Risk-taking Behavior in Adolescents:
The Paradigm"
CHARLES E. IRWIN, JR.,VIVIEN IGRA, STEPHEN E Y E ,
AND SUSAN MILLSTEIN
Division of Adolescent Medicine
Department of Pediatrics
School of Medicine
University of California, San Francisco
San Francisco, California 94143
INTRODUCTION
The majority of adolescents emerge from the second decade of life relatively unscathed by serious social, psychological, or physical disability.I4 Risk behaviors are
largely responsible for the negative health outcomes among the remaining significant
minority of adolescents. The primary immediate threats to adolescent health are preventable behaviors such as substance use, dangerous vehicle use, and sexual behavior. Unhealthy eating behaviors have generally not been considered under the risk
construct; however, there are many similarities between the traditional behaviors
classified as risky and eating behaviors. These behaviors have a developmental trajectory as well. Typically, they increase in prevalence over the adolescent years,5 with
rates of substance use, unhealthy eating behaviors, and injury-related behaviors peaking in late adolescents and declining in young adulthood.Gs These potentially healthdamaging behaviors, established during adolescence, often have lasting deleterious
effects throughout the life span. There is considerable debate about whether these behaviors are risk-taking behaviors or just risky behaviors with negative health outcomes. A critical question continues to be debated: Does the risky nature of the behaviors initiated during adolescence in conjunction with their volitional nature imply
that there is a conceptual link among the behaviors? A number of investigators and
theorists argue that adolescent risk taking is a valid construct, although they disagree
about its conceptual basis.
Since the focus of these proceedings is Adolescent Nutritional Disorders, whenever possible the issues specifically related to eating and the associated behaviors will
be highlighted. It is hard to pinpoint an absolute reason why adolescents are at increased risk for risky behavior and the associated negative nutritional outcomes. As
with all health problems, there is a complex relationship among the host, the agent,
and the environment that are associated with an increased likelihood of engaging in
risky behaviors.
This paper reviews the theoretical constructs that provide a framework for under"Partial support for development of this paper was provided by the Maternal and Child Health
Bureau, Health Resources and Services Administration, Department of Health and Human Services (MCJ000978, MCJ063ASOA, and MCJ060564) and the William T. Grant Foundation.
1
2
ANNALS NEW YORK ACADEMY OF SClENCES
standing the covariation of risk-taking behaviors, the nature of adolescence itself, the
prevalence of adolescent sexual activity, substance use, injury-related behaviors, eating behaviors, and delinquency, and the mechanisms by which these behaviors covary. In addition, we present data from our longitudinal study of risk taking during
adolescence.
RISK-TAKING BEHAVIOR: A DEFINITION
Two assumptions have driven much of the work on risk taking in adolescents.
First, teenagers engage in a series of "risky" behaviors and feel invulnerable to the
consequences of these behaviors as compared to adults.
Second, all risk-taking behaviors occurring through the second decade of life are
dnven by similar mechanisms and the behaviors themselves are similar.
Risk-taking behaviors can be distinguished from developmentally appropriate exploratory behavior by their potentially serious, long-term, and negative consequences. Whereas adolescent exploratory behavior in a safe or positive context enhances competence and confidence, risk-taking behaviors jeopardize health and
well-being. Some risk-taking behaviors are defined by their adolescent age of onset.
For example, sexual activity, certain eating behaviors, driving a car, drinking alcohol,
or leaving home may be considered risk-taking behaviors at age 13, but may not be at
age 2 I . Some behaviors are risky regardless of age such as unhealthy eating behaviors, promiscuous sexual behavior, cocaine use, or driving under the influence of alcohol.
For the purposes of this paper, risk-taking behaviors will refer to volitional behaviors including sexual activity, eating behaviors, substance use, delinquency, and injury-related behavior^.^ Negative health outcomes of these behaviors include sexually transmitted diseases, unplanned pregnancy, school failure, habituation, premature
cardiovascular disease, hypertension, obesity and its associated medical sequelae,
physical and psychological disability, and death. Psychopathologic health-jeopardizing behaviors ( e g , suicidal behaviors and serious eating disorders) are not volitional
in the traditional sense, and are therefore not considered risk-taking behaviors for the
purposes of this paper.
The paradigm of risk is described in the following section using three general explanatory approaches: dispositional, ecological, or biological bases for risk-taking
behavior.'&l3 These distinctions are not clearly demarcated since theories often borrow from more than one of these approaches.
RISK TAKING AS A DISPOSITIONAL TRAIT
Dispositionally based theories of risk taking focus on individual differences associated with a tendency to engage in risky behaviors. These theories view risk taking
as deviant. The hypothesized deficits leading to deviancy include poor self-esteem,I4.Is depression,'(' inadequate social skills,I7 impulsivity characteristic of attention deficit disorder,I* or a general propensity for unconventionality and deviance.'y-2'
IRWIN et al.: ADOLESCENT RISK-TAKINGBEHAVIOR
3
With each of the proposed causal factors, there is some evidence that the factor is
associated with risky behaviors under a specific condition. There is also an emerging
body of evidence that may contradict some of the evidence. For example, in the area
of self-esteem, poor self-esteem appears to be associated with a number of risky behaviors and forms the basis of many intervention programs. However, a careful examination of the relationship of self-esteem and risk leads to a body of evidence that
risk taking itself can raise s e l f - e ~ t e e m ’ ~or*that
~ ~ .certain
~ ~ risky behaviors, for example, sexual behavior, are correlated with a higher level of self-esteem.22
Individual-difference models maintain that certain dispositional characteristics
may not be deficit models but reflect underlying differences in the organism (e.g.,
sensation seeking). Theories of sensation ~ e e k i n g state
~ ~ -that
~ ~ individuals differ in
terms of their underlying needs for stimulation (or sensation or thrill seeking), which
underlies much of risk-talung behavior. Sensation seekers are willing to take risks for
the sake of increases in stimulation and arousal. Sensation seeking has been associated with substance use, sexual behavior, dangerous sports, and motor-vehicle use. A
new body of evidence indicates that these biological differences may also be protective if the environmental context is ~upportive.~’
RISK TAKING FROM AN ECOLOGICAL PERSPECTIVE
In contrast to an individually oriented theory of risk taking, an ecological theory
considers the human organism within its social and environmental context. These
theories propose that contextual factors such as economic status, cultural background, and the general social environment provide social norms, models, opportunities, and reinforcements for adolescent participation in risky behaviors.
An ecological paradigm is used by Bronfenbrenner to describe the social world of
adolescents.28The most proximal contexts are referred to as microsystems, and include those elements of the social environment with which adolescents interact directly, such as peers, family members, and social institutions such as schools or
churches. More distal social contexts such as the community, the mass media, and
social policies are called exosystems. The most distal social contexts are referred to
as macrosystems, which reflect things such as cultural, economic, and political contexts. Many of the conceptual models of risk taking described in the literature can be
overlaid onto this general ecological model. Each of the four system levels described
by Bronfenbrenner is described below.
