Protecting teens

R. W. Blum
January 1, 2000

Teenage drinking and tobacco usage, involvement in
violence, early and unprotected sex, and certainly, suicidal
thoughts and attempts can lead to early death as
well as health problems later in life. What factors are associated
with increased or decreased risks within each racial/
ethnic group? Are these factors the same across ethnic groups
and gender? Are there unique factors that increase
the risk in some groups but not others? Preliminary answers to
these questions are the focus of this report.
Beyond Race, Income and
Family Structure
Protecting Teens:
Introduction . . . . . . . . . . . . . . . . . . . .4
Interrelationships Among Race,
Income and Family Structure . . . . . . . .8
Individual Factors,
Friends and Family
Influence the Behavior of Teens . . . . . .22
Conclusion . . . . . . . . . . . . . . . . . . . .37
Acknowledgements . . . . . . . . . . . . . . .39
All photographs in this publication by John Nolter, Minneapolis, Minnesota
Preparation of this report was assisted by a grant
from the Robert Wood Johnson Foundation
Princeton, New Jersey
Introduction
There are many things that influence the health of an
adolescent. Race, ethnicity, income and family structure, for example,
represent contexts affecting the attitudes, beliefs and behaviors
of youth that belong to the group. This influence is seen in the consistent
differences between groups in the prevalence of adolescent
drinking, smoking, violence, suicide risk, and sexual activity.
It is, however, important to note
that there are large numbers of
youth in every “high-risk group”
who do not engage in problem
behaviors. Likewise, in groups viewed
as “low-risk,” a significant number of
youths participate in the same health
compromising behaviors. What factors
are associated with more or less involvement
with health compromising behaviors
within demographic groups? Are
some factors the same regardless of race
or gender? Are there unique factors that
increase or decrease risk in some groups
but not others? How might the answers
to these questions help shape policy and
interventions to promote the health and
well being of all youth in the United
States?
Whether it is tobacco use or violence,
pregnancy rates or suicide, analysis after
analysis has led policy makers to believe
that an adolescent’s race, income, family
structure, and gender predict the
4
likelihood of a teen participating in risky
behaviors. Perhaps unintentionally,
reports on the prevalence of risk
behaviors are often interpreted as
suggesting a direct relationship between
individuals’ behavior and his or her
ethnicity, income, or family structure.
Because such an approach is often
misleading, the American Academy of
Pediatrics recently recommended that
child health analyses go beyond
demographic descriptions to identify the
underlying social mechanisms that
account for poor health and health-risk
behaviors.
How important is it to address health
risk behaviors within low-risk groups as
well as high-risk groups? What is it
within each group that is associated with
increased or decreased risk behavior? Do
the same risk and protective factors apply
to all youth, regardless of gender or
ethnicity? Are there unique factors for
some groups? This monograph reports
on these issues in greater depth.
Understanding Risk and
Protective Factors
Risk factors are those aspects of the
teen’s life that are associated with an
increased likelihood of substance use,
early sex, violence, and other behaviors
that threaten his or her health and wellbeing.
Protective factors are those aspects
of the teen’s life that are associated with a
reduced risk of engaging in problem
behaviors. Risk and protective factors are
often mirror images of each other (for
5
The Add Health Survey, from which this data comes, is a comprehensive
school-based study of the health-related behaviors of adolescents in
the United States.Add Health surveys were conducted in two phases. In the
first phase, some 90,000 students in grades 7 through 12 attending 134
schools across the United States answered brief questionnaires about their
lives, including their health, friendship, self-esteem, and expectations for the
future. Before students could participate, parents had to give their permission
through procedures approved by each school.
In the first year of the study, administrators from the participating schools
also completed a questionnaire dealing with school policies and procedures,
teacher characteristics, health service provisions or referrals, as well as student
body characteristics. In the spring of 1996, school information was
updated in a telephone interview.
In the second phase, with the written consent of both the parent and the
adolescent, over 20,000 in-home interviews of students were conducted
between April and December of 1995 (Wave I).This “in-home” sample is
composed of both a nationally representative core sample (approximately
12,000) and a dozen special samples that could be used to examine questions
for groups that would otherwise be too small for analysis (for example,
twins, Cuban Hispanics, and disabled youth). No paper questionnaires
were used. Instead, all data were recorded on laptop computers with sensitive
questions asked privately using a pre-recorded audiocassette.A followup
in-home interview (Wave II) of 15,000 adolescents was conducted
between April and August of 1996.
A parent of each adolescent who was interviewed at home, usually the
mother, was asked to complete an interview as part of Wave I. Eighteen
thousand parent interviews were completed.
An additional survey, Phase Three, is planned for 2001.At that time, the
entire original sample group will be interviewed once again.
THE NATIONAL LONGITUDINAL STUDY OF
ADOLESCENT HEALTH (ADD HEALTH)
example, low selfesteem
is a risk
factor while high
self-esteem is
protective). They
may be causal or
merely “red flags”
indicating that a
youth is at higher or
lower risk for
specific problems.
Research indicates
that the relationship is often complex.
For example, while teens may choose
friends who do what they like to do,
they are also influenced by their friends’
behavior and vice versa. Whether we
view them as“red flags”, causal or some
of both, understanding risk and
protective factors can help us in
improving the lives of youth by
identifying where we might intervene.
Race and Ethnicity
Teens were asked about their
racial/ethnic identities in three ways.
They could select one or more racial
groups — such as White or Black — to
which they belonged. If they said they
were from two or more groups, for
example, White and Black, they were
then asked with which group they
primarily identified. A separate question
inquired whether they were Hispanic.
Based on these responses, teens were
then assigned to one of three major
racial/ethnic categories: White (non-
Hispanic), Hispanic (any race), and
Black (non-Hispanic). Due to
insufficient numbers in the nationally
representative sample, youth in other
racial/ethnic groups were not included in
these analyses.
Over two thirds (71.1%) of the final
sample identified themselves as White
(non-Hispanic); 12.6% identified
themselves as Hispanic and 16.3% said
they were Black.
6
Almost 11,000 White,
Black and Hispanic youth
from a nationally
representative sample;
Add Health Wave 1 data
(1995-6) based on in-home
interviews with teenagers
and their parents.
