely to be enrolled, spend more
time in them, attend higher-quality programs,
or benefit more.
In discussing racial and ethnic gaps, we focus
on three groups: Hispanics, non-Hispanic
whites (whites), and non-Hispanic African
Americans (blacks). We note that these
groups are socially constructed and heterogeneous
categories that proxy for diverse ethnic
and cultural groups.3 Hispanic describes
first-generation immigrants, refugees from
Cuba, and Puerto Ricans, all of whom face
different circumstances in U.S. society, including
socioeconomic resources.4 In the
United States, the Hispanic and black categories
serve as markers for minority status
and its accompanying experiences of discrimination
and disadvantage.5 Hispanic and
black children face much higher rates of
poverty, particularly persistent poverty, than
do white children.
In this article, we first review the main types
of early childhood care and education and
their effects on school readiness. We then
summarize trends in enrollment and in the
quality of care for Hispanic, white, and black
children. We conclude by considering how
early childhood care and education might
help to narrow racial and ethnic gaps in
school readiness and by discussing the implications
for public policy.
Main Types of Early Childhood
Care and Education
Early childhood care and education programs
come in many forms. We categorize
these into three broad types: parental care,
informal care (by a relative, nanny, or
babysitter in the child?s own home or in a
babysitter?s or family day care provider?s
home), and center care or preschool (day
care center, nursery school, preschool, Head
Start program, or prekindergarten).
Katherine A. Magnuson and Jane Wa l d f o g e l
170 THE FUTURE OF CHILDREN
We focus most on the third category because
a host of studies has found that children who
attend center care or preschool programs
enter school more ready to learn. As noted,
this category includes many different types of
programs, and it is important to distinguish
between them.
Most children in preschool or center care attend
private programs, for which their parents
pay fees. Low-income working parents
may receive child care subsidies that offset
some of the costs, and other families with
working parents may also receive financial assistance
through tax provisions, including the
child and dependent care tax credit and the
dependent care assistance plan.6 Some center
care and preschool programs operate fullday
and year-round; others, only part-time or
during the school year.
Preschool attendance becomes more common
as children approach school age. Approximately
60 percent of four-year-old children
are in care during the year before they
enter kindergarten, up from about 17 percent
in care before their second birthday.7
The federal government does not regulate
preschool programs, and state regulations
vary widely in both stringency and enforcement.
8 One way to assess the quality of center
care is through ?structural? indicators,
such as more highly educated teachers,
smaller classes, and lower children-to-staff
ratios.9 Some studies suggest that caregiver
education may be particularly important.10
Quality varies widely from one program to
the next, but, on average, the quality of center
care programs, as measured by structural
indicators, is probably just ?mediocre.?11
A second, arguably better, way to measure
child care quality is for trained observers to
rate the quality of the ?process??the
warmth, responsiveness, and sensitivity of
caregivers, as well as the physical environment
and children?s activities.12 Thus measured,
few center-based programs are high in
quality; a substantial proportion rank low in
quality.13 The Cost, Quality, and Child Outcomes
Study, conducted in 1993, found good
or developmentally appropriate care in only
24 percent of centers serving preschool-age
children. Quality was poor in 10 percent.
Child-caregiver interactions were positive in
less than half.14 The National Institute of
Child Health and Human Development
(NICHD) Study of Early Child Care found
similarly low rates of positive child-caregiver
interactions in center care.15
A small but growing share of children attend
publicly funded preschools, most commonly
Head Start and prekindergarten (other public
programs exist, but they serve few children).
Head Start, the largest publicly funded early
education program, began in 1965 as part of
President Lyndon B. Johnson?s War on
Poverty. It serves children from families with
incomes below the federal poverty threshold,
as well as children with disabilities.16 Under
Head Start, federal grants are provided to
local community organizations that offer early
education and comprehensive health, nutrition,
and family services to three- and fouryear-
old children.17 In 2002 the federal government
distributed $6.3 billion to local Head
Start grantees, who served an estimated 65
percent of eligible three- and four-year-olds,
some 10 percent of all children in that age
group.18
To receive funding, Head Start programs
must meet twenty-four federal performance
guidelines. Centers undergo an on-site review
at least once every three years. In 2000
about 85 percent of reviewed centers met the
E a r l y C h i l d h o o d C a r e a n d E d u c a t i o n : E f f e c t s o n E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 171
standards of adequate care. According to a
recent study of Head Start, programs met or
exceeded recommendations of the National
Association for the Education of Young Children
(NAEYC, a leading group of experts in
the field) for class size and adult-to-child ratios.
Judged by process quality, on average
Head Start centers are on par with other
types of center care.19 Nevertheless, only
one-third of Head Start teachers hold fouryear
college degrees, and experts worry that
low pay and low levels of provider education
constrain program quality.20
Prekindergarten programs, often funded
through local school districts, are a more recent
type of early education.21 As the name
suggests, they provide a year (or two) of education
before children enter kindergarten.
Publicly funded programs rely mainly on
state dollars, although local school districts
may also use federal Title 1, disability, or
other types of funds. Prekindergarten programs
may operate in public schools, but
some states also directly fund, and school districts
may subcontract with, other programs
to provide early education services. Typically,
prekindergartens offer some services beyond
education, including meals and transportation,
but few provide a full array of services
such as health screenings.22
Since 1990, state funding for prekindergarten
has increased 250 percent, to approximately
$1.9 million in 2002, but state spending
varies widely.23 In 2000, thirty-nine states
had prekindergarten initiatives, but only
seven (Connecticut, Georgia, Illinois, Kentucky,
Massachusetts, Ohio, and Oklahoma)
made substantial per capita investments in
them.24 Most state programs target disadvantaged
three- and four-year-old children and
serve a small but growing share of children,
with an estimated 14 percent of four-yearolds
enrolled in public school?based
prekindergarten programs in 2002.25 Only
two states, Georgia and Oklahoma, and the
District of Columbia offer such programs to
all children; they serve slightly more than half
of their four-year-olds.
Structural quality indicators suggest that
prekindergarten programs provide relatively
high-quality care.26 Most states set guidelines
for class size and child-to-caregiver ratios
that meet or exceed NAEYC recommendations.
The average size of general education
prekindergarten classes in public schools is
well within NAEYC guidelines.27 Of schoolbased
prekindergarten teachers, 86 percent
have four-year college degrees, more than
twice the rate among center care and Head
Start teachers. Teachers? pay is also more
likely to be commensurate with that of elementary
school teachers (82 percent receive
public school teacher salaries) and considerably
higher than that of other child care
workers.28 State-funded prekindergarten
programs in private preschools, however, appear
to have lower structural quality than
programs in public schools.29
Katherine A. Magnuson and Jane Wa l d f o g e l
172 THE FUTURE OF CHILDREN
Most state programs target
disadvantaged three- and
four-year-old children and
serve a small but growing
share of children, with an
estimated 14 percent of fouryear-
olds enrolled in public
school?based prekindergarten
programs in 2002.
Data on process quality in prekindergarten
programs are in short supply. Because structural
indicators are linked to process quality
and are higher for prekindergarten than for
other types of center care, prekindergarten
classrooms could be expected to have higher
process quality, too. Indeed, an evaluation of
Georgia?s universal prekindergarten found
the classrooms to be of higher process quality
than private preschool classrooms in that
state and less likely than Head Start classrooms
to be of poor quality.30 But an evaluation
of New Jersey?s Abbott preschool program
argues for caution, because it found
classroom quality was lower than that in
Georgia and lower than national estimates of
center care quality.31 The lack of information
on prekindergarten classroom quality makes
any general conclusions about process quality
unwarranted.
Effects of Early Childhood Care
and Education on Children?s
School Readiness
Can early childhood care and education raise
children?s test scores and promote school
readiness? Because space does not permit a
comprehensive review of the literature, we
summarize the best evidence on preschool
and center care, as well as informal and
parental care.
The best estimates of the effects of early
childhood care and education come from
random-assignment experimental studies.
These compare children in a particular program
with children who were not in the program
but were otherwise equivalent on important
background characteristics, thus
assuring that any differences in children?s academic
outcomes must be due to their experiences
in care. Random-assignment studies,
however, are rare. And researchers who conduct
them typically evaluate high-quality programs
that serve only a few children, often at
a single site, making it hard to generalize
findings to large-scale programs or more diverse
populations of children.
Many nonexperimental studies consider the
effects of more typical early childhood care
and education on children?s school readiness
by taking advantage of naturally occurring
variation in child care arrangements. But
these observational studies may identify effects
that in fact reflect unobserved factors,
such as socioeconomic status, that cause children
to receive a particular type of care. Because
the analyses often include only a few
statistical controls for such factors, their findings,
although more generalizable to other
programs and children, typically do not provide
convincing evidence that an effect has
been caused by the child?s experience in
care.32
Experimental Evaluations of
High-Quality Model Programs
Over the past thirty years, researchers have
conducted experimental evaluations of several
high-quality model programs in compensatory
early education. These model programs,
which primarily enroll economically
disadvantaged children, provide developmentally
appropriate education, often in
combination with health, nutrition, parenting
education, and family support services. With
highly trained teachers and low child-to-staff
ratios, they offer quality far superior to most
typical early education programs.