Microsystems and Mesosystems: Peers and Parents
Among all the microsystems of potential relevance, the role of peer influences in
adolescent risk taking has generated the most interest. “Peer pressure” has been implicated as an etiologic factor in adolescent substance use, sexual behavior, and antisocial b e h a ~ i o r , 2and
~ * ~intervention
~
programs designed to teach adolescents how to
resist peer pressure have been developed and widely i m ~ l e m e n t e d . ~During
’ - ~ ~ adolescence, peers assume an increasing influence as adolescents begin to establish an
identity outside of the family group.30 Conformity increases during the adolescent
4
ANNALS NEW YORK ACADEMY OF SCIENCES
period, peaking during early adolescence and decreasing thereafter.34Social learning
theory would suggest that peers may promote risky behaviors by providing models.
Beyond modeling, peers may also provide opportunities for risky behavior to
Research supports an association, both cross-sectionally and longitudinally, between
the adolescents’ behavior and the behavior of their peer^.^^.^^ The initiation of risk
behaviors may emerge from a desire to conform to an existing peer group, or adolescents who are inclined to taking risks may specifically choose peers who are similarly i n ~ l i n e d . ~ ~ . ~ ~
Parents are important elements of the microsystem for adolescents. The linking of
parents and adolescent risk taking has focused on the role of parenting styles, monitoring, and modeling. Two dimensions of parenting styles appear to be important in
adolescent risk taking: the level of demand that parents place on their adolescents
(e.g., limit-setting and responsibility), and the degree of emotional re~ponsibility.~~
In support of this approach, some research indicates that adolescents whose parents
are authoritative (demanding and responsive) are less likely to use substances than either those with authoritarian (demanding but unresponsive) or permissive (nondemanding but responsive) parents. Adolescents with neglecting and rejecting parents
are the most likely to engage in substance abuse.”
Parents may also influence adolescent risk taking as a function of their monitoring
of their child’s behavior. Lack of supervision has been used, for example, to explain
the reported relationship between family structure and risk taking. Adolescents from
single-parent families are more likely to initiate intercourse and less likely to use
contraception than their peers from intact familie~,4~.~’
and are more likely to engage
in various forms of substance use.42,43Younger adolescents (ages 13 to 14 years) who
take care of themselves after school show a twofold increase in substance use compared to their supervised p e e r ~There
. ~ ~ is also evidence that family structure does
not act as an independent variable and that adolescents who are emotionally autonomous (independent of family structure) engage in more risks.4s
Lack of parental supervision could be an explanation for the increase in adolescent risk taking over the past four decades. Significant changes in family structure
have taken place in the United States. Consider, for example, the image of the “typical family,” with a working father, a housewife mother, and two or more school-age
children. In 1955, 60% of the households in the United States fit this description; in
1985, only 7% of families fit the d e ~ c r i p t i o n .Significant
~ ~ . ~ ~ increases in the divorce
rate, the number of children born to unmarried females, and the percentage of mothers in the work force have changed the nature of the typical family. By mid-1985,
nearly 70% of mothers with children between the ages of 6 and 17 years were in the
labor force, leaving three-quarters of the 13-14-year-old children in the United States
in self-care or with a younger sibling for more than 10 hours a week.48Other authors
argue that other things, such as political and cultural changes, are more critical in the
increase in risk t a k i r ~ g . ~ ~ . ~ ~
Beyond lack of supervision, parents may influence adolescent risk as a function of
modeling. The strongest evidence for the effects of modeling have come from research on parental and adolescent tobacco use,s1 which show a significantly higher
rate of smoking among adolescents whose parents also smoke. Adolescent children
of substance-abusing parents are also more likely than children of abstainers to abuse
substances,s2 although alternative models, such as genetic predisposition, have also
been offered to explain this.s3
IRWIN et al.: ADOLESCENT RISK-TAKING BEHAVIOR
5
Parental influence may also be viewed in relation to how it compares to peer influence. Jessor has argued that it is the relatively greater influence of peers compared to
parents that is associated with a greater tendency toward risk-taking behaviors.54This
explanation is supported by research showing the relative dominance of peer influence over parental influence to be predictive of marijuana use, problem drinking, and
early sexual i n t e r c o ~ r s e .The
~ ~ .concept
~~
of the "mesosystem," represents an interrelationship between various microsystems, such as peers or parents. The relative influence of parental versus peer values can be viewed as an effect of the mesosystem.
Exosystems: Neighborhoods, Communities, and the Media
Exosystems reflect social contexts (often referred to as the social environment)
that influence adolescent behavior. Media and community norms are examples.
There is some evidence to support the effects of role models for risk-taking behavior
that are regularly presented by the media.57Controversy remains about the role of the
medias8 and the direct effects of these media presentations on risk taking. The pervasive nature of the media makes it difficult to conduct controlled research that would
allow for clarification of the role of the media. Most of the research base is concentrated in the area of television violence and childhood a g g r e s s i ~ n More
. ~ ~ recently,
investigators have explored the role of the media on its portrayal of dieting behaviors
and body types.60The direction of the effects of the media remains uncertain.
The social context model focuses on the neighborhood and the community. Communities show marked variation in their rates of adolescent alcohol use, illicit substance use, and early sexuality and childbearing.50Some theorists suggest that certain
environments are risk-promoting by providing adolescents with access to, and motivation for, engaging in risk taking6' Access to risk taking is increased in communities that permit cigarette vending machines and food vending machines or that lower
required ages to purchase alcoh01.'~In some communities, such as economically depressed, inner-city neighborhoods, behaviors such as substance use may emerge as a
function of the availability of substances in these communities, and the establishment
of economic opportunities by selling drugs. Communities may also influence risk behaviors as a function of how well they link the achievement of desired future goals to
the absence of risk behavior^.^' This may be absent in environments where adolescents see few examples of adults who have achieved these desired goals.' These theories may help explain why covariation exists in some individuals but not in others.
Adolescents who see few opportunities to participate in conventional adult roles may
be particularly likely to embark on a lifetime risk-taking syndrome.'
Macrosystems: The Impact of Culture
From the macrosystem perspective, risk-taking behavior rates vary in different
cultural, economic, and political contexts. For example, despite similar ages of sexual debut, the United States has the highest rates of adolescent childbearing and abortion in the developed world.' This is thought to be related to differing normative
views about adolescent sexuality and contraception. Similarly, rates of contraceptive
use have been found to vary significantly by ethnicity and religious affiliation,'" al-
ANNALS NEW YORK ACADEMY OF SCIENCES
6
though the former may be confounded by factors related to socioeconomic status, another macrosystem variable.