Data Used in this
Report
Analyses are based on the nationally representative “core sample” of
teens and their parents interviewed at their homes. It is precisely
because the survey is so representative of teens across the United States
that, despite its size, some of the numbers are small. For example, Native
Americans represent 1.2% of the population, and they represent 1.2% of the
Add Health teen sample, so only 120 Native American youth actually participated
in the study – too small a number to be analyzed separately.The same
is true for other ethnic minorities as well.The sample is so small for certain
populations that there is a risk of generalizing to a whole group across the
country based on only a handful of young people.Thus, we have limited our
analyses to Black, Hispanic and White youth.While there are a number of
special populations available for analysis in a larger Add Health sample, we
choose not to use them because of difficulties of generalizability.
WHO IS INCLUDED? WHO IS NOT?
7
Income
Annual household income was defined
as all sources of money, including public
assistance, and was based on parent
reports. Income ranges were divided into
six categories:
$10,000 or less (9.0%)
$11,000 to $20,000 (13.4%)
$21,000 to $30,000 (33.6%)
$31,000 to $40,000 (26.8%)
$41,000 to $60,000 (9.5%)
$61,000 or more (7.6%)
This report used the six-category
measure of income in all analyses because
it was easier to make an accurate estimate
where data were missing. The results
were the same as when a continuous
measure of income was used.
Family Structure
Just under one-third (31.1%) of
students reported that they had one
resident parent, while over two thirds
(68.9%) reported living in two-parent
families. Although youth living in intact
two-parent families are at lower risk than
youth living in step, adoptive, or foster
two-parent families, this report clusters
all two-parent families together. (When
we ran the data separating two parent
biologic families from others, we saw no
significant differences from how we
analyzed the data initially).
For instance, while nearly two
thirds (65.2%) of adolescents in
the lowest income group came
from single-parent families,
only 6.4% of those in the upper income
group came from single-parent homes. If
you live in a single-parent home, you are
much more likely to be poor. Likewise,
children of color from single-parent families
are much more likely to be poor
than their white counterparts.
Income and race are also closely
related: four out of five youths in upper
income groups were white, while more
than half of those with incomes under
$20,000 were teens of color.
Risk Behaviors
We explored the relationships between
four key demographic factors (race,
income, family structure and gender).
We then selected five adolescent healthrisk
behaviors that represent some of the
major threats to adolescent health and
well-being.
Analyses throughout this report were
conducted using the following measures.
All but sexual intercourse reflected
increasing degrees of involvement in
risky health behaviors.
cigarette use (12 levels);
alcohol use (6 levels);
suicide risk (5 levels);
violence involving weapons
(8 levels) and
sexual intercourse (yes/no);
8
Interrelationships Among Race,
Income and Family Structure
There is a high degree of interrelationship among race,
income and family structure.
43.6
0
10
20
30
40
50
60
White Black Hispanic
25.5
59.2
10.8
52.1
10.4
Percent of youth in single-parent families
Under $20,000/yr
Over $40,000/yr
Intercorrelation of Income & Family Structure
To simplify the presentation of the
findings in the first section of the report,
the health risk behaviors were divided
into two categories for each, as described
in the “cut-points” box. Income was
combined into three categories for the
same reason.
Prevalence of Problem Behaviors
The prevalence of every health risk
behavior varied to some degree,
depending on race/ethnicity. Overall:
White teens were more at risk
for smoking and drinking,
regardless of gender.
Black and Hispanic teens were
more at risk for weaponrelated
violence, although the
differences were much smaller
among males than females.
9
Measuring Health and Behaviors
Cigarette Use
Twelve levels of use defined by a combination of items on frequency and
number of cigarettes smoked: Never tried; experimental smoker; former
occasional or regular smoker; former daily; occasional; transitional; light
regular; moderate regular; heavy regular; light daily smoker; moderate
daily smoker; heavy daily smoker.
Alcohol Use
A single item on frequency of use: none in the past year; 1-2 days in the
past year; 3-12 days in the past year; 2-3 days in the past month; 1-2
days in the past week; 3+ days in the past week.
Suicidal Thoughts and Attempts
Five levels of thought or attempt in the past year as defined by a
combination of items: no suicidal thoughts or attempts; thoughts; one
attempt in the past year; two attempts; three or more attempts.
Weapon-Related Violence
A scale of 13 items measuring the number of incidents of weapon use,
weapon carrying, and/or involvement in incidents where they or
someone else was injured by a weapon (alpha reliability=.83). Items were
selected based on review of the literature and a factor analysis of 26 items
involving delinquent behavior, fighting, weapon use, weapon carrying
and injury by a weapon.
Sexual Intercourse
One question: Ever had vaginal intercourse.
Smoking: smoked one or more
cigarettes in the past 30 days;
Alcohol use: drank any alcohol in the
past 12 months;
Suicide risk: any suicide thoughts or
attempts in the past 12 months;
Weapon-related Violence: any
weapon-use, weapon carrying, or any
incident where a weapon was used;
Sexual intercourse: ever had
intercourse.
“Cut-Points” of Teen Risk
Behaviors used in the
Descriptive Analyses*
* These outcome behaviors are treated as continuous or quasicontinuous
measures in the second part of the report.
Black teens were the most
likely to have had sexual
intercourse.
White and Hispanic teens
were more likely than Black
youth to report suicidal
thoughts and attempts.
The following analyses looked at the
unique relationship of each demographic
variable to the health risk behaviors after
controlling for the influence of the
others (for example, the relationship of
race to smoking after controlling for the
effects of income, family structure and
gender).
Cigarette Use
Over half (55%) of the 7th to 12th
graders in the study said they had never
smoked a full cigarette. One quarter of
the sample (27%), representing 5.4
million American teenagers, reported
having smoked in the past 30 days. The
remaining youth (18%) were former
smokers. The prevalence of smoking
nearly doubled between middle school
and high school, rising from 19% in the
7th-8th grade to 37% in the 11th-12th
grade.
10
14,380,500
Do not smoke
4,385,100
534,100
454,100
10
20
30
40
50
60
70
White Hispanic Black
(14,054,800) (2,490,100) (3,208,900)
31%
20%
14%
Percent of Teens Who Smoke
Number of Teens Who Smoke *
(N = 19,753,800)
White
Hispanic
Black
* Based on the percent of youth in the sample applied to national
adolescent census data.
White youth smoked
more than Black or
Hispanic teens;
Youth from wealthier
families smoked less than
youth from poorer families
regardless of race, gender, or
family structure;
Teens from single parent
homes were at increased risk
for smoking regardless of
grade, income or gender.
There were no gender
differences in cigarette
smoking at either younger or
older grades.