Not surprisingly, these programs enhance
children?s cognitive development and academic
skills at school entry.33 For example, in
the Infant Health and Development Program
(IHDP), which provided full-time highquality
center care to low birth weight children
between birth and age three, the heav-
E a r l y C h i l d h o o d C a r e a n d E d u c a t i o n : E f f e c t s o n E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 173
ier low birth weight children had IQ scores
close to 4 points higher than their counterparts
in the comparison group at ages five
and eight.34 Children from the most disadvantaged
backgrounds, as measured by
maternal education, gained the most.35
The academic benefits of these model programs
persist, although they fade over time.
Children who in their first five years received
high-quality care from the Carolina
Abecedarian project continued to outperform
a comparison group on IQ tests at ages
eight and fifteen by just over one-third of a
standard deviation.36 Furthermore, exemplary
programs reduce children?s special education
placement and grade retention.37 Children
who attended Perry Preschool, for
example, received special education services
for an average of 1.1 years, as against 2.8
years for comparison children.38
Because most programs were developed to
improve children?s academic skills and cognitive
development, few studies have considered
whether they also improve children?s social
skills and behavioral problems. Indeed,
only the IHDP has documented short-term
positive effects on children?s behavior.39 But
several long-term follow-up studies have
found lower rates of juvenile delinquency
and antisocial behavior, as measured by criminal
activity.40 It is not yet clear whether longterm
declines in problem behavior follow
from positive effects on young children?s behavior
or emerge later in childhood.
Head Start
Clearly, high-quality model early childhood
programs can enhance the school readiness of
disadvantaged children, but what about other
types of programs? Has Head Start done the
same for the disadvantaged or disabled children
it serves? Answering this question is difficult
because the program has never been
evaluated by a random assignment study (although
one is now under way). Researchers
using nonexperimental designs must find an
appropriate comparison group, and as Head
Start enrollees became increasingly disadvantaged
during the 1980s and 1990s, constructing
an appropriate comparison group may
have become even more difficult.41
A series of observational studies with data
collected during the 1970s and 1980s found
generally modest, short-term positive effects
of Head Start participation on disadvantaged
children?s school readiness.42 For example,
Valerie Lee and colleagues found that black
children who attended Head Start gained
0.25 of a standard deviation more on a test of
verbal skills by the end of first grade than did
black children who attended no early education
program.43 Head Start also improved
children?s social competence.
The studies that have most successfully controlled
for the disadvantaged background of
the children enrolled in Head Start may be
those that compare children who attended
the program with their siblings who did not.
Using this method, a series of parallel analyses
across two large data sets finds that attending
Head Start enhanced children?s cognitive
development. Six-year-old Head Start
children scored close to 7 percentile points
higher on a vocabulary test than their siblings
who did not attend preschool.44 The benefits
appeared to persist through elementary
school for white and Hispanic children, but
not for black children.45 Furthermore, follow-
up analyses found that Head Start children
engaged in less criminal activity as they
grew older.46
Thus, Head Start appears to have beneficial
cognitive and behavioral effects for the chil-
Katherine A. Magnuson and Jane Wa l d f o g e l
174 THE FUTURE OF CHILDREN
dren it serves, though how large the effects
are, how long they persist, and whether they
vary by race and ethnic group remain unclear.
Evidence from the random assignment
study now under way should shed further
light on these questions.
Quasi-Experimental and Observational
Studies of Prekindergarten Programs
Do prekindergarten programs improve children?s
school readiness? In the absence of
large-scale experiments, we cannot answer
this question with certainty. Researchers have
undertaken at least twenty evaluations of state
prekindergarten programs, but many are so
methodologically weak as to raise questions
about their findings.47 Several rigorous quasiexperimental
and observational studies, however,
suggest that school-based early education
programs can enhance readiness.
The first of these studies evaluated the
Chicago Child Parent Centers (CPC), a
prekindergarten program provided by the
Chicago public school system to predominantly
African American children living in
poor neighborhoods.48 CPC, a part-day preschool
for three- to four-year-olds, was
staffed by teachers with college degrees and
early childhood certification; it offered a
follow-on program during the early elementary
school years. The preschool program
emphasized early language development,
promoted parental involvement, and offered
comprehensive services such as meals and
health screenings. The follow-on program
provided smaller classes and programming to
keep parents involved in their children?s
schooling. Because the program was neighborhood
based, the researchers were able to
compare CPC children with children from
poor communities that did not have CPC
programs. Children who attended CPC during
the year before kindergarten scored 0.64
of a standard deviation higher on an assessment
of academic skills in the fall of kindergarten.
49 Accumulated evidence suggests
that preschool contributed to lasting improvements
in CPC children?s reading and
math achievement, as well as high school
graduation.50
More recently, researchers evaluated the
Tulsa prekindergarten program, part of Oklahoma?s
universal prekindergarten initiative.
Tulsa?s program offers part- or full-day early
education to any child who turns four by September
1; classes are held at local public
schools, and teachers have at least a college
degree. Taking advantage of the program?s
strict age cutoff for entry, evaluators compared
children at kindergarten entry who had
met the age cutoff and attended prekindergarten
with those who had missed the age
cutoff. Prekindergarten boosted children?s
language skills by 0.39 of a standard deviation,
with the largest effects for Hispanic and
black children who attended full-day.51
Observational studies also find positive
prekindergarten effects on school readiness.
E a r l y C h i l d h o o d C a r e a n d E d u c a t i o n : E f f e c t s o n E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 175
Head Start appears to have
beneficial cognitive and
behavioral effects for the
children it serves, though
how large the effects are,
how long they persist,
and whether they vary
by race and ethnic group
remain unclear.
One such study evaluated Georgia?s universal
prekindergarten program, delivered by private
providers and public schools.52 In our
own analyses, we used national data from the
Early Childhood Longitudinal Study?
Kindergarten Cohort (ECLS-K). In this national
sample of children entering kindergarten
in 1998, the 17 percent who had
attended prekindergarten scored 0.19 of a
standard deviation higher on a reading and
math skills assessment at school entry than
otherwise comparable children who spent
the previous year in exclusively parental care.
The children who had attended prekindergarten
also performed better at school entry
than children who had attended other types
of center care.53 From their review of states?
prekindergarten evaluations, William Gilliam
and Edward Zigler conclude that although
most studies are methodologically weak, evidence
is accumulating that prekindergarten
programs have positive short-term effects on
children?s academic skills.54
The evidence on the effects on social skills
and behavior is more mixed. The CPC studies
have not explored effects on children?s social
skills or problem behavior at school entry,
but have found lower levels of adolescent
delinquency, as measured by arrest records.
The Tulsa prekindergarten evaluation found
no effect on children?s behavior as they entered
school. Our own work with the ECLSK
finds that children who attend prekindergarten
have more problem behavior at school
entry than do children in parental care.55
Likewise, evaluations of state prekindergarten
programs do not consistently find improved
behavior at school entry, though, as
noted, many of these studies are methodologically
flawed.56
Research on prekindergarten programs is still
in its infancy, and much remains to be
learned. Few studies follow children long
enough to know whether benefits to school
readiness are likely to persist. In addition,
few studies describe well the quality of
prekindergarten programs being studied or
identify program characteristics that might
contribute to or hinder children?s school
readiness. Finally, whether prekindergarten
has short- or long-term effects on children?s
behavior is unclear.
Observational Studies of Other Types
of Early Childhood Care and Education
Most children do not attend model programs,
prekindergarten, or Head Start. What do we
know about the effects of privately funded
preschools, nursery schools, and day care
centers, as well as informal care and parental
care? Most observational studies lump together
several care arrangements into broad
categories, providing estimates, for example,
of the effects of center-based care or informal
care.
Whereas estimating the effects of Head Start
is complicated by the disadvantaged background
of the children, evaluating centerbased
care is problematic because of the
children?s relatively advantaged family backgrounds.
The best observational studies use
various techniques to reduce bias from the
characteristics of children that cause or coincide
with center care enrollment. Methodological
concerns notwithstanding, these
studies find that attending center care at, for
example, a day care center, nursery school, or
preschool, particularly at ages three and four,
promotes children?s academic skills and cognitive
development.57 Center care during a
child?s first three years may also enhance cognitive
development, particularly for disadvantaged
children, although evidence is less consistent
for infants and toddlers than for
preschool-age children.58
Katherine A. Magnuson and Jane Wa l d f o g e l
176 THE FUTURE OF CHILDREN
A particularly informative study, by Greg
Duncan and colleagues, used data from the
NICHD Study of Early Child Care to model
changes in children?s cognitive development
as a function of time spent in child care.59 By
relying on intra-individual change to identify
effects, the authors greatly reduced the likelihood
of bias caused by the children?s advantaged
family backgrounds. They found that
by attending center care at ages three and
four, children gained between 0.22 and 0.33
of a standard deviation more on measures of
academic achievement than children in
parental or informal care. And children
whose cognitive ability was lowest gained the
most. Yet, they also found that attending center-
based care from birth to age three was not
consistently linked to higher academic
achievement.60
We and our colleagues have used data from
the Early Childhood Longitudinal Study?