Eating behaviors vary markedly by culture-black youth eat more fat food and
food with higher sugar control than does white youth.RThe history of research on
race ethnicity and poverty points to a frequent confounding of the
While we
generally consider impoverished environments to be associated with higher levels of
adolescent risk taking, we actually know little about risk taking in poor adolescent^.^^
Cultural factors, while viewed as critical, also have not undergone the level of study
that is required. For example, little research has explored the meaning of risk taking
in different cultures.'8~6s
BIOLOGICAL MODELS OF RISK-TAKING BEHAVIOR
Biological models of risk-taking behavior have considered the roles of genetic
factors, neuroendocrine processes, including hormonal influences, and the timing of
pubertal events. The first two of these are hypothesized as having direct effects on
behavior. Genetic predisposition has been implicated in adolescent alcohol abuse.
The biological children of alcoholic parents are more likely to later abuse alcohol
than are the children of nonalcoholics, even when they are separated from their pare n t ~Neuroendocrine
.~~
factors are viewed as the etiological agent in some sensationseeking theories.25The most familiar model positing direct effects of biology on behavior comes from theories about the role of sex hormones in behavior. Udry and his
colleagues have demonstrated the importance of testosterone levels to heterosexual
intercourse and the onset of other risk behaviors in
In females, the link between testosterone and sexual behavior appears to be mediated by the social environment.67.6xThe effects in nonwhites have yet to be e ~ t a b l i s h e d . ~ ~
Biological factors may also have indirect effects on risk taking.69This has been
most clearly addressed in the area of hormonal influences. The onset of puberty,
which is often seen as the onset of adolescence, is driven by hormonal changes. Puberty is associated with major and rapid transformations in body shape, size, and appearance. The youngster who looked like a child now looks like an adult. These physical changes may be associated with differences in how the social environment responds to the adolescent, including different expectations for b e h a v i ~ r ,and
~ ~ some
.~~
research supports this, showing an association between pubertal onset and changes in
family interaction~,4~.~'
parental feelings,6O and peer expectation^.^^
The adolescent's response to the physical changes of puberty have also been implicated in risk taking. In particular, decrements in self-esteem and depressive responses to puberty in females, combined with possible neuroendocrine changes, have
been hypothesized to contribute to the increased incidence of depression during the
adolescent year^.^^.^^ There is also research indicating that pubertal onset is associated with changes in peer relationship^,^^ patterns of intimacy,75and opposite sex intere s t ~To
. ~the
~ degree that these factors also predict risk taking in adolescents, this
would support a role for the indirect effects of biological factors.
The timing of pubertal maturation has also been postulated as indirectly influencing risk taking.i0.7" Normal ranges of pubertal onset for females are between the
ages of 8 and 13 years; for males, the range is between 9.5 and 13.5 years. The
''
IRWIN et al.: ADOLESCENT RISK-TAKING BEHAVIOR
7
biopsychosocial model l o considers the onset and timing of pubertal development in
terms of its effects on psychological and social factors such as self-perceptions, perceptions of the social environment, and personal values. According to the model, the
biological changes trigger the psychological and social changes, which themselves
have both direct and indirect effects on risk taking; the impact of these effects on the
adolescent will depend in part on the age of the adolescent.
Some elements of the model can be supported. For example, the effects of early
maturation in females include greater dissatisfaction with ones physical appearance,
greater general unhappiness, lower self-esteem, and lower levels of educational
a ~ h i e v e m e n t .Socially,
~ ~ . ~ ~ early-maturing girls have greater social prestige and popularity among older maless4 The adult physical appearance of these girls, and their
popularity with older boys, is accompanied by earlier heterosexual interests and behavior, including an earlier age for the onset of sexual i n t e r ~ o u r s e .Early-matur~~.~~
ing boys also appear to be prone to early onset of risky behavior^.^^.^'
The biopsychosocial model (see FIG. I), although not fully tested, recognizes the
importance of direct and interactive effects of multiple factors. The model integrates
the biological aspects of development with the social environment and the dispositional characterization of the organism. It is a disposition model in that individual
differences such as self-esteem are viewed as moderating the response to puberty. It
is most definitely an ecological model, in that changes in the social environment are
viewed as important cofactors in the onset of risk taking, interacting with individual
differences. It is a biological model; the effects of the social and environmental
changes are activated by pubertal change. Additionally, the model is appealing from a
developmental perspective, viewing risk taking within the context of the developmental changes of adolescence. We turn now to a discussion of important developmental issues.
THE NATURE OF ADOLESCENCE
Adolescence is a period of risks and opportunities. The dramatic biological
changes that accompany this transition are essentially the same as they have been for
the past several centuries, but the social context in which they occur is very different
from earlier times and continues to change rapidly. Changes in the family, economic
structure, the media, and the community have all affected the way adolescents live
and interact with peers and with the rest of society. Within this social context, adolescents need to find ways to develop a vision of the future, to formulate an image of
what adulthood offers and demands, and to work out a perception of opportunity and
paths toward practical implementation of such opportunity.
Developmental Issues of Adolescence
Adolescence does not occur in a vacuum. It is highly dependent upon what precedes it and likewise adulthood is highly influenced by what occurs during adolescence. The environmental context is critical to understand the process of adolescence. Adolescence is a developmental transition in which there is considerable
8
ANNALS NEW YORK ACADEMY OF SCIENCES
change, both within the individual and within the social context. In fact, there are two
developmental transitions during the second decade of life: the first being the transition into adolescence from childhood and the second being the transition from adolescence into adulthood. These dramatic biologic, psychosocial, and environmental
changes are elucidated below.
Biological Development during Adolescence
Biological development characterized by the rapid hormonal, physiological, and
somatic changes of puberty is dramatic and interwoven with the other aspects of maturation. With the exception of sexual differentiation during fetal growth and hormonal changes during senescence, there is no other period in the life-span development in
which such significant hormonal and biological changes take place. The onset and
duration of these changes have broad and different ranges both between and within
genders. There are also geographical and racial differences.
The somatic manifestations of puberty begin in females between the ages of 8 and
13, with completion at 13 to 18 years.81For males, the initial somatic changes occur
between the ages of 9.5 and 13.5, with completion at 13.5 to 19 years of age.82Endocrine studies have documented that the changes in the central nervous system that
control the onset of puberty begin around 7 years of age.83
Psychosocial Significance of Puberty
Perhaps more important than the actual somatic changes themselves are the responses of the self and others to physical changes. Adolescents are acutely aware of
the changes taking place within their bodies.
Two aspects of psychological functioning may be influenced by puberty: pubertal
status and pubertal timing. Pubertal status is defined by the changes that all individuals experience as they undergo the physical changes. Pubertal timing is the timing of
the changes relative to same-aged peers. Pubertal status changes have been studied as
a direct-effect phenomenon and an indirect-effect phenomenon mediated by social
and psychological responses of the individual or of others in the social environment.
Pubertal development influences adolescents’ satisfaction with their appearance.