Alcohol Use
Only half (53%) of the 7th to 12th
grade youth reported that they had not
had a glass of beer, wine, or liquor in the
past year. Over one-quarter of the
sample (29%), representing 5.8 million
youth nationwide, said they drank
between once a year and once a month.
Another 8%, representing 1.6 million
youth, drank 2-3 days per month. The
final 10%, representing 2.0 million
youth, drank weekly. The prevalence of
alcohol use more than doubled between
11
middle school and high school, from
28% in the 7th-8th grade to 63% in the
11th-12th grade.
White youth use alcohol more
than Hispanic youth and
much more than Black youth,
regardless of gender;
Black teens reported drinking
less than either White or
Hispanic youth. Among 9-
12th graders, Hispanic teens
reported drinking less than
White youth as well. This
finding challenges previous
findings that Hispanic youth
are at especially high risk for
alcohol abuse.
Among 9th to 12th graders,
youth from wealthier families
reported more drinking than
their lower income-peers.
Teens in single parent families
were more likely to drink than
those in two-parent families.
Older adolescent females use
alcohol less frequently than
same age boys. There were no
gender differences in drinking
among younger teens.
12
10
20
30
40
50
60
70
White Hispanic Black
(14,054,800) (2,490,100) (3,208,900)
50%
46%
36%
Percent of Teens Who Drink Alcohol
Number of U.S. Teens Who
Drink Alcohol *
(N = 19,753,800)
White
Hispanic
Black
10,390,500
Who report not drinking
alcohol in past year
7,027,400
1,145,500
1,155,200
* Based on the percent of youth in the sample applied to national
adolescent census data.
Suicidal Thoughts and Attempts
Overall, 12.6% of the adolescents,
representing 2.5 million youth in school
nationwide, reported suicidal thoughts or
attempts in the past year. Approximately
one third (0.7 million), indicated that
they had already made at least one recent
attempt. The percentage of youth
reporting suicidal thoughts or attempts
was relatively stable across grades,
income levels, and family structure but
varied with gender and race/ethnicity.
Females were at greater risk than males
(16% versus 9%), while White and
Hispanic youth were at somewhat greater
risk than Black youth (9% versus 7%).
White and Hispanic youth
were more likely than Black
youth to report suicidal
thoughts and attempts in all
grades.
Among 9th to 12th graders,
suicidal thoughts and attempts
were slightly less common
among wealthier youth.
Among 9th to 12th graders,
suicidal thoughts and attempts
were more common among
youth from single-parent
families.
13
10
20
30
40
50
60
70
White Hispanic Black
(7,018,700) (1,237,600) (1,568,600)
10% 9% 7%
Percent of MaleTeens Who Have
Suicidal Thoughts or Attempt Suicide
Number of U.S. Male Teens Who Have
Suicidal Thoughts or Attempts *
(N = 9,824,900)
White
Hispanic
Black
8,900,900
No thoughts or attempts
in past year
701,900
111,400
111,800
* Based on the percent of youth in the sample applied to national
adolescent census data.
Females were more likely than
males to report suicidal
thoughts and attempts in
every grade.
Weapon-Related Violence
Weapon-related violence —defined as
using a weapon, carrying a weapon, or
being in an incident where someone was
injured by a weapon in the past year—
14
10
20
30
40
50
60
70
White Hispanic Black
(7,036,700) (1,252,500) (1,640,200)
16% 17%
13%
Percent of FemaleTeens Who Have
Suicidal Thoughts or Attempt Suicide
Number of U.S. Female Teens Who
Report Suicidal Thoughts or Attempts *
(N = 9,928,800)
White
Hispanic
Black
8,376,900
No thoughts or attempts
in past year
1,125,800
212,900
213,200
* Based on the percent of youth in the sample applied to national
adolescent census data.
may be far more common among
American youth than previously
suspected. Overall, 26% of the sample,
representing 5.3 million students
nationwide, reported being involved in
weapon-related violence. Among those
with any involvement in weapon-related
violence (one or more incidents), fully
35% (representing 1.8 million students),
said they had used a weapon to threaten
or hurt someone in the past year. This
proportion rose to 55% (1.3 million
students) among those who were
involved in three or more violent
incidents and to 78% (0.7 million
students) among those who were
involved in six or more violent incidents.
The prevalence of weapon-related
violence was surprisingly stable across
grades 7 through 12 (24% to 29%).
Black and Hispanic youth
were more likely than White
youth to report involvement
in weapon-related violence —
independent of income, family
structure or gender.
Youth from wealthier families
were less likely to be involved
in weapon-related violence
than their lower income peers
— independent of race, family
structure, or gender.
15
10
20
30
40
50
60
70
White Hispanic Black
(7,018,700) (1,237,600) (1,568,600)
33%
44% 46%
Percentage of Teenage Males Involved
in Weapon-Related Violence
Number of U.S. Male Teens Involved
in Weapon-Related Violence *
(N = 9,824,900)
White
Hispanic
Black
6,242,650
Males not involved in
weapon-related violence
2,316,000
544,600
721,600
* Based on the percent of youth in the sample applied to national
adolescent census data.
Youth in single-parent families
were more likely to be
involved in weapon-related
violence than youth in twoparent
families —
independent of race, family
structure, or gender.
Males were more likely than
females to be involved in
weapon-related violence —
regardless of race, income, or
family structure.
16
10
20
30
40
50
60
70
White Hispanic Black
(7,036,100) (1,252,500) (1,640,200)
12%
25%
30%
Percentage of Teenage Females Involved
in Weapon-Related Violence
Number of U.S. Female Teens Involved
in Weapon-Related Violence *
(N = 9,928,800)
White
Hispanic
Black
8,279,400
Females not involved in
weapon-related violence
844,300
313,100
492,000
* Based on the percent of youth in the sample applied to national
adolescent census data.
Sexual Intercourse
Reports of ever having had sexual
intercourse increased dramatically with
grade, from 16% among 7th to 8th
graders to 60% among 11th to 12th
graders, with a corresponding rise in
risks of pregnancy and sexually
transmitted diseases like AIDS and
hepatitis.
Black youth were more likely
to have had intercourse than
White or Hispanic youth.
Those from wealthier families
were less likely to have had
intercourse than those from
lower-income families.
Youth in single-parent families
were more likely to have had
intercourse than youth in twoparent
families.
Among 7th and 8th graders,
females were less likely to have
had intercourse than males.
What Does It Mean?