Kindergarten Cohort of 1998?99 to analyze
the effects of center care on children?s reading
and math skills.61 Children who attended
center care (including prekindergarten) the
year before entering school performed better
on academic skills assessments than their
peers. After controlling for a host of family
background and other factors that might be
associated with center care attendance, we
found positive effects at school entry (effect
sizes of about 0.14) that persisted into first
grade (effect sizes of about 0.06). In most instances,
the effects were largest (ranging from
0.16 to 0.23) for disadvantaged groups, measured
by such indicators as family income,
parental education, and family structure.
Center care may have some adverse effects.
Observational studies link all types of nonmaternal
care, including center care, with increased
problem behavior and aggression in
preschool and early school.62 Effects are
more pronounced for children who enter
nonmaternal care at an early age, are in care
for many hours, and attend center care. Although
the links between center care and increased
problem behavior are consistent, we
are uncertain what to make of these findings,
for several reasons. First, because all the evidence
comes from observational studies, the
links may not be causal. Second, the effects
are relatively small. The NICHD study suggests
that attending center care from birth to
age fifty-four months would result in an increase
of only 0.10 of a standard deviation in
teacher reports of conflict, and most children
in center care did not exhibit serious behavior
problems or aggression.63 Whether such
small differences in children?s behavior have
any long-term implications for their wellbeing
is unclear. Finally, researchers do not
understand what explains the problem behaviors
or how much effects may differ depending
on program and child characteristics.
Some children attend no center care or preschool
before starting formal education. They
are cared for by their parents or informal
caregivers, such as relatives, babysitters, nannies,
or family day care providers. Informal
child care is most prevalent during children?s
earliest years; it is the primary child care
arrangement for about 38 percent of infants.
64 Again, studies of informal and
parental care are limited by their reliance on
observational, rather than experimental, data.
Most find that, on average, informal care
does not influence children?s cognitive development
or academic skills, though, as noted,
it may be linked to increases in problem behavior.
However, these average effects may
mask considerable variability in effects because
of differences in the quality of care.
Research consistently links higher-quality informal
care to better cognitive development
and positive behavior.65
E a r l y C h i l d h o o d C a r e a n d E d u c a t i o n : E f f e c t s o n E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 177
In the cohort of children in kindergarten in
1998?99, about 17 percent had been in
parental care the year before, and 12 percent
had been in informal child care (including
care by a relative, babysitter, or nanny).66 In
terms of school readiness, children in
parental and informal child care fared similarly.
Compared with children who attended
some form of preschool, children who had
only parental or informal care entered school
with lower reading and math scores, but with
better behavior and self-control, even after a
host of child and family characteristics had
been taken into account.
Racial and Ethnic Differentials in
Enrollment in Early Childhood
Care and Education
To consider how children?s experiences in
early childhood care and education may be
affecting racial and ethnic gaps in school
readiness, we examine racial and ethnic differences
in enrollment in different types of
care. We start by comparing rates of Hispanic,
black, and white children?s enrollment
in center care or preschool programs over
time, making use of data from the October
Current Population Survey (CPS) from 1968
to 2000.67 Despite minor changes in question
wording over the period, the October CPS
provides fairly consistent data on the enrollment
of three- to five-year-olds in center care
and preschool (including nursery schools,
Head Start, and prekindergarten).68 We
focus on enrollment trends for three- and
four-year-olds, because kindergarten is now
almost universal for five-year-olds.
In recent decades, preschool enrollment has
grown steadily for three- and four-year-olds
from all racial and ethnic groups (figures 1
and 2).69 Yet racial and ethnic differences in
enrollment are still evident. From the late
1960s through the early 1980s, black threeand
four-year-olds were slightly more likely
than their white peers to attend preschool.
Starting in the mid-1980s, however, black
children?s enrollment stagnated, while white
children?s enrollment continued to increase.
Trends since the mid-1990s suggest that
black children may have regained their enrollment
advantage. Rates of preschool enrollment
for Hispanic children have remained
consistently below those of other
children. In 2000, only 23 percent of Hispanic
three-year-olds were in preschool compared
with 49 percent and 43 percent of their
black and white peers, respectively. Gaps are
also apparent for Hispanic four-year-olds.
In fact, racial and ethnic differences in enrollment
in center care or preschool pro-
Katherine A. Magnuson and Jane Wa l d f o g e l
178 THE FUTURE OF CHILDREN
Figure 1. Preschool Enrollment of Three-Year-Olds, by Race and Ethnicity, 1968?2000
60
50
40
30
20
10
0
White
Hispanic
Black
2000 1996 1992 1988 1984 1980 1976 1972 1968
Percent
Source: Current Population Survey.
grams exist for young children in all age
groups. Table 1 describes the care and education
arrangements of children under age six
in 1999.70 As expected, young white children
are somewhat less likely to be enrolled in
center care or preschool than black children
(panel A). Black children are more likely than
white children to attend center care as their
primary arrangement (33 percent versus 26
percent) or to attend any center care,
whether as a primary or secondary arrangement
(40 percent versus 30 percent). Again,
Hispanic children are the least likely to be in
center care (22 percent).
If one looks only at children with employed
mothers (panel B), the patterns remain quite
similar, suggesting that different rates of maternal
employment do not explain the disparities
in enrollment. Thus, the fact that black
mothers are more likely to be employed fulltime
than white mothers is not the only reason
why a greater share of black children is
enrolled in center care.71 Even within families
with employed mothers, black children
are more likely to be in center care than
white children.72
As table 1 shows, the type of early childhood
care and education also varies by family income.
Families with the highest incomes (at
or above 200 percent of the poverty threshold)
are most likely to use preschool or center
care. Because child care subsidies and Head
Start and prekindergarten programs are targeted
to economically disadvantaged families,
families in poverty are more likely to use
center care than are those with incomes between
100 percent and 200 percent of the
poverty threshold.
Although black children are more likely to be
in center care than white children, they are
not enrolled in the same types of programs.
As noted, black and Hispanic children are
more likely to be economically disadvantaged
than white children, and thus are more likely
to participate in publicly funded preschool
programs. More than 20 percent of black and
15 percent of Hispanic three- and four-yearolds
are enrolled in Head Start, compared
with about 4 percent of white children.73
These racial and ethnic differentials in participation
suggest that Head Start probably has
played an important role in equalizing rates of
black and white children?s participation in early
education. Assuming that children attending
Head Start centers would not receive any center
care in its absence, then relative to white
E a r l y C h i l d h o o d C a r e a n d E d u c a t i o n : E f f e c t s o n E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 179
Figure 2. Preschool Enrollment of Four-Year-Olds, by Race and Ethnicity, 1968?2000
80
70
60
50
40
30
20
10
0
White
Hispanic
Black
2000 1996 1992 1988 1984 1980 1976 1972 1968
Percent
Source: Current Population Survey.
children gaps in enrollment might be as large
as 9 percentage points for black children and
31 percentage points for Hispanic children.74
What does this imply for Head Start?s effectiveness
in narrowing the black-white
achievement gaps? Answering this question
requires an accurate estimate of Head Start?s
effects on children, which to date have not
been established. We offer an upper bound of
the possible effects by using estimates from
the quasi-experimental study of the Chicago
Child Parent Centers.75 The estimate is likely
to be an upper bound because the CPC had
more highly qualified teachers than most
Head Start centers.76 Arthur Reynolds reported
that the effect of participating in CPC
for one year was 0.64 of a standard deviation
increase in academic skills in the fall of
kindergarten.77 If Head Start boosts skills as
much as CPC, then with 19 percent of black
children in Head Start, black children?s skills
Katherine A. Magnuson and Jane Wa l d f o g e l
180 THE FUTURE OF CHILDREN
Table 1. Share of Children under Age Six in Child Care, by Type of Child Care, 1999
Percent
Primary care arrangement
Race/ethnicity and Other Center-based Any center-based
poverty status of Maternal Paternal Relative nonrelative Family day care and care and
children care care care care care education education1
Panel A: All Children
All children 28 12 21 7 7 25 30
Race/ethnicity
White 29 13 18 8 7 26 30
Hispanic 37 10 25 4 6 19 22
Black 17 9 30 5 7 33 40
Other 30 11 32 2 5 20 27
Poverty status
Below 100% poverty 38 6 23 3 6 23 27
100?200% poverty 33 12 25 4 5 20 24
Above 200% poverty 23 13 19 9 8 28 33
Panel B: Children of Employed Mothers
All children 5 19 27 11 9 29 37
Race/ethnicity
White 5 21 22 13 9 30 38
Hispanic 5 20 37 7 12 19 25
Black 3 13 34 5 8 38 45
Poverty status
Below 100% poverty 5 16 34 6 10 28 36
100?200% poverty 5 23 34 7 8 23 30
Above 200% poverty 4 19 23 13 9 32 39
Other 4 19 45 3 7 23 5
Source: Authors? calculations of 1999 SIPP data. Distribution of children across primary care arrangements may not sum to 100 because
of rounding of numbers.