For boys, physical maturation leads to improved body image, probably because increased size and muscular development enhance their social status. For girls, physical maturation leads to greater dissatisfaction with their a p p e a r a n ~ e . ~ ~ . ~ ~
There is some evidence that hormonal changes are related to emotional states
such as anger and depression.86
Some investigators have also found relationships between timing of puberty and
psychosocial functioning. Most of these studies have attempted to look at the effects
of early or late timing and compare with those adolescents who are progressing “on
time.” (It should be noted that the definitions of early and late are not consistent
across studies. In addition, most of the research has been done in the female population because of the well-defined self-report marker of menarche for pubertal maturation in females.) Early physical maturation in girls is associated with greater dissatisfaction with physical appearance, greater unhappiness, lower self-esteem, and lower
levels of educational a c h i e ~ e m e n t . ~
Although
~ , ~ ~ . ~their
~ accelerated maturation gives
these girls greater social prestige and popularity among older males, it is accompa-
IRWIN et al.: ADOLESCENT RISK-TAKING BEHAVIOR
9
nied by a decrease in recognition from other females.89 Later physical development
in girls is associated with better affective adjustment.”’?
Less is known about the effects of pubertal timing because of the lack of validated
and reliable measures for males.93 Most investigators have portrayed late-developing
boys at a social disadvantage: they complete their physiological development after all
other chronologically similar age p e e r ~ . I ~ , ~ ’ . ’ ~Late-developing
.’~
males are more
likely to have ectomorphic body types and a negative self-concept, particularly
around body-image types.96In addition, these boys have a greater incidence of major
identity crises. Behaviorally, these boys also have been described as impulsive and
rebelli~us.~~,~’
There are some data that raise the issue of the adverse effects of both early development for males and females.98
Several investigators continue to investigate the specific effect of hormones on
behavior. Pubertal hormones appear to affect the behavior of boys, especially aggresion.^^.'^ Udry and his colleagues have been attempting to identify the contribution
of pubertal hormones to adolescent risky behavior (in particular, onset of sexual behavior). Testosterone appears to increase the interest in sexual behavior for both sexes; however, males are more likely to actually initiate sexual
More recent
work by most investigators postulates a model that looks at the mediation of social
and psychological factors by biological mechanism^.^^
Psychosocial Development
In psychosocial dimensions, normal adolescent development encompasses increasing independence, autonomy from the family, greater peer affiliation and importance, sexual awareness, identity formation, and increasing cognitive capacity with
the ability to process complex information.
In early adolescence, the peer group begins to fulfill the needs of separation. This
increased identification has special relevance for risk taking because peer pressure is
well established as a factor in the onset of risk behaviors (e.g.. sexual behavior and
substance use). Peer pressure as early as 10 years of age is operational in encouraging
young people to participate in dangerous activities including sexual behavior and aggressive behaviors in males such as fighting.Io0
In middle to late adolescence when issues associated with mastery, autonomy, and
individuation are operational, many activities associated with dangerous outcomes
are pursued including substance use and motor vehicle use. In order to achieve mastery, adolescents may attempt to test or verify their physical strength and limits by engaging in risky behaviors.
Cognitive functioning undergoes a major developmental shift during adolescence
with the onset of more complex abstract processing of information and formal operations. There is increasing evidence that brain growth continues into adolescence. Although the number of neurons does not increase, there is proliferation of the support
cells that brace and nourish the neurons. In addition, myelination continues at least
until puberty, establishing faster neural processing. The number of interconnections
between adjacent neurons decreases in the second decade of life, probably reflecting
the disappearance of redundant or inappropriate neural connections. This fine-tuning
of the neural system coincides with the development of formal operational thought.
Although many adolescents will have the ability to reason abstractly and consider
10
ANNALS NEW YORK ACADEMY OF SCIENCES
cause-and-effect relationships, they also have had little experience in applying these
skills to decisions in a more autonomous manner. For the young adolescent, this
translates into a belief in the power and possibilities of thought itself, in which possibility is secondary to reality. One effect of this cognitive immaturity is what Elkind
calls cognitive egocentrism, which includes the inability to recognize ones’ similarities with other people.I0l If the adolescent believes that he or she is not subject to the
same laws of chance as others, perceptions of invulnerability result, affecting the
adolescent’s perceptions of the risk associated with specific behaviors.’’*
There is an increasing ability for adolescents to think hypothetically, apply formal
logic, and use abstract concepts.Io3 Thinking becomes more relative and less absolute, as well as more self-reflective.Io4Adolescents are also increasingly capable of
considering an extended time perspective rather than being tied to the here and now.
Cognitive capacity as measured by tests is quite different from typical performance. Io3 Laboratory studies optimize conditions so that maximal performance levels can be attained; everyday situations are usually more time-pressured and often
personally stressful. Perhaps for this reason, spontaneous processing, compared to
that in the laboratory, is typically less systematic and reflective and depends extensively on prior knowledge. In particular, thinking on unfamiliar or emotionally arousing topics is likely to be less sophisticated.loSUnfortunately, many of the health-related decisions that adolescents confront, such as those surrounding substance use,
driving, and sexual behavior, involve such “hot” cognition. Under such conditions,
adolescents are less likely to exercise their capacity for abstract, formal reasoning.
Decision-making ability changes as young people proceed through adolescence:
there is an increase in awareness of possible risks, consideration of future consequences, and there is a tendency to consult with independent experts. By middle adolescence, most adolescents are able to reason as well as adults with similar reasoning
flaws.
Environmental Context
As one examines the specific cognitive changes, psychosocial changes, and biological changes that characterize the adolescent period within the changing environmental context, the emergence of risk taking is not surprising. With the developmental task and biological change providing the push and cognitive abilities being immature (and less practical) accompanied by lack of experience with the behaviors, the
adolescent period emerges as a critical developmental period in the life cycle for the
onset of life-long risky behaviors and the associated negative outcomes.
The psychological and biological issues are of critical importance to the developing young person; however, these biological and psychological processes occur within the framework of a rapidly changing environment. The environmental context of
adolescence has undergone major changes. There has been a deterioration in the
structure and economic support of many more single females heading households
with less income. Associated with these changes in the family, there has been a decrease in effective monitoring of adolescents due to decreased resources within communities and lack of adult time to provide effective supervision. This lack of supervision may have effects on the development of risky behavior. In addition, there has
been decreasing access to the health care system, with adolescents and young adults
IRWIN et al.: ADOLESCENT RISK-TAKING BEHAVIOR
11
having the highest rates of no insurance coverage of any age cohort in the United
States.Iob School districts in urban centers have deteriorated and there are decreased
opportunities for vocational training. Over the past two decades the number of adolescents involved in the juvenile justice system has increased, with increasing numbers being incarcerated or in foster care settings. I n 7
MORBIDITY DURING ADOLESCENCE
To further elucidate the concept of health-damaging behaviors within the context
of the risk paradigm, morbidity during adolescence will be reviewed.
In general, morbidity of a population is generally described by use of physicians,
hospital discharge area, disability data, and chronic conditions. Within the adolescent
population, these data do not adequately capture the health status of the population.