While some adolescent health-risk
behaviors appear to be disproportionately
prevalent among Black and Hispanic
youth, lower income adolescents, and
17
10
20
30
40
50
60
70
White Hispanic Black
(7,018,700) (1,237,600) (1,568,600)
33%
41%
65%
Percentage of Teenage Males Who
Have Ever Had Sexual Intercourse
Number of U.S. Male Teens Who
Have Ever Had Sexual Intercourse *
(N = 9,824,900)
White
Hispanic
Black
5,981,800
Males who never
2,316,200
507,400
1,019,600
* Based on the percent of youth in the sample applied to national
adolescent census data.
had intercourse
youth living in single-parent families,
other behaviors, such as substance use,
are higher among upper income White
youth. No matter what the association, it
is unwise to conclude that if we know
these things about an adolescent we can,
with any degree of accuracy, predict his
or her health-risk behavior.
How much does knowing an
individual’s race/ethnicity, income, and
family structure together help us explain
that the individual will likely participate
in adolescent health-risk behaviors? A
measure called R2 (the square of the
correlation between a health-risk
behavior and a set of predictors) provides
a model by which to proceed. It indicates
what percentage of the individual
differences regarding involvement in a
health-risk behavior (ranging from none
to a lot) can be explained by knowing
other things about that individual. In
this case, regardless of gender, we want to
see the joint effect of knowing a youth’s
race/ethnicity, family structure, and
income.
Cigarette Smoking: Only
4.1% of the individual
differences in the amount of
cigarette use among younger
and 7.2% of the differences in
amount of smoking among
18
10
20
30
40
50
60
70
White Hispanic Black
(7,036,100) (1,252,500) (1,640,200)
35%
32%
65%
Percentage of Teenage Females Who
Have Ever Had Sexual Intercourse
Number of U.S. Female Teens Who
Have Ever Had Sexual Intercourse *
(N = 9,928,800)
White
Hispanic
Black
6,245,300
Females who never
2,462,600
400,800
820,100
* Based on the percent of youth in the sample applied to national
adolescent census data.
had intercourse
older teens can be explained
merely by knowing the youth’s
race/ethnicity, income, and
family structure.
Drinking: The same three
demographic variables
explained only 1.1% of the
individual differences in
frequency of alcohol use
among younger teens and
2.3% of the individual
differences among older
youth.
Suicidal thoughts and
attempts: Less than 0.5% of
the individual differences in
degree of suicide risk could be
explained by race/ethnicity,
income, and family structure
in either grade group.
Violence: Only 2.7% of the
individual differences in the
level of involvement in
weapon-related violence could
be explained from the same
demographic variables
considered together.
Ever had sex: An analogue of
R2 for logistic regression
(which is used for dichotomous
“yes/no” rather than
continuous measures)
indicated that race/ethnicity,
income, and family structure
explained only 9.7% of the
individual differences as to
whether a younger teen had
ever had sexual intercourse,
and 2.9% of the individual
differences in sexual
intercourse among older teens.
19
20
Thus, while the cultural contexts of
race/ethnicity, family structure, and
income help shape behavior, knowing
these factors is insufficient to accurately
target interventions or policies. Other
influences in the lives of teens are
important to our understanding and
effectively addressing adolescent healthrisk
behaviors.
Understanding Individual
Differences within Groups
If race/ethnicity, income, and family
structure are weak explanations for
youth health-risk behaviors, then what
might help us better to understand the
factors that contribute to some youth
participating in risk behaviors while
others avoid them?
What is Beyond Race, Income,
and Family Structure?
Teen smoking, drinking, weaponrelated
violence, suicide attempts and
unprotected sex are among the major
public health concerns in the United
States today. Perhaps one of the factors
that has made it difficult to address
these problems has been the focus on
Cigarette
smoking
Differences in Health-Risk Behavior
Explained By Race/Ethnicity,
Income and Family Structure
Alcohol use
Weapon-related
violence
Suicidal
thoughts and
attempts
Ever had
intercourse
0% 5% 10% 15% 20% 25%
Percent of Behavior Explained
7th–8th Grade
9th–12th Grade
4.1%
7.2%
1.1%
2.3%
2.7%
2.7%
0.8%
0.5%
9.7%
2.9%
21
group differences and the belief that
targeting policies and programs towards
the “highest risk” groups is the primary
solution.
The findings presented in the first
part of this report suggest that such an
approach, especially when based on
demographics, cannot solve the larger
public health problem. While it is
important to continue to direct
resources and programs to groups where
rates of health-risk behavior are high, it
is equally important to remember that,
in absolute numbers, the majority of
health-risk behaviors occur among
populations who, according to
demographics, would be considered
low-risk.
How else might we identify
vulnerable youth? Are there key risk and
protective factors that generalize across
racial/ethnic groups vis-a-vis risk
behaviors? Are certain risk behaviors
unique to one or another group?
Answering these questions present
implications for developing policies and
programs that would meet the needs of
American teens. The second part of this
report focuses on these questions.
22
n ecological systems
model views the adolescent
as developing
within the
context of family, peers, school,
community, and culture. The
closer the context is to the
teenager, the more directly it
influences his or her healthrelated
attitudes and behavior.
Linkages within and across contexts
play an important role,
since community and cultural
contexts often influence behavior
indirectly through their
impact on peer and family
norms. Links exist between the
health-risk behaviors as well,
according to Problem Behavior
Theory. For example, the extent
to which an adolescent tends to
violate social norms as a way of
establishing individuality is seen
as the underlying motivation
for behaviors as diverse as substance
use, delinquency, and
early sex.
A large body of research provides
general support for these theoretical
perspectives. What is less clear is
whether these influences are
expressed in similar ways within
various groups. The substantial
gender and racial/ethnic differences
in the prevalence of health-risk
behaviors, for example, suggest that
risk and protective factors may
differ among culturally distinct
groups. Alternatively, the risk and
protective factors may be the same
but simply occur more often, or
accumulate to a greater degree, in
some groups than others.
The unusually large sample and
broad scope of the Add Health
survey makes it possible to examine
relationships separately by gender
for the three racial/ethnic groups
described earlier, using measures
collected at the same time and
defined in identical ways for each
group. Interactions among
individual, peer and family factors
were also tested as well, since
Individual Factors, Friends and Family
Influence the Behavior of Teens
Sample. The same nationally
representative samples of
White (non-Hispanic), Black (non-
Hispanic) and Hispanic youth were
used in the first section, but divided
by gender instead of grade.
Since sample size affected the sensitivity
of the analyses, the White
sample was randomly divided into
four sub-samples that were similar
in size to the other groups. Only
results that were significant in at
least three of the four White subsamples
are reported in tables on
pages 23-33.This helps ensure that
comparisons among racial/ethnic
groups are not biased by differences
in statistical sensitivity and
that very weak effects are not
given undue importance.