1. Includes center-based care or education that was reported as a secondary care arrangement.
would be about 0.12 of a standard deviation
lower, on average, if they did not attend Head
Start or other early education programs. Since
the black-white test score gap is estimated at
close to 0.50 of a standard deviation, such a
reduction implies that the black-white test
score gap would be about 24 percent larger
(at 0.62 of a standard deviation) in the absence
of Head Start. The proportions of Hispanic
and black children in Head Start are
similar; it is therefore likely that the program
also has reduced Hispanic-white test score
gaps. In terms of lower bounds, we think it is
likely that Head Start?s effects are greater
than zero and thus are fairly confident that
the program has played an equalizing role.
Have other public preschool programs also affected
racial and ethnic patterns of preschool
enrollment or achievement gaps? Prekindergarten
is more likely to be offered in schools
with a large percentage of racial and ethnic
minority children, which suggests that black
and Hispanic children may be more likely
than white children to attend publicly funded
prekindergarten. However, precise national
estimates of the number of black, Hispanic,
and white children attending publicly funded
prekindergarten programs are not available.78
Racial and Ethnic Differences in
the Intensity and Quality of Early
Childhood Care and Education
Comparing racial and ethnic enrollment
trends tells only part of the story. Other important
pieces of evidence are the time spent
in preschool and the quality of programs attended
by white, black, and Hispanic children.
Unfortunately, information on racial
and ethnic patterns in hours and quality of
center care is hard to find.
Lacking published estimates of the number
of hours a week spent in preschool and center
care by children of different racial and
ethnic groups, we turn to the ECLS-K data
set for estimates of the average number of
hours that children were in center care (including
Head Start, prekindergarten, and
preschool) during the year before kindergarten.
Racial and ethnic differences are evident:
both black and Hispanic children spent
significantly more time in center care each
week (thirty-one and twenty-three hours, respectively)
than did white children (twenty
hours). National data sets find similar patterns
for hours spent by young children in all
types of nonparental care.79
Should one conclude that the longer time
spent by black and Hispanic children in center
care narrows the gap? Again, we are uncertain,
because the answer should be based
on precise estimates of the additional benefits
derived from thirty hours of care rather
than twenty hours, but none is available.
Finding no evidence that minority children
are spending less time in preschools than
white children, however, we are confident
that differences in the number of hours that
children spend in center care are not widening
achievement gaps.
As noted, the quality of child care can be
measured by structural indicators, such as
teacher certification and education, class
size, and child-to-staff ratios, and by process
measures, such as observations of interaction
between caregivers and children.80 Here, we
use evidence on differences between the
quality of care experienced by African American
and white children from a study by Margaret
Burchinal and Debby Cryer.81 One of
their data sources, the Cost, Quality, and
Outcomes (CQO) study, collected information
on the quality of center care received by
four-year-old children in four states (and thus
was not nationally representative). It in-
E a r l y C h i l d h o o d C a r e a n d E d u c a t i o n : E f f e c t s o n E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 181
cluded four different measures of quality of
care, assessing teacher?s interactions and responsiveness
to children as well as the extent
to which the program was child centered
(rather than didactic). Across all measures,
white children on average experienced
higher-quality care than black children, but
the differences were less pronounced for
caregivers? responsiveness and sensitivity
than for other measures. The difference on a
summary measure of quality, which combined
these indicators, was about 0.3 of a
standard deviation.82
Burchinal and Cryer present results from
similar analyses for three-year-olds from the
NICHD Study of Early Child Care, which
followed a large (but not nationally representative)
sample of children born in 1991. In
contrast to the CQO study, this research included
children in all types of care and education
settings, not only center care. Consequently,
differences in the quality of care
may be caused not only by variations in quality
within types of care, but also by the different
distribution of children across types of
care. The measure used by the NICHD
study, the Observational Record of the Caregiving
Environment (ORCE), captures the
quality of caregiver interactions with children,
including their sensitivity and responsiveness.
Again, black children experienced
lower-quality care than white children; the
gap was even larger than in the CQO study,
at more than 0.7 of a standard deviation.
Taken together, these studies suggest that
black children may receive lower-quality care
than white children, both within centers and
across other types of care.
How Much Do Differences in
Early Childhood Care and
Education Matter for Racial
and Ethnic Gaps in Readiness?
To sum up, racial and ethnic differences exist
both in enrollment in early childhood care and
education and in the quality of care received.
Black children are more likely than white children
to be enrolled in some form of preschool,
although almost 20 percent of these are Head
Start programs. Black children also may attend
lower-quality preschool programs than
their white peers. Although Hispanic children
are much less likely than white children to be
in preschool, they are also more likely than
white children to be in Head Start. If Head
Start programs are of lower quality or less academic
in focus than other types of preschools,
the high rates of black and Hispanic enrollment
in Head Start may mean that these programs
are doing less than they might to alleviate
early achievement gaps.83
How might early childhood care and education
policies narrow racial and ethnic
achievement gaps at school entry? First,
funds might be targeted to promote the enrollment
of racial and ethnic minority children
in center care or preschool. Given the
current low enrollment of Hispanic children
relative to white children, such initiatives
could be particularly effective in closing Hispanic-
white school readiness gaps. Second,
Katherine A. Magnuson and Jane Wa l d f o g e l
182 THE FUTURE OF CHILDREN
Across all measures,
white children on average
experienced higher-quality
care than black children,
but the differences were less
pronounced for caregivers?
responsiveness and sensitivity
than for other measures.
additional funds might be used to increase
the quality of the preschools that black and
Hispanic children attend (including Head
Start programs).84 The magnitude of effects
will depend on how much quality is improved
and on the number of children affected.
How much might such changes in enrollment
and quality narrow racial and ethnic test
score gaps? We conducted some back-of-theenvelope
estimates that, although rough,
allow us to place some bounds on the likely
share of the school readiness gaps that could
be closed by changing current patterns of
preschool enrollment or quality. We assume
at the outset that the role of incremental
changes in early child care and education is
likely to be limited, given the many other influences
on the school readiness gaps (documented
in the other articles in this volume).
We do not attempt to identify specific policies
that might increase center care enrollment
or quality or to model the effects of
specific policies. Rather, we demonstrate
how changes in early childhood care and education
might narrow racial and ethnic gaps in
school readiness.
Increasing Enrollment
We begin by considering the potential effect,
by race and ethnicity, of five different
changes in enrollment (table 2). Each scenario
involves boosting the enrollment in
preschool of three- to five-year-olds who are
not now in Head Start, prekindergarten, or
any other form of preschool. Clearly, the size
of the benefit from increases in enrollment
depends on how much preschool improves
children?s school readiness. For each scenario,
we draw on the most reliable research
to give three different estimates of preschool
effects on children?s reading scores at school
entry: 0.15, 0.25, and 0.65 of a standard
deviation.85
In the first scenario, Hispanic children?s enrollment
rises from 40 percent to 60 percent
to match that of white children. Depending
on the size of the preschool effect, this scenario
could narrow the Hispanic-white reading
gap at school entry by 0.03 to 0.13 of a
standard deviation. Given that the average
Hispanic-white gap in reading at school entry
is about 0.50 of a standard deviation, this
amounts to closing between 6 percent and 26
percent of the gap.86 (Although we use the
estimate of 0.50 of a standard deviation
throughout the remainder of our discussion,
it is important to recognize that these figures
will overstate the percentage reductions if
racial and ethnic school readiness gaps are in
fact larger.) In the second scenario, both Hispanic
and black children?s preschool enrollment
rates increase to 80 percent, 20 percentage
points above that of white children.
Such changes would narrow the black-white
gap by 0.02 to 0.10 of a standard deviation
(about 4 percent to 20 percent of the gap)
and the Hispanic-white gap by 0.06 to 0.26 of
a standard deviation (about 12 percent to 52
percent of the gap), again depending on how
much children benefit from preschool.
Although both of these scenarios reduce
school readiness gaps, particularly that between
Hispanic and white children, it may be
difficult to implement race- or ethnicityspecific
policies. For this reason, we also consider
the effect of increases in preschool enrollments
across all racial and ethnic groups.
In the third scenario, the enrollment of all
children living in poverty rises to 100 percent;
in the fourth scenario, enrollment for all lowincome
children (under 200 percent of the
poverty threshold) rises to 100 percent; and in
the fifth scenario, enrollment is universal
without regard to income. Initiatives that
boost preschool enrollment without regard to
racial or ethnic backgrounds (scenarios 3 to 5)
E a r l y C h i l d h o o d C a r e a n d E d u c a t i o n : E f f e c t s o n E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 183
would be less effective at closing racial and
ethnic school readiness gaps than the more
targeted initiatives (scenarios 1 and 2). In scenarios
3 to 5, the Hispanic-white gap would
fall by between 0.02 and 0.17 of a standard
deviation; but the black-white gap might either
slightly increase (by up to 0.02 of a standard
deviation) or slightly decrease (by up to
0.06 of a standard deviation).