Therefore, for the purposes of this paper, the behaviors that account for the majority
of morbidity in the adolescent population will be discussed as well as health-damaging behaviors that are initiated during adolescence, yet do little damage until adulthood. It is important to note that the majority of morbidity and mortality in the adolescent populations is secondary to health-damaging behaviors. Furthermore, these
behaviors are also responsible for the majority of morbid conditions in the young
adult and adult population.
Sexual Behavior
For the past three decades, adolescents have been initiating sexual intercourse at
an earlier age. The most recent comprehensive survey (1988) showed that 53% of
adolescent females between the ages of 15 and 19 were sexually active and 76% of
adolescent males were sexually
Black adolescents were more likely than
white or Hispanic adolescents to have had sexual intercourse. The proportion of sexually active adolescents increased with age and grade levels. Adolescents who are
younger at initiation of sexual intercourse are more likely to have multiple partners
and are more likely to not utilize contraception con~istent1y.l~~
The most recent data
from the Youth Risk Behavior Survey indicates that 66 percent of adolescents have
had intercourse by their senior year in high school and 50 percent are currently sexually active.*
There are no national data on same-sex sexual behavior among adolescents in the
United States. A study conducted in a New York City high school found that 10% of
females and 9% of males reported same-sex sexual behavior. I '('
Substance Use
Until 1992, the use of most illicit substances had declined in the United States
since 1970 despite the fairly stable availability of those substances. In 1993, we saw
an increase in illicit substance use (lifetime prevalence increased from 6.1 to 6.6%).
In 1995, slightly less than half of high school seniors surveyed reported using an il-
12
ANNALS NEW YORK ACADEMY OF SCIENCES
licit substance at some time in their lives, more than one-third had used marijuana,
and 8% had used cocaine. I * I The declining rate of substance use had been associated
with an increasing perception of risks of harm associated with the use of those substances. Over the past two years, there has been a decrease in the perception of risk
associated with substance use. At the same time, substance use has increased. One
could hypothesize similar processes for the issue in unhealthy eating behavior; if
adolescents do not see risks associated with unhealthy eating, they are more likely to
eat poorly.
Alcohol and tobacco use have remained relatively constant over the past two
decades at consistently high rates. Ninety percent of high school seniors reported
having consumed alcohol on at least one occasion, and a third had consumed five
drinks or more in a row at least once in the past two weeks (a measure of heavy drinking or drinking to intoxication). Two-thirds of high school seniors reported smoking
cigarettes at some time in their lives, and one in five reported smoking cigarettes daily. By the eighth grade, close to half of the students surveyed nationally had experimented with cigarettes, and almost a third had drunk alcohol to intoxication at least
0nce.l I I
Patterns of Substance Use
Experimentation with substances commonly begins in early adolescence, with the
greater proportion of substance use initiated in the mid-to late-teens. The major risk
for initiation of cigarette, alcohol, and marijuana use is completed for the most part
by the age of 20, and for illicit drug use, initiation by 2 1. Those who have not experimented with any of these substances by age 2 1 are unlikely to do so
Longitudinal studies of adolescent substance use initiation and progression have
demonstrated fairly consistent sequences of drug involvement. Specifically, similar
but less-serious substance use has been found to predict subsequent use of more serious substances. Adolescent tobacco or alcohol use usually precedes marijuana use.
The impact of cigarette smoking on the progression to higher states of substance use
is greater for adolescent females than for male youth. Few adolescents will try other
illicit substances without prior use of marijuana.
Patterns of substance use may vary by region of the United States, gender, or ethnicity. Males are more likely to consume alcohol, use marijuana, or use cocaine than
were female students8 Among high school seniors in the United States, Native
Americans have been found to have the highest prevalence rates for cigarettes, alcohol, and most illicit substance^."^ Rates for white students were only slightly less
than for Native-American youth for most substances. Black youth had consistently
lower rates of substance use than whites, and Asian American youth had the lowest
rates overall. Hispanic high school seniors reported more substance use than blacks
and only slightly less than white youth.
Eating Behaviors
The most current data on unhealthy dietary behaviors comes from the Youth Risk
Behavior Surveillance Program.* The data reported here come from a nationwide
IRWIN et al.: ADOLESCENT RISK-TAKING BEHAVIOR
13
sample of high school students. The sample size is 10,904. Approximately one-quarter (27.6%) of all students thought they were overweight. Female students (across all
racial and ethnic groups) were significantly more likely to think of themselves as
overweight. Also, 41.4% of students were trying to lose weight. Across all racial/ethnic and grade subgroups, female students were more likely than male students to
have been trying to lose weight.
Only 17.7% of students had eaten five or more servings of fruits and vegetables
during the day preceding the survey. Male students more often reported consumption
of h i t s and vegetables than their female counterparts. In addition, white students reported consuming more fruits and vegetables than their minority peers.
Approximately two-thirds of students (72.3%) nationwide had eaten two or fewer
servings of foods typically high in fat content during the day preceding the survey.
Females were more likely than males to have eaten foods low in fat. TABLE1 highlights these data.
Delinquent Behavior
The study of adolescent delinquent behavior has been impeded by the lack of reliable and valid information regarding the extent of adolescent involvement in illegal
behavior, and a lack of consensus regarding the definition of delinquency. “Conduct
disorder,” and “antisocial personality disorder,”-psychiatric diagnoses that may
manifest through delinquent behaviors-are sometimes used interchangeably. However for the purposes of this paper, we will treat these psychiatric conditions as potential antecedents to delinquent behavior.
Sources of information on adolescent delinquent behavior include national crime
reports and self-report surveys. The former give arrest data, which may underestimate rates of status offenses and minor infractions of the law, and the latter has no independent source of validation. According to the National Youth Survey, a large majority of adolescents have engaged in some, usually minor, illegal activity. A minority
have committed a more serious offense at least once in the past, and a small percentage repeatedly engage in both serious and nonserious delinquent behavior.’I4 Chronic offenders are more likely than their peers to have initiated delinquent behaviors at
an earlier age. In 1988, of the 1.6 million arrests of U.S. adolescents ages 10 to 17,
TABLE1. Percentage of High School Students Engaging in Dietary Behaviors ( N =
10,904)“
Total (%)
Thought overweight
Trying to lose weight
Not eating fruitdvegetables
Eating high fat foods
27.6
41.4
72.3
39.5
Female (YO)
33.6
59.8
77.3
28.4
Male (X)
22.1
24.3
67.4
49.7
“L. Kann, C. W. Warren, W. Harris, et a/., Youth Risk Behavior Surveillance-US, 1995, Oz
Centers for Disease Control and Prevention, CDC Surveillance Summaries, September 27,
1996 (MMWR 1996; 45 (no SS-4).)
14
ANNALS NEW YORK ACADEMY OF SCIENCES
most were for serious property crimes rather than for violence against others.’lSRecently, however, there has been a relative rise in the arrest rates of 13- to 18-year-olds
for violent offenses, including murder, nonnegligent manslaughter, and aggravated
assault.