Risk and protective factors.
Based upon the ecological
model and problem behavior theory,
over 50 potential risk and protective
factors derived from the
youth and parent surveys were
screened; those that had a simple
association of r >.10 with any risk
behavior in any gender/ethnic
group were retained for testing in
the models described below.
Where possible, risk and protective
factors were defined by more
than one survey item in order to
improve the validity and stability
with which the underlying constructs
were measured.
METHODS
Theories of adolescent development and health-risk behavior
provide the framework for this report.
A
23
resiliency theory suggest that the joint
occurrence of risk and protective factors
may increase or reduce risk to a greater
degree than either factor operating alone.
Although such interactions, or “enhanced
effects,” are an integral part of the
theoretical frameworks cited above, they
have not been widely studied in practice.
This study therefore provides new
evidence of how important these
“enhanced effects” may be in
understanding adolescent health-risk
behavior.
Cigarette Use
Considered jointly, the risk and
protective factors shown on page 25
explained 31% to 42% of individual
differences in the extent of smoking
among males and 24% to 49% of
individual differences in the extent of
smoking among females.
The most cross-cutting risk and
protective factors, regardless of gender or
race/ethnicity, were:
Frequent problems with school
work;
Frequency of “just hanging
out” with friends each week;
Number of best friends who
smoke daily.
Depending on the group, the effect of
having best friends who smoke was
enhanced when self-esteem was low, if
there was a smoker at home, or if the
teenager hangs out a lot with friends.
Females: A history of rape or sexual
abuse (defined as intercourse before the
age of 12) was associated with increased
cigarette use among females, regardless of
race/ethnicity.
Models of association. Multiple or logistic regression was used to select
a set of risk and protective factors that, taken together, did the best
job of reproducing individual scores on the risk behavior in question.
These models are often called “prediction models” even though in this
case they were based on a snapshot of data collected at one point in time.
A hierarchical stepwise method was used to select the optimal set of factors.
After entering the control variables, blocks of factors were forced
into the model in a pre-determined order based on their theoretical
closeness to the adolescent’s attitudes and behavior: individual-level, peer,
and then family. Within each block, statistical criteria determined which
specific factors would be retained, if any.
The multiple correlation coefficient (R) provided a measure of the
strength of association between each health risk behavior and the final set
of statistically selected factors.When squared (R2), the multiple correlation
indicates how well a set of risk and protective factors, considered
jointly, reproduces individual scores on the measure of health risk behavior.
Although not without its drawbacks, R2 provides a single measure for
comparing how well the models developed for each group do at explaining
individual differences about the extent to which youth engaged in the
health risk behavior. Since this study was actually a snapshot of associations
taken at one point in time, however, the results should be viewed as
preliminary.Additional research is needed to determine the usefulness of
these models for predicting behavior at a later point in time, as is research
that attempts to determine which factors are causes and which are merely
markers of risk.The reader should also keep in mind that the R2 for each
domain (individual, peer, family) was influenced somewhat by the number
of variables and order of domains in the analyses, although changing the
order did not materially alter the conclusions presented in this report.
ANALYTIC APPROACH
Minority Males: Frequently attending
youth groups or services at church, high
self-esteem, and expectations of
attending college were protective against
smoking for minority males, while
greater self-assessed physical maturity was
associated with increased risk.
Black Youth: A positive relationship
with parents and family was uniquely
protective against smoking among Black
youth.
The strength and consistency of peer
influences on smoking is upheld by other
research, although the relationship is
known to be complex. Youth both pick
friends who do what they want to do,
and are influenced by those friends’
behaviors. In this analysis, the influence
of having friends who smoked was
enhanced by risk factors in other
domains. This suggests that the
association may be at least partly due to
the influence of friends.
24
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
White
Black
Hispanic
White
Black
Hispanic
Males
Females
How much knowing individual, peer,
family and socio-economic factors help
explain which teens are likely to smoke (R2)
Individual
Peer
Family
Socio-economic
Unexplained
Individual
Peer
Family
Socio-economic
Unexplained
25| A dichotomous (yes/no) variable. History of rape was asked only of females. History of sexual abuse was defined as first intercourse before age 12.
1 The following risk and protective factors were not associated with amount of smoking in any gender/ethnic subgroup: ever repeated a grade; physical
recreation; extent of religious beliefs; belief that successes were earned; family member suicide or attempt; presence of extended family in the home (youth;
adult); whether Spanish was primary language at home (Hispanic); parent present at dinner; parent presence after school; number of siblings; extent of joint
parent-youth decision making.
2 The following risk and protective factors were treated as non-significant for Whites in the table because they were not consistently related to smoking among
White youth when their sample size was matched to the Hispanic and Black sample sizes: religious activities; hobbies; physical maturity; temper; history of
sexual abuse; belief that successes were earned; parent-family relationship; parent-presence at dinner; number of siblings; friend suicide or attempt.
3 Risk enhanced by friends smoking.
4 Risk enhanced if there is a smoker in the home.
5 Risk enhanced by just hanging out.
6 Risk enhanced by low self-esteem.
Factors Associated with Cigarette Smoking1 = Risk = Protective
Males Females
White2 Black Hispanic White2 Black Hispanic
Individual Level
Frequent problems with school work
Frequently just hangs out with friends 3 3
Frequenty religious activities
Frequency of hobbies
Work 20 hrs/wk during school*
Degree of physical maturity
Rape/sexual abuse history*
Youth has bad temper*
Degree of self-esteem
Wants & expects to attend college
Peer Context
Number of best friends who smoke 4 5/6 6 4 5 4/5
Prejudice among students at school
Family Context
Smoker in the home* 3 3 3
Teen sets his/her own curfew*
Positive parent/family relationship)
Alcohol Use
Considered jointly, the risk and
protective factors shown in the Table on
page 27 explained 35% to 44 % of
individual differences in the frequency of
drinking among males and 28% to 43%
of individual differences in the frequency
of drinking among females.
The most important risk factors,
regardless of gender or race/ethnicity, were:
Frequent problems with
school work, and
Number of best friends who
drink at least monthly.
Depending on the group, the effect of
having best friends who drink was
enhanced by having parents who drink
frequently or by adolescent reports of
frequently just hanging out with friends.