Although boosting Hispanic or black preschool
enrollment rates beyond that of white
children would be the most effective means
of closing racial and ethnic gaps, the universal
programs may offer benefits that our estimates
do not capture. For example, if universal
programs are of higher quality or if
children benefit from attending preschools
with peers of diverse socioeconomic backgrounds,
then our estimates may be too low.87
Improving Quality
What about improving the quality of center
care that black and Hispanic children re-
Katherine A. Magnuson and Jane Wa l d f o g e l
184 THE FUTURE OF CHILDREN
Table 2. Effects on Reading Scores at School Entry of Increasing Preschool Enrollment
for Children Aged Three to Five Who Are Not in Head Start or Preschool
Standard deviation
Increase in population average Decrease in gap
Scenario Preschool effect Blacks Hispanics Whites Black-white Hispanic-white
1. Boost Hispanic enrollment to the level of
white enrollment (from 40% to 60%) .15 - .03 0 - .03
.25 - .05 0 - .05
.65 - .13 0 - .13
2. Increase Hispanic and black enrollment
to 80%, no change in white enrollment (60%) .15 .02 .06 0 .02 .06
.25 .04 .10 0 .04 .10
.65 .10 .26 0 .10 .26
3. Preschool for all children below 100% of
poverty; full enrollment .15 .02 .03 .01 .01 .02
.25 .04 .05 .02 .02 .03
.65 .10 .12 .04 .06 .08
4. Preschool for all children below 200% of
poverty; full enrollment .15 .03 .06 .02 .01 .04
.25 .06 .09 .03 .03 .06
.65 .14 .25 .08 .06 .17
5. Preschool for all children; full enrollment .15 .05 .10 .06 ?.01 .03
.25 .10 .14 .10 0 .04
.65 .24 .38 .26 ?.02 .12
Sources and notes: Estimates of the percentage of children in preschool are taken from National Center for Educational Statistics, The Condition
of Education 2002 (U.S. Department of Education, Office of Educational Research and Improvement, 2000). National rates of preschool
attendance among all children, by race and ethnicity, are as follows: white, 59 percent; black, 63 percent; Hispanic, 40 percent. For
poor children, the corresponding estimates are white, 46 percent; black, 60 percent; Hispanic 36 percent. For nonpoor children, the estimates
are white, 60 percent; black, 66 percent; Hispanic, 42 percent.
Poverty rates were taken from the National Center for Children in Poverty. Estimates are based on the following poverty rates for 2002: children
below 100 percent of poverty line: whites, 13 percent; blacks, 38 percent; Hispanics, 30 percent (Child Trends Database, ?Children in
Poverty,? www.childtrendsdatabank.org/indicators/4Poverty.cfm [July 20, 2004]). Children below 200 percent of poverty line: whites, 25
percent; blacks, 58 percent; Hispanics, 62 percent (National Center for Children in Poverty, ?Low-Income Children in the United States,
2004,? www.nccp.org/pub_cpf04.html [July 20, 2004]).
ceive?88 We answer this question, again, by
considering the effect of several different
scenarios for quality improvement (see table
3). And, again, because these estimates will
be sensitive to the extent to which quality influences
children?s outcomes, we provide a
range of estimates, reflecting the incremental
effects of increased preschool quality on children?s
reading skills of 0.1, 0.2, or 0.3 of a
standard deviation. However, we note that to
bring about such large increases in children?s
outcomes would involve large increases in
the process and structural measures of quality,
in some cases over a full standard deviation
increase in the quality of care.89
The first scenario involves raising the quality
of Head Start programs. Depending on the
size of the increased quality effects, this scenario
would reduce the black-white school
readiness gap by 0.02 to 0.05 of a standard
deviation (4 percent to 10 percent of the gap)
and narrow the Hispanic-white gap by 0.02 to
0.04 of a standard deviation (4 percent to 8
percent of the gap). The second scenario entails
raising the quality of all preschool programs
(including Head Start) for currently
enrolled children. It would improve the
achievement of black children somewhat
more than scenario 1 because they have the
highest rates of enrollment in center care.
But reductions in black-white gaps would still
be fairly modest, ranging from 0 to 0.07 of a
standard deviation, depending on whether
the quality increase were universal (scenario
4) or targeted to low-income children (sce-
E a r l y C h i l d h o o d C a r e a n d E d u c a t i o n : E f f e c t s o n E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 185
Table 3. Effects on Reading Scores at School Entry of Improving Quality of Head Start
and Preschool Programs for Children Aged Three to Five
Standard deviation
Increase in population average Decrease in gap
Scenario Quality effect Blacks Hispanics Whites Black-white Hispanic-white
1. Increase quality of Head Start .1 .02 .02 0 .02 .02
.2 .04 .03 .01 .03 .02
.3 .06 .05 .01 .05 .04
2. Increase quality of Head Start and other
preschools for children below 100% of poverty .1 .02 .01 .01 .03 .00
.2 .05 .02 .01 .04 .01
.3 .07 .03 .02 .05 .01
3. Increase quality of Head Start and other
preschools for children below 200% of poverty .1 .04 .02 .01 .02 .01
.2 .07 .05 .03 .05 .02
.3 .11 .07 .04 .07 .03
4. Increase quality of Head Start and other
preschools for all children .1 .06 .04 .06 .0 ?.02
.2 .13 .08 .12 .01 ?.04
.3 .19 .12 .18 .01 ?.06
Notes: See sources and notes for table 2. Current levels of enrollment are assumed for all scenarios. Estimates of the number of children
served by Head Start for scenario 1 are taken from data published by the Head Start Bureau, but the numbers of children in Head Start and
preschool are taken from the National Household Education Survey (NHES), 1995. Thus it is not possible to compare directly scenarios 1
and 2, the effect of increasing the quality of Head Start and the effect of increasing the quality of all Head Start and preschools for poor
children. Although the NHES indicates that only 36 percent of poor Hispanic children are in center care, the numbers from the Head Start
Bureau suggest that 18 percent of all Hispanic children are in Head Start, and if Head Start primarily serves poor children this would imply
that close to 60 percent of poor Hispanic children were in Head Start.
narios 2 and 3). Because Hispanic children
are less likely to experience center care, raising
the quality of preschools without changing
current enrollment patterns would do little
to narrow the Hispanic-white gap and
could even increase it (scenario 4).
The estimates in table 3 lead us to conclude
that even large increases in the quality of
center care would have only a small effect on
the black-white school readiness gap and
even less of an effect on the Hispanic-white
gap. However, we note that raising the quality
of preschools attended only by black and
Hispanic children would result in slightly
larger reductions in school readiness gaps.
Increasing Quality and Enrollment
The estimates thus far have shown what
could result from initiatives that either increase
enrollment or increase quality. How
much more effective would initiatives be if
they attempted to do both? In table 4, we
show estimates for three different scenarios
that increase center care quality and enrollment
at the same time. As in table 3, for each
scenario we model the effects of a range of
quality improvements, again with increases in
center care and preschool effects ranging
from 0.1 to 0.3 of a standard deviation.
In the first scenario, preschool enrollment of
children in poverty becomes universal and
the quality of programs they attend increases.
We assume that before the increase in quality,
preschool raised children?s school readiness
by 0.25 of a standard deviation (our middle-
ground estimate from table 2); with the
quality improvement, preschool raises school
readiness by 0.35, 0.45, or 0.55 of a standard
deviation.90 Universal enrollment in higherquality
care of children in poverty would narrow
the black-white school readiness gap at
school entry by 0.05 to 0.10 of a standard deviation
(that is, 10 percent to 20 percent of
the gap) and would narrow the Hispanic-
Katherine A. Magnuson and Jane Wa l d f o g e l
186 THE FUTURE OF CHILDREN
Table 4. Effects on Reading Scores at School Entry of Improving the Quality of and
Increasing Enrollment in Head Start and Preschool for Children Aged Three to Five
Standard deviation
Increase in population average Decrease in gap
Scenario Quality effect Blacks Hispanics Whites Black-white Hispanic-white
1. Increase quality of Head Start and other
preschools for children below 100% poverty
with 100% enrollment .1 .08 .08 .03 .05 .05
.2 .11 .11 .04 .07 .07
.3 .15 .14 .05 .10 .09
2. Increase quality of Head Start and other
preschools for children below 200% poverty
with 100% enrollment .1 .11 .16 .05 .06 .10
.2 .17 .22 .08 .09 .14
.3 .23 .28 .10 .12 .18
3. Increase quality of Head Start and other
preschools for all children with 100% enrollment .1 .19 .25 .20 ?.01 .05
.2 .29 .35 .30 ?.01 .05
.3 .39 .45 .40 ?.01 .05
Notes: See sources and notes for tables 2 and 3. All scenarios assume 100 percent enrollment and an effect of 0.25 before increase in
quality.
white gap by 0.05 to 0.09 of a standard deviation
(10 percent to 18 percent of the gap). In
the second scenario, enrollment in preschool
becomes universal for children from families
with household incomes below 200 percent
of the poverty threshold. Such a change
would narrow the black-white school readiness
gap by 12 percent to 24 percent, and the
Hispanic-white gap by 20 percent to 36 percent.
The third scenario, universal enrollment
and higher-quality care for all children
regardless of family income, would do little
to close racial and ethnic gaps, primarily because
white children would also benefit from
this change.