Adolescent delinquent behavior generally peaks between 15 and 17 years of age,
and declines thereafter. Male adolescents predominate both in self-report and arrest
data for delinquent behavior. Sex differences are relatively small for adolescent status
offenses (acts committed by minors that would not be considered offenses if committed by an adult, e.g., running away from home, truancy), but are substantial for more
serious offense^.^' The sex differential for arrest rates appears to be narrowing. Between 1965 to 1987, the male-to-female arrest-rate ratios decreased for serious property and violent offenses from 7: 1 to 4: 1 and from 1 1: 1 to 7.5: I , respectively.’I s
Arrest rates for black adolescents far outnumber those for whites. Disparities in
self-reports of serious offenses are much smaller than those typically reported, based
on arrest statistics.115Black adolescents are overrepresented among those living in
urban areas with low family incomes and high crime rates, placing them at increased
risk for engaging in delinquent activity. As a result of the disparities in arrest rates,
40% of adolescents in state-operated juvenile detention facilities are black and 15%
are Hispanic. Recent increases in rates of juvenile detention have occurred for nonwhites only. Between 1985 and 1987, the number of black and Hispanic adolescent
public detainees increased by 15 and 20%, respectively.
Injuries and Injury-related Behavior
Injuries as a whole contribute significantly to overall morbidity within the adolescent population. Injuries constitute a major cause of visits to physician offices and
emergency rooms. Within the National Ambulatory Care Survey, 7.2% of all visits to
physicians were for injuries. Injuries ranked fifth in terms of frequency of diagnostic
categories. Among children younger than 15, injuries ranked as the second leading
cause of hospitalization after respiratory conditions and account for 13.5% of discharges. In the 15- through 44-year-old population, injury ranked as the number one
cause for hospitalization, if one excludes pregnancy. With the exclusion, injuries account for 15% of the discharges in the 15 through 44 category. Data from emergency
room surveys have also documented the vulnerability of the adolescent population:
one of every five 15- to 24-year-old received injury-related care in emergency rooms.
Males once again contribute disproportionately to the emergency room visits, with
one of every four males receiving care versus one in every seven females receiving
care.116
In 1995, among 10,904 high school students surveyed nationally, 22% had rarely
or never used a seat belt. Thirty-nine percent of high school students reported riding
during the past month with a driver who had used alcohol or other drugs before getting behind the wheel of a car8
Injury-related, risk-taking behavior may be seen as a continuum that is influenced
by a number of external and internal factors. The continuum may begin with exploration within a controlled context, for example, using a bicycle or skateboard with
protective gear, or driving a car using a seat belt. Sensation-seeking may increase the
IRWIN et al.: ADOLESCENT RISK-TAKING BEHAVIOR
15
likelihood of taking chances, and thrill-seeking may result in recklessness." Some
have argued that beyond thrill-seeking, risk taking becomes self-destructive behavior.
Slap and others found, however, that there were no similarities between the risk profiles of youth admitted with injuries and those admitted with medical complications
of suicide attempts. Suicidal youth were more likely to have dyshnctional families,
to have attempted suicide in the past, and to have used alcohol, cigarettes, marijuana,
and other drugs in the past. Internal factors influencing injury-risk behavior include
lack of experience, as with young drivers, and lack of cognitive ability to take into account potential negative future consequences when making decisions. External factors include substance use, as with the association of alcohol to motor vehicles and
drowning fatalities among adolescent males. I
COVARIATION OF RISK BEHAVIORS
Studies have increasingly documented that certain health-damaging behaviors begun during adolescence are associated in predictable ways and that involvement in
one type of risk behavior increases the likelihood of becoming involved in other risk
behavior^.^',^^ Osgood and his colleagues found, for example, that covariation among
various types of substance use, and delinquent behaviors among older adolescents
and young adults is due in large part to a general tendency toward "deviance" and to
a lesser extent to unique antecedents of individual behaviors and that influence of one
behavior on another.*OThe following section provides examples of the complex ways
that risk-taking behaviors cluster and interact. From our review of the literature, we
were unable to identify studies that specifically linked eating and risk behaviors.
Vehicle Use and Substance Use
The concurrence of alcohol and motor-vehicle injuries is well established. In approximately half of the motor-vehicle fatalities involving an adolescent driver, the
driver had a blood alcohol level above 10%."8,"yBeyond the well-established relationship of alcohol use to motor-vehicle injuries, data are emerging to support the relationship of alcohol use to other injuries including firearm injuries, burns, and
drowning1'*
Our own data point to the association of substance use and vehicle use. Fourteen
percent of our entire sample reported having used a bicycle or skateboard under the
influence of alcohol or other drugs. Among adolescents in the middle-school setting
(1&14), 7.7% admitted using recreational vehicles under the influence of alcohol
These data underestiand/or drugs. In high school students, the rate was 19.5%.13.120
mate the percentage of risky users because they reflect the percentage among the
sample as a whole and not just among users of the vehicles. With regard to motor-vehicle use, riding with a driver under the influence was reported by 37% of the middle-school students and 58% of the high school students. Driving a car under the influence of alcohol or other drugs was reported by 6.4% of the entire sample. Within
the high school sample, 9.4% reported this behavior.
Bierness and Simpson have also reported that risky driving behavior is also asso-
16
ANNALS NEW YORK ACADEMY OF SCIENCES
ciated with other risky health habits, such as smoking, illicit drug use, or diet, and
lack of sleep.'2'.122Substance Use Covariation
Substance use patterns have been well described in the patterns of substance use
section.
Delinquent and Other Risk Behaviors
Delinquent youths are also at increased risk for contracting STDs. In a San Francisco study, incarcerated youth were compared to high school students.123The former
were more likely to be sexually active, to have initiated sexual activity at earlier ages,
to have had more sexual partners, and were less likely to use condoms than similarly
aged high school students. Adolescent substance use, in turn,has been found to be
associated with involvement in delinquent activities. According to a Bureau of Justice Statistics' survey, illicit substance and alcohol use preceded the commission of
23% of offenses for which detainees under age 18 were arrested. Among male adolescents, delinquency has been found to predict illicit substance use in young adulthood and among females, illicit substance use predicts some delinquent behaviors in
early adulthood (e.g.,
Sexual Behavior, Substance Use, and Other Risk Behaviors
Associations among sexual activity and other risk behaviors have not been as well
documented as in the area of injuries and substance use where outcome data such as
motor-vehicle deaths and hospitalization for injuries are more quantifiable outcome
measures. With the emergence of acquired immune deficiency syndromes (AIDS),
there is a greater focus on this area of covariation.12sResearch indicates that adolescents who are having sex are also engaging in other risk behaviors. ZabinIz6looked at
the association of cigarette smoking and sexual behavior in a sample of 1200 female
teenagers attending 32 contraception clinics. Within this sample that had a 25%
prevalence rate of cigarette smoking, there was an association between onset of
coitus at an early age and less effective use of contraceptives with cigarette use.126
Zabin and colleagues have also shown that at each age, sexually active teens are significantly higher on a substance-use index than virgins.I2'
Jessor and Jessors4 documented the association between early sexual activity and
use of marijuana, cigarettes, and alcohol. With drinking status as a marker for at-risk
youth, they found that 80% of their subjects were marijuana users and better than
50% had initiated coitus.