Males: The frequency with which
males “just hang out” with friends was
associated with increased alcohol use in
every racial/ethnic group. For White and
Black males, the frequency of hanging
out also increased the risk associated with
having best friends who drink.
Minority Youth: Greater self-assessed
physical maturity was associated with
frequency of alcohol use among Black
and Hispanic males and females, while
the number of siblings at home was
associated with reduced risk for minority
females.
As with smoking, the strength and
consistency of peer influences on
drinking is reflected in prior research. In
this analysis, the influence of having
friends who drink was enhanced by risk
factors in other domains. This suggests
that the association may be at least partly
due to peer influences.
26
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
White
Black
Hispanic
White
Black
Hispanic
Males
Females
How much knowing individual, peer, family
and socio-economic factors help explain
which teens are likely to drink alcohol (R2)
Individual
Peer
Family
Socio-economic
Unexplained
Individual
Peer
Family
Socio-economic
Unexplained
27| A dichotomous (yes/no) variable. History of rape was asked only of females.
1 The following risk and protective factors were not associated with drinking in any gender/ethnic subgroup: ever repeated a grade; physical recreation; wants
and expects to attend college; frequency of religious activities; belief that successes were earned; self-esteem; extent of religious beliefs; family member suicide of
attempt; presence of extended family in the home (youth; adult); whether Spanish was the primary language at home (Hispanic); extent of prejudice at school;
youth sets his/her curfew; frequency of parent presence after school.
2 The following risk and protective factors were listed as non-significant for whites in the table because they were not consistently related to drinking among
White youth when their sample size was matched to the Hispanic and Black sample sizes: frequency of religious activities; physical maturity; history of rape
(females); parent-family relationship; parent presence after school.
3 Risk enhanced by friends drinking.
4 Risk enhanced by parents drinking.
5 Risk enhanced by just hanging out with friends.
Factors Associated with Alcohol Use1 = Risk = Protective
Males Females
White2 Black Hispanic White2 Black Hispanic
Individual Level
Frequent problems with school work
Frequently just hangs out with friends 3 3 3
Frequency of hobbies
Work 20 hrs/wk during school*
Degree of physical maturity
Rape/sexual abuse history*
Youth has bad temper*
Peer Context
Number of best friends who drink 5 5 4/5
Friend suicide or attempt*
Family Context
Frequency of parent drinking 4
Frequency parent present at dinner
Positive parent/family relationships
Extent of joint decision-making
Number of siblings
Weapon-related Violence Table
Considered jointly, the risk and
protective factors shown in the table on
page 29 explained 20% to 31% of
individual differences in the extent of
violence reported by males and 12% to
25% of individual differences in the
extent of violence reported by females.
The most important risk and
protective factors, cutting across gender
and ethnic groups were:
Frequent problems with school
work;
Number of best friends who
drink;
Friend’s suicide or attempt
(except among Black males);
Degree of positive parent and
family relationship (except
among White females).
In contrast to substance use, there was
only scattered evidence of enhanced
effects due to other factors. Surprisingly,
easy access to a gun at home was not
consistently related to the degree of
involvement in weapon-related violence.
It was a specific risk factor for White and
Black males and Black females.
Males: Frequency of “just hanging out
with friends” was associated with
increased involvement in weapon-related
violence among males, regardless of
race/ethnicity.
Minority Males: In addition to the
above, degree of self-assessed physical
maturity, having repeated a grade, and a
28
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
White
Black
Hispanic
White
Black
Hispanic
Males
Females
How much knowing individual, peer,
family and socio-economic factors help
explain which teens are likely be
involved in weapon-related violence (R2)
Individual
Peer
Family
Socio-economic
Unexplained
Individual
Peer
Family
Socio-economic
Unexplained
29| A dichotomous (yes/no) variable. History of rape was asked only of females.
1 The following risk and protective factors were not associated with extent of weapon-related violence in any gender/ethnic subgroup: work 20+
hours/week; frequency of hobbies; frequency of religious activities; extent of religious beliefs; belief that successes were earned; self-esteem;
frequency of parent presence after school; sets own curfew; extent of prejudice at school; number of siblings; whether Spanish was the primary language at
home (Hispanic).
2 The following risk and protective factors were listed as non-significant for Whites in the table because they were not consistently related to
violence among White youth when their sample size was matched to the Hispanic and Black sample sizes: college plans; physical recreation (a risk); physical
maturity; history of rape/sexual abuse; self-esteem; family member suicide or attempt; joint decision-making; sets own curfew.
3 Risk enhanced by friends drinking.
4 Risk enhanced by low joint decision-making.
5 Risk enhanced by just hanging out with friends.
6 Protection enhanced if no bad temper.
Factors Associated with Weapon-Related Violence1 = Risk = Protective
Males Females
White2 Black Hispanic White2 Black Hispanic
Individual Level
Frequent problems with school work
Frequently just hangs out with friends 3 3
Degree of physical maturity
Rape/sexual abuse history*
Youth has bad temper* 4
Wants & expects to attend college
Ever repeated a grade*
Peer Context
Number of best friends who drink 5 5
Friend suicide or attempt*
Family Context
Positive parent/family relationships
Joint decision-making 6
Family member suicide or attempt*
Extended family in the home (adults)*
Extended family in the home (youth)*
Frequency parent presence at dinner
Easy access to guns at home*
family member’s suicide or attempt were
all associated with an increased
involvement in weapon-related violence
among Black and Hispanic males.
Among Black males, according to
parental report, those who have a bad
temper have more reported weaponrelated
violence.
30
Minority Females: Percentages of
students’ involved in weapon-related
violence were surprisingly high among
Black and Hispanic females – 30% and
25% respectively. Most of the risk and
protective factors found among Black
males applied to Black females as well,
Frequency of “just hanging out”, having
repeated a grade, having a bad temper,
and a history of rape, further increased
their risk. However, there was no such
pattern for Hispanic females. Selfassessed
physical maturity, expectations
of attending college, and whether or not
there were adult extended family
members at home were associated with
their degree of involvement in violence.
The findings reported here are unique
in two respects. First, they are limited to
violence involving weapons. Most
previous research has included fighting in
the measure of violence because reports of
using a weapon are rare. However,
fighting is relatively common, especially
among early and middle adolescents. It
declines with grade as adolescents mature;
fighting is not nearly as great a threat to
adolescent health as violence involving a
weapon. So this study used a more
sensitive index of violence that included
reports of carrying a weapon and seeing
someone injured or being injured by a
weapon — all of which are associated
with increased weapon use by youth (See
Measuring Health and Behavior on page 2)
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
White
Black
Hispanic
White
Black
Hispanic
Males
Females
How much knowing individual, peer,
family and socio-economic factors help
explain which teens are likely to have
suicidal thoughts or attempt suicide (R2)
Individual
Peer
Family
Socio-economic
Unexplained
Individual
Peer
Family
Socio-economic
Unexplained| A dichotomous (yes/no) variable. History of rape was asked only of females.