As table 4 shows, initiatives that substantially
raise both enrollment in and the quality of
center care for low-income children could
narrow racial and ethnic school readiness
gaps considerably, reducing black-white gaps
by up to 24 percent and Hispanic-white gaps
by up to 36 percent. In addition, table 2 indicates
that race- or ethnicity-specific increases
in enrollment?in particular, increasing the
enrollment of Hispanic children but not that
of white children?could also narrow school
readiness gaps. Other changes would also improve
black and Hispanic children?s school
readiness, but would not reduce racial and
ethnic gaps much, because they would also
improve white children?s achievement. If
raising black and Hispanic children?s school
readiness regardless of their relative levels of
achievement is a goal, then these changes
should be considered.
Implications for Policy
We draw two conclusions about the role of
early childhood care and education in closing
racial and ethnic gaps in readiness at school
entry. First, public funding of early education
programs is probably already reducing ethnic
and racial gaps. Large shares of Hispanic and
black children are attending Head Start; as an
upper bound, we estimate that the blackwhite
test score gap at school entry might be
as much as 24 percent larger in the absence of
Head Start. Yet questions remain about the
extent to which Head Start provides lasting
academic benefits for children, particularly of
differing ethnic and racial backgrounds, making
conclusions about Head Start?s role in reducing
test score gaps speculative.
Second, the effects of incremental increases
in enrollment or improvements in quality will
depend on the specific changes adopted. For
example, boosting the enrollment of Hispanic
children in center care to meet or exceed
the enrollment of white children would
raise their test scores at school entry and narrow
the gap between their scores and those
of non-Hispanic white children. The overall
effect could be quite large (because the gap
in enrollment between Hispanic and white
children is fairly large), but would depend on
the quality of the preschools. Thus, our
analysis affirms the wisdom of policies that
specifically boost the enrollment of Hispanic
children, starting at age three, for example,
by funding early education programs in Hispanic
neighborhoods.
Likewise, improving the quality of center
care would modestly boost children?s test
scores. Such improvements in quality would
do more to close black-white school readiness
gaps than Hispanic-white gaps, because
more black children are now enrolled than
Hispanic children. Yet these effects would be
fairly small for both groups, because quality
improvements would also benefit white children
attending preschool.
What about simultaneous increases in children?s
preschool enrollment and quality?
Universal enrollment in higher-quality center
E a r l y C h i l d h o o d C a r e a n d E d u c a t i o n : E f f e c t s o n E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 187
care or preschools for low-income children
could close a substantial portion of school
readiness gaps based on race and ethnicity,
narrowing the black-white reading gap at
school entry as much as 24 percent and the
Hispanic-white reading gap as much as 36
percent. Such findings point to the potential
for policies that raise enrollment in Head
Start, prekindergarten, and other preschool
programs for children in and near poverty,
while substantially improving the quality of
these programs.
In keeping with the focus of this issue, and
given data limitations, in this article we have
concentrated mainly on test scores as a measure
of school readiness. But school readiness
encompasses many aspects of development
in addition to academic skills, including
health, social skills, positive and problem behaviors,
and motivation to learn.91 As noted,
early childhood care and education programs
may affect these other aspects of school
readiness, positively or negatively, and such
effects should also be taken into account.92
Finally, we need to keep in mind that the
benefits even of the best early childhood programs
tend to fade over time. Preschool programs
may need to be followed up with interventions
for school-age children, as in the
successful Chicago CPC program.93 As others
have observed, it is not realistic to expect
a preschool program, however effective, to
?inoculate? a child for life against the risk of
low academic achievement.94 But we can and
should expect such programs to help narrow
racial and ethnic differentials in young children?s
academic skills, so that they enter
school on a more even footing.
Katherine A. Magnuson and Jane Wa l d f o g e l
188 THE FUTURE OF CHILDREN
Endnotes
1. NICHD Early Child Care Research Network, ?Early Child Care and Children?s Development prior to
School Entry: Results from NICHD Study of Early Child Care,? American Educational Research Journal
39 (2002): 133?64.-
2. Few prior studies have explicitly considered these questions. For general discussions, see David Grissmer,
Ann Flanagan, and Stephanie Williamson, ?Why Did the Black-White Score Gap Narrow in the 1970s and
1980s?? in The Black-White Test Score Gap, edited by Christopher Jencks and Meredith Phillips (Brookings,
1998), pp. 182?226; Marcia K. Meyers and others, ?Inequality in Early Childhood Education and
Care: What Do We Know?? in Social Inequality, edited by Kathryn M. Neckerman (New York: Russell
Sage Foundation, 2004).
3. Cynthia T. Garcia Coll and others, ?An Integrative Model for the Study of Developmental Competencies in
Minority Children,? Child Development 67 (1996): 1891?914. The racial and ethnic categories and terms
we use in this article reflect the terminology and categorizations used in the bulk of studies we review.
4. Alejandro Portes and Ruben Rumbaut, Ethnicities: Children of Immigrants in America (New York: Russell
Sage Foundation, 2003). Unfortunately, few studies distinguish among Hispanic children based on characteristics
such as immigration status or language ability, so this review is unable to make these important
distinctions.
5. Garcia Coll and others, ?An Integrative Model? (see note 3).
6. Child care subsidy programs reach only a small share of eligible children: in 1998, only about 15 percent of
eligible low-income families.
7. Forum on Childhood Family Statistics, ?America?s Children in Brief: Key National Indicators of Wellbeing.?
Accessed from http://childstats.gov, on November 3, 2004.
8. Gina Adams and Monica Rohacek, ?More than a Work Support? Issues around Integrating Child Development
Goals into the Child Care Subsidy System,? Early Childhood Research Quarterly 17 (2002): 418?40;
Suzanne W. Helburn and Barbara Bergmann, America?s Child Care Problem (New York: St. Martin?s Press,
2002).
9. NICHD Early Child Care Research Network, ?Child Care Structure ?Process ?Outcome: Direct and Indirect
Effects of Child-Care Quality on Young Children?s Development,? Psychological Science 13 (2002):
199?206; NICHD Early Child Care Research Network and Greg J. Duncan, ?Modeling the Impacts of Child
Care Quality on Children?s Preschool Cognitive Development,? Child Development 74 (2003): 1454?75.
10. Nicholas Zill and others, Head Start FACES: A Whole Child Perspective on Program Performance (U.S.
Department for Health and Human Services, Administration for Children and Families, 2003).
11. Helburn and Bergmann, America?s Child Care Problem (see note 8); Eugene Smolensky and Jennifer A.
Gootman, eds., Working Families and Growing Kids: Caring for Children and Adolescents (Washington:
National Academy Press, 2003).
12. NICHD Early Child Care Research Network, ?Child Care Structure ?rocess ?Outcome? (see note 9);
Elizabeth Votruba-Drzal, Rebecca L. Coley, and P. Lindsay Chase-Lansdale, ?Child Care and Low-Income
Children?s Development: Direct and Moderated Effects,? Child Development 75 (2004): 296?312.
E a r l y C h i l d h o o d C a r e a n d E d u c a t i o n : E f f e c t s o n E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 189
13. See recent reviews in David Blau, The Child Care Problem (New York: Russell Sage Foundation, 2001);
Helburn and Bergmann, America?s Child Care Problem (see note 8); Smolensky and Gootman, Working
Families and Growing Kids (see note 11).
14. Suzanne W. Helburn, ed., Cost, Quality, and Child Outcomes in Child Care Centers: Technical Report
(University of Colorado at Denver, Department of Economics, Center for Research in Economic and Social
Policy, 1995).
15. NICHD Early Child Care Research Network, ?Early Child Care and Children?s Development prior to
School Entry? (see note 1); NICHD Early Child Care Research Network, ?Child Care Structure ?
Process ?Outcome? (see note 9).
16. In 2003, 12.5 percent of children in Head Start programs had disabilities.
17. A small number of children under the age of three are served by the Early Head Start program, which
began in 1995.
18. These estimates are based on the number of funded Head Start slots and the U.S. poverty rates; see Janet
Currie and Mathew Neidell, ?Getting Inside the ?Black Box? of Head Start Quality: What Matters and
What Doesn?t,? Working Paper 10091 (Cambridge, Mass.: National Bureau of Economic Research, 2003).
19. Zill and others, Head Start FACES (see note 10).
20. Carol H. Ripple and others, ?Will Fifty Cooks Spoil the Broth?? American Psychologist 54 (1999): 327?43;
Edward Zigler and Sally. J. Styfco, ?Head Start: Criticisms in a Constructive Context,? American Psychologist
49 (1994): 127?32.
21. Private schools also offer such programs for a fee, but to simplify our discussions we use ?prekindergarten?
to refer to publicly funded programs.
22. Ripple and others, ?Will Fifty Cooks Spoil the Broth?? (see note 20); Karen Schulman, Helen Blank, and
Danielle Ewen, Seeds of Success: State Prekindergarten Initiatives 1998?1999 (Washington: Children?s Defense
Fund, 1999).
23. ?Quality Counts 2002: Building Blocks for Success,? Education Week 21, no. 17 (January 10, 2002)
(www.edweek.org/sreports/qc02/templates/article.cfm?slug=17exec.h21[June 27, 2003]).
24. ?Quality Counts 2002? (see note 23); Schulman, Blank, and Ewen, Seeds of Success (see note 22); Ripple
and others, ?Will Fifty Cooks Spoil the Broth?? (see note 20).