Analyses from the National Longitudinal Survey of the Labor Market Experience
of Youth show the strong correlation between prior substance use and sexual initiation by age 16.128-129
Even when such covariates as race, religion, parental education,
family structure, and personality (including delinquency and school characteristics)
IRWIN eta/.: ADOLESCENT RISK-TAKING BEHAVIOR
17
are controlled for, there is still a strong association of drug use and sexual behavior.
Early sex is 1.4 times more frequent for boys who have used alcohol or cigarettes or
both than for boys who did not report any prior drug use. It is 2.7 times more frequent
for boys who have used marijuana and 3.4 times more frequent for boys who have
used other illicit drugs. For females, the association is even stronger, with 1.8, 3.5,
and 4.9 times more frequent use of alcohol or cigarettes, marijuana, and illicit drugs,
respectively, for female users than for nonusers. The association is stronger for
whites and Hispanics than for black^.'^'
Additional studies to document the relationship between substance use and sexual
activity are highlighted below. Elliott and Morsel3” reanalyzed data from the
1976-1980 waves of the National Youth Survey (a national probability sample of
2360 adolescents aged 11 to 17 at the time to the first interview in 1976). Among
males in the survey, the percentage of those who were sexually active increased with
increasing involvement in substance use: 10% of boys with no history of substance
use, 23% of boys with a history of alcohol use only, 48% for boys with a history of
combined alcohol and marijuana use, and 72% with a history of multiple illicit
drugs. In these analyses, Elliot and Morse attempted to establish the temporal sequence of sexual activity and substance use. They found that males and females tended to initiate substance use prior to sexual activity: 5 times as many females and 2.25
times as many males initiated substance use prior to sexual activity rather than initiating sexual activity prior to substance use.
Beyond the covariation of substance use and sexual activity, other behaviors covary with sexual a ~ t i v i t y . ’Miller
~’
and Simon132studied the relationship of sexual
intercourse with other adolescent behaviors in a random stratified sample of 2064
white adolescents aged 14-1 7 living in Illinois households. Sexual intercourse was
once again associated with drug use in their data. In males more than females, sexual
intercourse was also related to delinquent activity. Additionally, they found that adolescents who have had sexual intercourse are less likely to aspire to advanced education and less likely to report being very religious.
In our own data on risk-taking behaviors and intention to become sexually active
during the next year, we have found that the number of risk behaviors (e.g.. substance
use, cigarette use, dangerous vehicle use, etc.) reported by white females correlates
positively with their intention to become sexually active.’33Among the entire sample,
sexually active youth engaged in significantly more risk behaviors than confirmed
virgins. There was no significant difference between the sexually active teens and
those in transition in the number of risks engaged in for any of the subjects. Among
white adolescents, males and females in transition had engaged in significantly more
risk behaviors than the confirmed virgins. Among females, engaging in risk behaviors is predictive of intentions to become sexually active for whites, but only age is
associated with sexual activity intentions for blacks. This relationship does not hold
up in black females. A factor analysis of our cross-sectional data on risk behaviors
demonstrates an interrelationship of substance use and other physical risk behaviors
in males, but not in females.’33
Other factors associated with intention to initiate sexual behaviors include knowledge of transmission, beliefs and attitudes regarding sexually transmitted
diseaseshuman immunodeficiency virus (STDsklIV) and the adolescent’s personal
vulnerability for STD, and peer influences, including perceptions of peer norms.
18
ANNALS NEW YORK ACADEMY OF SCIENCES
Among urban high school students surveyed as a part of a STD/HIV intervention
project (mean age 14.6 years), nonsexually active students who anticipated commencing sexual activity over the next 12 months were found to be less anxious about
acquiring STD and HIV, perceived that peers do not believe in preventive health behaviors, including condom use with sexual intercourse, and engaged in more risk behaviors as reflected by the increased use of alcohol and drugs.134
MECHANISMS OF COVARIATION
In the initial section of this paper, we highlighted various theories of risk taking.
In this section we utilize theories of risk taking to propose mechanisms for covariation of risky behavior. Even though some of the behaviors covary, few investigators
have attempted to develop a theoretical framework for consideration of the mechanisms. Jessor has proposed the problem behavior theory as a mechanism to explain
the relationship^.^^.'^^ Udry and his colleagues provided important data on the importance of gonadal steroids in the initiation of coitus in males and heterosocial behavior in fern ale^.^'.^* Recently, Udry has proposed an integrated model of the biological and developmental history of adolescents that affects their predisposition toward problem behavior^.^^ Beyond the work by Udry, few investigators have attempted to integrate both biological and psychosocial factors in developing theories to
explain the onset of risky behaviors during adolescence.
We maintain that risk taking during adolescence is a part of the developmental
process and that a careful understanding of the process of development itself may
give clues to the mechanisms that explain the onset and maintenance of risky behaviors.I0 If one examines the data on behaviors for prevalence, age of onset, gender distribution, and covariation and the nature of factors that generate these behaviors, certain issues regarding development emerge.
First, there is a developmental trajectory with certain behaviors preceding other
behaviors. For example, alcohol use and unhealthy eating behaviors begins early in
adolescence. In fact, the trajectory may be different for certain groups of adolescents,
with substance use being a behavior that is generally begun prior to sexual behavior
in white adolescents. In black adolescents, sexual behavior may begin the initial transitional behavior without substance use. Unhealthy dietary behaviors also begin early
in adolescence with a progression of increasingly unhealthy eating patterns as adolescents traverse the second decade of life.8
Second, the behaviors do not occur in isolation. Adolescents who engage in one
behavior tend to engage in another behavior, especially if the onset of the behavior
occurs early chronologically in a d o l e ~ c e n c eIn
. ~fact
~ Jessor’s problem behavior theory suggests that the adolescent propensity for (or proneness to) risk taking or problem behaviors is a manifestation of their “c~nventionality.”~~
The theory suggests that
adolescents whose values and behaviors conform to conventional societal norms are
less likely to engage in risk-taking behavior, and that problem behavior represents
such a syndrome that adolescents who engage in one type of risk behavior are more
likely to engage in others. Problem Behavior Theory encompasses three systems of
explanatory variables: personality, perceived environment, and behavior. In the personality system, unconventionality is reflected by a lower value on achievement com-
IRWIN el al.: ADOLESCENT RISK-TAKING BEHAVIOR
19
pared to independence, greater attitudinal tolerance for deviance, and lower religiosity. In the perceived environmental system, unconventionafzfy involves the relative
predominance of peer influence over parental influence, lower parental disapproval
of problem behavior, and greater involvement of peers in problem behavior. Unconventionality in the behavior system is determined by a greater involvement in problem behaviors and less involvement in school activities and church attendance. According to the Problem Behavior Theory, the greater the unconventionality across the
three systems, the greater the proneness to engage in problem behavior, such as substance use and sexual activity. Recently, the theory has been expanded to include
such biological factors as genetic predisposition and protective factors that may mitigate risk proneness in otherwise moderate to high-risk youth.*'
Third, there appear to be gender differences among behaviors. For example, injuries are highly correlated with male gender. In addition males are more likely to use
substances and engage in delinquent behavior. Females are most likely to engage in
certain unhealthy eating behaviors, and yet when it comes to fat intake, females engage in less fat intake than males.