1 The following risk and protective factors were not associated with extent of suicide risk in any gender/ethnic subgroup: ever repeated a grade; college plans;
working 20+ hours/week; frequency of hanging out; hobbies; religious activities; physical recreation; bad temper; belief that successes were earned; extent of
prejudice at school; easy access to a gun at home; youth sets own curfew; frequency of parent at dinner; frequency of parent presence after school; joint
parent-youth decision making; extended family in the home (adult); whether Spanish was the primary language at home (Hispanic).
2 The following risk and protective factors were listed as non-significant for Whites in the table because they were not consistently related to suicide risk among
White youth when their sample size was matched to the Hispanic and Black sample sizes: physical recreation; physical maturity; history of rape; religious belief;
frequency of just hanging out; self-esteem; parent presence after school; family member suicide or attempt; joint decision-making; temper.
3 Risk enhanced by parent drinking.
4 Risk enhanced by friends drinking.
Factors Associated with Suicidal Thoughts and Attempts1 = Risk = Protective
31
Males Females
White2 Black Hispanic White2 Black Hispanic
Individual Level
Frequent problems with school work
Degree of self-esteem
Degree of physical maturity
Rape/sexual abuse history*
Extent of religious beliefs
Peer Context
Friend suicide or attempt*
Number of best friends who drink 3
Family Context
Frequency of parent drinking 4
Positive parent/family relationships
Extended family in the home (youth)
Secondly, this is perhaps the first report
using a nationally representative sample
that reports the rates weapon-related
violence among females; thus it is the first
comparison of correlates of weaponrelated
violence among females from
different racial/ethnic groups.
Suicidal Thoughts and Attempts
Considered jointly, the risk and
protective factors shown in table on page
31 explained 9% to 10% of individual
differences in extents of suicide risk
among males and 12% to 23% of
individual differences in extents of
suicide risk among females.
The most important risk factor,
regardless of gender or race/ethnicity,
was:
Friend suicide completion or
attempt.
Frequent problems with schoolwork
further increased risk among White and
Hispanic youth. A positive parentfamily
relationship was protective among
females, regardless of race/ethnicity.
The sparseness of significant risk and
protective factors is consistent with the
fact that the major risk factors for suicide
are related to personality, social skills and
genetic factors that could not be
measured and/or analyzed in the present
report. The fact that there is only a
limited overlap with identified risk and
protective factors for weapon-related
violence points to the need for caution
when drawing comparisons between selfdirected
and weapon-related forms of
violence directed at others. Furthermore,
the weak association observed between
measures of suicide risk and weaponrelated
violence (r = .18 to .22 for all
groups except Black males where r = .10)
should be a cautionary note about
equating suicidal thoughts and attempts
with weapon-related violence. While
there are some shared risk factors,
including problems with school, friend’s
suicide or attempt, and the parent/family
relationship, there are a number of
factors that are distinct beetween the
two.
Sexual Intercourse
Considered jointly, the risk and
protective factors shown in the table on
page 33 explained 25% to 34% of the
individual differences among males
regarding whether they had ever engaged
in sexual intercourse and 35% to 49% of
the individual differences among females.
32
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
White
Black
Hispanic
White
Black
Hispanic
Males
Females
Individual / Opportunity &
Motivation
Individual
Peer
Family
Socio-economic
Unexplained
How much knowing individual, peer,
family, socio-economic factors and
individual opportunity & motivation
for sexual activity help explain which teens
are likely to have sexual intercourse (R2)
Individual / Opportun
Motivation
Individual
Peer
Family
Socio-economic
Unexplained| A dichotomous (yes/no) variable. History of rape was asked only of females.
1 The following risk and protective factors were not associated with ever had intercourse in any gender/ethnic subgroup: whether would keep child if got pregnant; frequency
of religious activities; or physical recreation; youth has bad temper; believes successes were earned; self-esteem; parent presence after school; parent presence at dinner; sets
own curfew; frequency parent drinking; family member suicide or attempt; joint decision-making; whether Spanish was the primary language at home (Hispanic).
2 The following risk and protective factors were listed as non-significant for Whites in the table because they were not consistently related to ever had intercourse among
White youth when their sample size was matched to the Hispanic and Black sample sizes: ever dated; virginity pledge; ever repeated a grade; hobbies; college plans; parentfamily
relationship; extended family in the home (adults); friend suicide or attempt; number of best friends who drink; worked 20+ hours/week.
3 Extent of perceived knowledge of birth control was only weakly related to extent of actual knowledge regardless of grade or sexual experience.
4 Type “analogue of R2” explanation from section 1 as a footnote here or combine with “*” comment instead, if possible.
5 Risk enhanced if see benefit to sex.
6 Risk enhanced if repeated a year. 10 Protection enhanced if sees risk of pregnancy.
7 Risk enhanced if feels knowledgable about birth control. 11 Protection enhanced if strong religious belief.
8 Risk enhanced if sees few social costs. 12 Protection enhanced if good school attendance.
9 Risk enhanced if had a romantic relationship. 13 Protection enhanced if see many social costs.
Factors Associated with Whether or Not Youth Had Sexual Intercourse1 = Risk = Protective
33
Males Females
White2 Black Hispanic White2 Black Hispanic
Individual Sexual Experience
Opportunity
Ever dated*
Ever kissed or necked*
Romantic relationship in 18 months 5/6 5
before survey*
Motivation
Made public or written virginity pledge*
Perceived personal & social benefits to sex 7 8 9 9
Perceived personal and social costs to sex 10 10 11
Perceived costs of get/make someone pregnant 12 12 12 12 12
Perceived (not actual) knowledge
of birth control3
Individual
Ever repeated a grade* 9
Frequent problems with school work
Wants & expects to attend college
Religious beliefs 13
Physical maturity
Peer Context
Number of best friends who drink
Prejudice among teens at school
Family Context
Parents disapprove of youth having sex
at this time in their life
Positive parent/family relationships
Number of siblings
The most important risk factor,
regardless of gender or race/ethnicity, was:
Having been in a romantic
relationship in the prior 18
months.