25. William Gilliam and Edward Zigler, ?State Efforts to Evaluate the Effects of Prekindergarten,? mimeo,
Yale University, 2004; Timothy Smith and others, Prekindergarten in U.S. Public Schools (U.S. Department
of Education, National Center for Education Statistics, 2003).
26. Ripple and others, ?Will Fifty Cooks Spoil the Broth?? (see note 20).
27. Smith and others, Prekindergarten in U.S. Public Schools (see note 25).
28. Blau, The Child Care Problem (see note 13).
29. Dan Bellm and others, Inside the Pre-K Classroom: A Study of Staffing and Stability in State-Funded
Prekindergarten Programs (Washington: Center for the Child Care Workforce, 2002).
Katherine A. Magnuson and Jane Wa l d f o g e l
190 THE FUTURE OF CHILDREN
30. Gary Henry and others, Report of the Findings from the Early Childhood Study (Andrew Young School of
Policy Studies, Georgia State University, 2003).
31. Cynthia Esposito Lamy and others, ?Inch by Inch, Row by Row, Gonna Make This Garden Grow: Classroom
Quality and Language Skills in the Abott Preschool Program,? Working Paper (National Institute for
Early Education Research, Rutgers University, 2004).
32. W. Steven Barnett, ?Long-Term Effects of Early Childhood Programs on Cognitive and School Outcomes,?
The Future of Children 5, no. 3 (1995): 25?50; Blau, The Child Care Problem (see note 13).
33. Barnett, ?Long-Term Effects of Early Childhood Programs on Cognitive and School Outcomes? (see note
32); Janet Currie, ?Early Childhood Intervention Programs: What Do We Know?? Journal of Economic
Perspectives 15 (2001): 213?38; Lynn A. Karoly and others, Investing in Our Children: What We Do and
Don?t Know about the Costs and Benefits of Early Childhood Interventions (Santa Monica, Calif.: RAND,
1998); Jane Waldfogel, ?Child Care, Women?s Employment and Child Outcomes,? Journal of Population
Economics 15 (2002): 527?48.
34. Cecilia McCarton and others, ?Results at Age 8 Years of Early Intervention for Low-Birth-Weight Premature
Infants: The Infant Health and Development Program,? Journal of the American Medical Association
277 (1997): 126?32.
35. Jeanne Brooks-Gunn and others, ?Enhancing the Cognitive Outcomes of Low Birth Weight, Premature Infants:
For Whom Is Intervention Most Effective?? Pediatrics 89 (992): 1209?15.
36. Frances A. Campbell and Craig T. Ramey, ?Cognitive and School Outcomes for High-Risk African American
Students at Middle Adolescence: Positive Effects of Early Intervention,? American Educational Research
Journal 32 (1995): 743?72; Craig T. Ramey and others, Early Learning, Later Success: The
Abecedarian Study (Chapel Hill, N.C.: Frank Porter Graham Child Development Institute, 1999).
37. Barnett, ?Long-Term Effects of Early Childhood Programs on Cognitive and School Outcomes? (see note
32); Karoly and others, Investing in Our Children (see note 33); Waldfogel, ?Child Care, Women?s Employment
and Child Outcomes? (see note 33).
38. Lawrence J. Schweinhart, Helen V. Barnes, and David P. Weikart, Significant Benefits of the High-Scope
Perry Preschool Study through Age 27 (Ypsilanti, Mich.: High-Scope Press, 1993).
39. McCarton and others, ?Results at Age 8 Years of Early Intervention for Low-Birth-Weight Premature Infants?
(see note 34).
40. Hiro Yoshikawa, ?Prevention as Cumulative Protection: Effects of Early Family Support and Education as
Chronic Delinquency and Its Risk,? Psychological Bulletin 115 (1994): 28?54.
41. Michael Foster, ?Trends in Multiple and Overlapping Disadvantages among Head Start Enrollees,? Children
and Youth Services Review 24 (2002): 933?54.
42. Ron Haskins, ?Beyond Metaphor: The Efficacy of Early Childhood Education,? American Psychologist 44
(1989): 274?82; Ruth H. McKey, The Impact of Head Start on Children (Department of Health and
Human Services: Government Printing Office, 1985).
43. Valerie Lee and others, ?Are Head Start Effects Sustained? A Longitudinal Follow-Up of Disadvantaged
Children Attending Head Start, No Preschool, and Other Preschool Programs,? Child Development 61
(1990): 495?507.
E a r l y C h i l d h o o d C a r e a n d E d u c a t i o n : E f f e c t s o n E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 191
44. Janet Currie and Duncan Thomas, ?Does Head Start Make a Difference?? American Economic Review 85
(1995): 341?64; Janet Currie and Duncan Thomas, ?Does Head Start Help Hispanic Children?? Journal of
Public Economics 74 (1999): 235?62; Eliana Garces, Duncan Thomas, and Janet Currie, ?Longer-Term Effects
of Head Start,? American Economic Review 92 (2002): 999?1012. For an exception, see Alison Aughinbaugh,
?Does Head Start Yield Long-Term Benefits?? Journal of Human Resources 36 (2001): 641?65.
45. Currie and Thomas, ?Does Head Start Make a Difference?? (see note 44). But see also Garces, Thomas,
and Currie, ?Longer-Term Effects of Head Start? (see note 44), who find some evidence that Head Start
might reduce crime among black children; and Lee and others, ?Are Head Start Effects Sustained?? (see
note 43), who find larger effects of Head Start on academic and social outcomes for black children than
white children.
46. Currie and Thomas, ?Does Head Start Help Hispanic Children?? (see note 44).
47. Gilliam and Zigler, ?State Efforts to Evaluate the Effects of Prekindergarten? (see note 25).
48. Arthur J. Reynolds, ?Effects of a Preschool Follow-On Intervention for Children at Risk,? Developmental
Psychology 30 (1994): 787?804.
49. Arthur J. Reynolds, ?One Year of Preschool or Two: Does It Matter?? Early Childhood Research Quarterly
10 (1995): 1?31.
50. Reynolds, ?Effects of a Preschool Follow-On Intervention for Children at Risk? (see note 48); Arthur J.
Reynolds and others, ?Long-Term Effects of an Early Childhood Intervention on Educational Achievement
and Juvenile Arrest: A Fifteen Year Follow-Up of Low-Income Children in Public Schools,? Journal
of the American Medical Association 285 (2001): 2339?46.
51. William Gormley and Ted Gayer, ?Promoting School Readiness in Oklahoma: An Evaluation of Tulsa?s Pre-
K Program,? mimeo, Georgetown University, 2003.
52. Henry and others, Report of the Findings from the Early Childhood Study (see note 30).
53. Katherine A. Magnuson and others, ?Inequality in Preschool Education and School Readiness,? American
Educational Research Journal 41 (2004): 115?57.
54. Gilliam and Zigler, ?State Efforts to Evaluate the Effects of Prekindergarten? (see note 25).
55. Katherine A. Magnuson, Christopher Ruhm, and Jane Waldfogel. ?Does Prekindergarten Improve School
Preparation and Performance?? Working Paper 10452 (Cambridge, Mass.: National Bureau of Economic
Research, 2004).
56. Gilliam and Zigler, ?State Efforts to Evaluate the Effects of Prekindergarten? (see note 25).
57. Barnett, ?Long-Term Effects of Early Childhood Programs on Cognitive and School Outcomes? (see note
32); Meyers and others, ?Inequality in Early Childhood Education and Care? (see note 2); Smolensky and
Gootman, Working Families and Growing Kids (see note 11).
58. Deborah Phillips and Gina Adams, ?Child Care and Our Youngest Children,? Future of Children 11, no. 1
(2001): 35?51.
59. This study included Head Start and prekindergarten as center-based care.
Katherine A. Magnuson and Jane Wa l d f o g e l
192 THE FUTURE OF CHILDREN
60. NICHD Early Child Care Research Network and Duncan, ?Modeling the Impacts of Child Care Quality
on Children?s Preschool Cognitive Development? (see note 9).
61. Magnuson and others, ?Inequality in Preschool Education and School Readiness? (see note 53). Unfortunately,
data were not available on the quality of care children received so we were unable to explore its influence
on school readiness.
62. Jay Belsky, ?Developmental Risks (Still) Associated with Early Child Care,? Emanuel Miller Lecture, Journal
of Child Psychology and Psychiatry and Allied Disciplines 42, no. 7 (2001): 845?59; NICHD Early
Child Care Research Network, ?Does Amount of Time Spent in Child Care Predict Socioemotional Adjustment
during the Transition to Kindergarten?? Child Development 74 (2003): 976?1005.
63. NICHD Early Child Care Research Network, ?Does Amount of Time Spent in Child Care Predict Socioemotional
Adjustment during the Transition to Kindergarten?? (see note 62).
64. Authors? calculation of 1999 SIPP data.
65. NICHD Early Child Care Research Network, ?Early Child Care and Children?s Development prior to
School Entry? (see note 1); Phillips and Adams, ?Child Care and Our Youngest Children? (see note 58).
66. These estimates and those that follow are from our analyses of the ECLS-K. For further details, see Magnuson
and others, ?Inequality in Preschool Education and School Readiness? (see note 53).
67. The October CPS began collecting data in 1964, but the microdata for 1964?67 are not readily available.
The 2000 data were the most current available at the time the analysis was conducted.
68. From 1968 to 1984, the survey asked: ?Is [name] attending or enrolled in school?? In 1985, the question
was changed to read: ?Is [name] attending or enrolled in regular school?? Then, in 1994, a prompt was
added after the question, so that the full question now reads: ?Is [name] attending or enrolled in regular
school? (Regular school includes nursery school, kindergarten, or elementary school and schooling which
leads to a high school diploma).? The October CPS and the National Household Education Survey find a
similar share of three- to five-year-old children enrolled in preprimary school programs (for instance, both
surveys find 68 percent in 1999). In contrast, two major child care surveys, the National Survey of American
Families (NSAF) and the Survey of Income and Program Participation, find a lower share of three- to
five-year-olds enrolled in center- or school-based programs; this is likely because these two surveys do not
ask explicitly about school programs and also because they interview some families during the summer
months, when such programs would be closed.
69. These figures chart the preschool enrollment of three- and four-year-olds, using data from the October
Current Population Survey. See note 68 for more details.
70. Table 1 uses SIPP data. Rates of center care enrollment are lower than in the CPS, because infants and
toddlers are less likely to experience nonparental care and more likely to experience informal child care
than children aged three and four. Estimates of child care arrangements from various data sources should
be compared with caution, because of differences in question wording, timing of data collection, and coding
categories (see also note 68).
71. Office of the Assistant Secretary for Planning and Evaluation, Trends in the Well-Being of America?s Children
and Youth: 2002 (U.S. Department of Health and Human Services, 2002).
E a r l y C h i l d h o o d C a r e a n d E d u c a t i o n : E f f e c t s o n E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 193
72. It is beyond the scope of this article to consider whether racial and ethnic differences in parental employment
have affected differentials in test scores, above and beyond any effects that may work through early
childhood care and education. Parental employment could affect child development through several pathways,
such as economic resources, parenting, and the home environment. For recent reviews on the effects
of parental employment on child development, and discussion of how they may vary by racial and ethnic
group, see Jack P. Shonkoff and Deborah Phillips, eds., From Neurons to Neighborhoods: The Science of
Early Childhood Development (Washington: National Academy Press, 2000); Smolensky and Gootman,
Working Families and Growing Kids (see note 11).
73. Authors? estimation of enrollment rates, using data from the Head Start Bureau and the 2000 decennial
census. The calculation is based on the number of Head Start slots available, not the number of children
served, which is larger because of turnover. Consequently, our estimates likely understate the number of
children who have ever participated in Head Start.
74. These estimates are derived by subtracting each group?s rate of attendance in Head Start from its October
CPS rates of preschool attendance. For example, because 19 percent of black four-year-olds are in Head
Start and CPS data indicate that 72 percent of black four-year-olds are in preschool, without access to Head
Start (and without enrollment in other programs), their enrollment rate would be 53 percent. By comparison,
only 5 percent of white four-year-olds are in Head Start, so without access to Head Start (and without
enrollment in other programs) their enrollment rate would fall from 67 percent to 62 percent, resulting in
a black-white enrollment gap of 9 percent (compared with black children?s current enrollment advantage of
5 percent).
75. Reynolds, ?Effects of a Preschool Follow-On Intervention for Children at Risk? (see note 48).
76. Per pupil expenditures in CPC and Head Start were comparable (in the early 1990s), and both programs
emphasize parental involvement and deliver comprehensive services. See Reynolds, ?One Year of
Preschool or Two? (see note 49). It is also important to keep in mind that Reynolds?s estimated effect sizes
are considerably larger than estimates derived from Head Start studies. For example, Lee and others, ?Are
Head Start Effects Sustained?? (see note 43) find Head Start effects of 0.25 on verbal skills at school entry;
Currie and Thomas, ?Does Head Start Make a Difference?? (see note 44) find almost no lasting effects
from Head Start on the academic skills of black children.
77. Reynolds, ?One Year of Preschool or Two? (see note 49).
78. Smith and others, Prekindergarten in U.S. Public Schools (see note 25).
79. Calculations from the 1999 National Survey of American Families by the Urban Institute found the average
numbers of hours spent in nonparental care by children under age three with employed mothers were
as follows: twenty-two for white non-Hispanic children, thirty-two for black children, and twenty-one for
Hispanic children. See Jennifer Erhle, Gina Adams, and Kathryn Tout, ?Who?s Caring for Our Youngest
Children? Child Care Patterns of Infants and Toddlers,? Occasional Paper 42 (Washington: Urban Institute,
2001). Data from the National Household Education Survey in 1995 found the average numbers of
hours spent in center-based nonparental care by children under age six were as follows: twenty-eight for
white non-Hispanic children, thirty-six for black children, and thirty-one for Hispanic children. See National
Center for Educational Statistics, Digest of Education Statistics, 2002 (Washington, 2003), table 44.
80. Smolensky and Gootman, Working Families and Growing Kids (see note 11).
Katherine A. Magnuson and Jane Wa l d f o g e l
194 THE FUTURE OF CHILDREN
81. Margaret R. Burchinal and Debby Cryer, ?Diversity, Child Care Quality, and Developmental Outcomes,?
Early Childhood Research Quarterly 18 (2003): 401?26. Burchinal and Cryer also provide information on
the quality of care received by Hispanic children, but sample sizes for this group are so small that we do not
include them in our summary. In addition, they examine whether the measures of quality of care had
equivalent effects across ethnic and racial groups They conclude that these measures are equally reliable
across groups and that higher quality care was linked to higher levels of cognitive and social skills among all
groups. For a discussion of racial and ethnic differences in measuring the quality of child care, see Deborah
L. Johnson and others, ?Studying the Effects of Early Child Care Experiences on the Development of
Children of Color in the United States: Toward a More Inclusive Research Agenda,? Child Development 74
(2003): 1227?44.
82. Black children in the CQO sample were much more likely than white children to be poor (30 percent versus
6 percent) or working poor (32 percent versus 11 percent). See Burchinal and Cryer, ?Diversity, Child
Care Quality, and Developmental Outcomes? (see note 81).
83. Zigler and Styfco, ?Head Start? (see note 20).
84. Although improving the quality of informal child care might reduce racial and ethnic gaps, it is much more
difficult for policies to influence the quality of informal care given by, for example, babysitters or grandparents,
than to improve formal child care.
85. The estimate of 0.15 is from Magnuson and others, ?Inequality in Preschool Education and School Readiness?
(see note 53). The estimate of 0.25 is from NICHD and Duncan, ?Modeling the Impacts of Child
Care Quality on Children?s Preschool Cognitive Development? (see note 9). The estimate of 0.65 is from
Reynolds, ?Effects of a Preschool Follow-on Intervention? (see note 48).
86. See Greg Duncan and Katherine Magnuson?s article in this volume.
87. Pam Sammons and others, Measuring the Impact of Pre-School on Children?s Cognitive Progress over the
Pre-School Period (Institute of Education, University of London, 2002).
88. We do not attempt to estimate how children may benefit from improvements in informal care, because it is
difficult to construct effective policies to this end.
89. Burchinal and Cryer, ?Diversity, Child Care Quality, and Developmental Outcomes? (see note 81);
NICHD Early Child Care Research Network, ?Early Child Care and Children?s Development prior to
School Entry? (see note 1).
90. If we estimate the effects of preschool or Head Start to be 0.65 (and increase the quality of care by 0.2), we
find that the population effects would be larger; for example, closing 0.10 of the black-white gap and 0.11
of the Hispanic-white gap (compared with 0.07 for both gaps) for children under 100 percent of the
poverty line. If we assume the estimated effects of preschool or Head Start care to be 0.15 (and increase
the quality of care by 0.20), we find the population effects are smaller, resulting in slightly smaller reductions
in the gap; for example, the black-white gap does not decrease, whereas the Hispanic gap would decrease
by 0.06 (compared with 0.07) for children below 100 percent of the poverty line.
91. Shonkoff and Phillips, From Neurons to Neighborhoods (see note 72).
92. Zigler and Styfco, ?Head Start? (see note 20).
E a r l y C h i l d h o o d C a r e a n d E d u c a t i o n : E f f e c t s o n E t h n i c a n d R a c i a l G a p s i n S c h o o l R e a d i n e s s
VOL. 15 / NO. 1 / SPRING 2005 195
93. Arthur J. Reynolds and Judy A. Temple, ?Extended Early Childhood Intervention and School Achievement:
Age Thirteen Findings from the Chicago Longitudinal Study,? Child Development 69 (1998):
231?46.
94. Jeanne Brooks-Gunn, ?Do You Believe in Magic? What Can We Expect from Early Childhood Intervention
Programs?? SRCD Social Policy Report 17, no. 1 (2003); Zigler and Styfco, ?Head Start? (see note 20).
Katherine A. Magnuson and Jane Wa l d f o g e l
196 THE FUTURE OF CHILDREN


#

tags