Two factors were related to sexual
behavior among White and Black males
and among Hispanic and Black females:
Perceived benefits and
perceived costs of having sex
(except Black females);
Perceived personal and social
costs of getting/making
someone pregnant (except
Hispanic males);
Perceived (not actual) degree
of knowledge about birth
control (expect Hispanic
males).
Males: Ever having kissed or necked,
the extent of perceived personal and social
costs as well as the perceived personal and
social benefits of having sex were risk or
protective factors for males in all three
racial/ethnic groups
Minority Males: Having made a public
or written virginity pledge was associated
with not having had intercourse. No such
association was found among White
males.
Minority Females: The adolescent’s
perception of parents’ strong disapproval
of her having sex at this time in her life
was associated with the Black and
Hispanic females reporting of not having
sexual intercourse. No such association
was found among White females.
Black Females: Having made a virginity
pledge, having a larger number of siblings,
34
and the extent of positive parent/family
relationships, were associated with
reporting not having had sexual
intercourse. Reporting ever having dated
and having a large number of friends who
drink were associated with having sexual
intercourse for Black females.
Hispanic Females: Ever having kissed
or necked, perceived costs and perceived
benefits of having sexual intercourse were
risk or protective factors, as were
problems with school work, degree of
self-assessed physical maturity, and extent
to which other students at school were
thought to be prejudiced.
The analyses provide extensive
evidence of enhanced effects among risk
and protective factors within the domain
of individual factors specific to sex, and
suggest that the decision to have sexual
intercourse may be heavily influenced by
both opportunity and the perceived trade
offs between costs and benefits. It is
consistent with the fact that sexual
intercourse is a normative behavior
which we merely seek to delay rather
than prevent.
Two other findings were noteworthy.
First, the variables we analyzed were
predictive of sexual intercourse among
Black males as well as for other ethnic
groups and for females. What
differentiates the present study from
many others was the large number of
individual-level variables specifically
related to sexual intercourse. Secondly, a
surprisingly large percentage of
individual differences in sexual
experience was explained in every
gender/ethnic group in this study, and
nearly all of the explanatory power was
attributable to individual-level factors
specific to sex. General factors that were
found to predict other health risk
behaviors did not seem to make nearly as
much difference when it came to sex.
35
Summary
When we examine the factors, events
and experiences that apply across gender,
most ethnic groups and health risk
behaviors, the following risk and
protective factors stand out:
1. Youth who have problems with
schoolwork are more likely than
others to experience or be
involved with every health risk
studied. This is evident, with very
little exception, across the groups
studied. School failure is a public
health problem.
2. Teens who spend a lot of time
“just hanging out” with friends,
especially friends involved with a
specific risk behavior, are more
likely to be involved themselves.
Clearly, one’s choice of friends
matters. There may also be health
consequences to substantial
amounts of unstructured leisure
time.
3. Friends’ drinking behavior is
strongly associated with not only
teen drinking, but weapon-related
violence, and, in some groups,
with suicidal thoughts and
attempts. Whether this is just a
“red flag” or is directly related is
unclear. In the present study there
was no measure of friends’
involvement with violence.
4. No protective factor cut across all
health-risk behaviors. However,
the one most consistently
protective factor found was the
presence of a positive parentfamily
relationship.
36
37
nd while these factors influence
behavior they do not cause a
young person to engage in a
high-risk behavior. If our
goal is to improve outcomes for young
people and reduce risk, we must move
away from our focus on these demographic
factors and abandon them as
useful ways of understanding adolescent
health risk behaviors. We have seen that
they are weak predictors of adolescent
behavior. Additionally, they are distal,
indirect influences on young peoples’
lives, and do not directly control those
factors that have a greater determinative
influence on what young people do.
Finally, they are not especially amenable
to change.
What this report has also shown,
however, is that there are a number of
factors that are powerfully associated
with exacerbating or minimizing risks to
young people. Many of these are
amenable to change. Being at academic
risk was nearly universally associated
with every health risk behavior we
studied. We need to understand that
health and education are closely
intertwined and that school failure needs
to be viewed as a health as well as an
education crisis.
Friends have a powerful influence on
the lives of young people; they in turn
influence their friends. Parents need to
be involved with their teens’ friends—
know who they are — attend to what
they do, and supervise the amount of
time their children spend “hanging out”
with their friends.
Conclusion
Race, ethnic, and cultural affiliations, family structure and
income are all contexts within which young people live.
A
Frequently we see that when young
people are close to their parents and
family they are less likely to report
involvement with health-risk behaviors.
Parents need both the skills and support
to develop and maintain close, caring
relationships and connect with their
children as they progress though teenage
years.
When parents and family are involved
in the lives of their teenagers, young
people benefit. When they are involved
in their teenagers’ schooling, young
people benefit. And when parents and
family are involved with their teenagers’
friends, they benefit, regardless of
whether they are White, Black or
Hispanic, male or female.
In a society with high rates of divorce
and separation and where work demands
increasingly encroach on parent and
family availability, this report stands as a
warning and a promise. The warning is
that when parents are not personally and
psychologically available for their teenage
children, teenagers pay a high price.
When we as a society do not support
parents to be effective as well as available,
teenagers suffer. On the other hand, this
report also indicates that, in the final
analysis, it matters less if an adolescent
comes from a single or dual parent
family than what happens within the
family. When we nurture the capacity of
parents and of families to be involved in
the lives of their teenage children, young
people are the beneficiaries.
38
39
The authors wish to thank the Robert Wood Johnson Foundation for its support of the analysis and production of this
report. Deep appreciation goes to Chris Bachrach, PhD, Michael Resnick, PhD, and Clea McNely, DrPH for their careful reading
and thoughtful critiques of previous drafts. In the final analysis, however, the authors are responsible for what is reported.We
are grateful to Jonathon Chanetsa who provided detailed copy-editing.A very special debt of gratitude goes to Linda Boche
who tirelessly revised the tables, figures and text too many times to count.Additionally, we wish to acknowledge the commitment
and efforts of Elizabeth Latts, MSW and and Melissa Bishop for identifying the network and compiling the mailing lists
that comprise the first mailing of this report. Finally, we wish to thank the 18 federal agencies lead by the National Institute of
Child Health and Human Development that had the vision to support the development
of the National Longitudinal Study of Adolescent Health. It is to the
young people of the United States from all ethnic groups, all religions, all different
types of families, and all income levels that this report is dedicated with the hope
that what is presented will help those committed to working with young people
so that all young people will benefit.
Acknowledgements


#

tags