Fourth, peers and family play a key role in the onset of the behaviors.
Biopsychosocial Model
Our own investigative work has been based on a biopsychosocial causal modello
that integrates two areas of research that have often been considered separately: ( 1 )
the relationship of biological development to psychosocial processes during adolesc e n ~ e : ~ .and
~ ~ ( 2 ) the relationship of risk-taking behaviors to psychosocial correlates
of these behavior^.^^.'^^,^^^ Please see FIGURE1.
According to our model, biological, psychological, and social or environmental
factors influence adolescent risk-taking behaviors. Specifically the timing of biological maturation directly influences four areas of psychosocial functioning: cognition,
perceptions of self and the social environment, and personal values. According to the
model, biological, psychological, and social or environmental variables, mediated by
perceptions of risk and peer-group characteristics, then predict adolescent risk taking. Biological variables influencing adolescent risk-taking behavior include pubertal
timing, hormonal effects, and genetic predispositions. Psychological variables associated with risk taking include self-esteem, sensation seeking, and cognitive and affective states. Social influences on adolescent risk taking include peers, parents, and
school. This model has been well described in previous sections.
Investigative Work: UCSF Bay Area Teen Health Project
The data reported here constitute a subsample of data from various facets of our
longitudinal research project. For a complete description of our study see Irwin and
Mill~tein.~~~'~~
Our study involves a cohort sequential longitudinal study involving 1306 adolescents from San Francisco County. The sample was racially and ethnically diverse:
29% were white, 18% were black, 17% were Asiaflacific Islander, and 14% were
ANNALS NEW YORK ACADEMY OF SCIENCES
20
.Egocentrism
*Future time perspective
Risk
Perception
.
*Self-esteem
*Body image
*Self-reliance
'Tolerance
*Identity
Perceptions of
Social Environment
.Parental influence *Peer influence
*Parental controls *Peer controls
'Parental supports *Peer supports
-0ptimiatic bias
Controllability
.costs
*Benefits
Risk-Taking
Behavior
4
Chatacteristics
of Pear Group
*Peer age
*Peer values
*Peer behavior
4
I
Values
*Independence
*Affection
*Achievement
FIGURE 1. Causal model of adolescent risk-taking behavior. (Modified and adapted from: C.
E. Irwin, Jr. and S. G. Millstein, Biopsychosocial correlates of risk talung behaviors during
adolescence, J. Adolesc. Health Care (November Supplement) 6: 89S, 1986. C. E. Irwin, Jr.,
Adolescence and risk taking: How are they related? In N. Bell and R. Bell, Eds., Perspectives
on Adolescent k s k Taking, Sage, Newbury Park, Calif., 1993, pp. 7-20.)
Hispanic. An additional 15% of the sample reported being of mixed racial/ethnic
background and 7% reported other backgrounds. Five hundred and sixty-three adolescents were selected for longitudinal study entry, with 202 being followed for two
consecutive years.
Frequency of Substance-use Behaviors
The frequency of adolescents engaging in substance-use behaviors is identified in
TABLE2. Behaviors reported by greater than 50% of the sample include lifetime
prevalence use of tobacco (57% of middle school and 75% of high school); lifetime
prevalence of alcohol use (71 3 % of the middle school sample and 84% of the high
school sample); lifetime prevalence of marijuana (61.4% of the high school sample).
Males were more likely to report use of all substances than females.
Frequency of Sexual Behaviors
Sexual behavior was measured using three questions: sexual intercourse, causal
sex, and unprotected sex. High school students were more likely to have had sexual
intercourse. The rates of sexual activity are listed in TABLE3. Forty-three and nine
tenths percent of the high school sample was sexually active, and 2 1.4% of the middle school sample was sexually active.
IRWIN ei al.: ADOLESCENT RISK-TAKING BEHAVIOR
21
TABLE
2. Percentage of Adolescents Engaging in Substance Use Behaviors"
Tobacco
Alcohol (ever used)
Alcohol (2 1 x month)
Marijuana (ever used)
Marijuana (21 x month)
% Middle Schoolb
% High School'
57.0
71.8
13.8
31.9
9.4
75.4
84.4
42.8
61.4
33.0
"Adapted from Millstein and Irwin.
bn = 561, ages 11-14.
cn = 680; ages 1 4 1 8 .
Frequency of Unhealthy Dieting/Eating Attitudes and Behaviors
We measured a number of behaviors associated with eating and dieting during
adolescence. A number of the questions are similar to those asked by the Centers for
Disease Control Youth Risk Behaviors Surveillance Instrument. Those behaviors include: How much salt do you usually add to your food? How many times per day do
you usually eat foods that are high in cholesterol or fat? How many times per day do
you usually eat foods that are high in fiber? How many times per day do you usually
eat foods that are high in sugar?
In addition, we added a number of questions that were used to identify attitudes
about food and eating. They included the following: I think about dieting; I am terrified about gaining weight; I am preoccupied with the desire to be thinner; If I gain a
pound, I worry that I will keep gaining; I have gone on eating binges where I have felt
that I could not stop; I eat moderately in front of others and stuff myself when they
are gone; I think about bingeing; I eat or drink in secrecy.
The Frequencies of Eating/Dieting Attitudes and Behaviors
The frequencies of eatingldieting attitudes and behaviors are reported in TABLES
4
and 5.
The eating behaviors of adolescents demonstrate a series of unhealthy behaviors
except in the area of salt, in which adolescents report adding very little additional salt
TABLE
3. Percentage of Adolescents Engaging in Sexual BehavioP
Sexual intercourse
Casual sex
Unorotected sex
"Adapted from Millstein and Irwin.
bn = 561, ages 11-14.
cn = 680; ages 14-18.
% Middle Schoolb
% High School'
21.4
8.0
11.0
43.9
19.7
32.6
22
ANNALS NEW YORK ACADEMY OF SCIENCES
TABLE
4. Percentage of Adolescents Engaging in Unhealthy Eating Behaviors
Add salt
Eat high-fat foods
Eat high-sugar foods
Do not eat food high in fiber
YOMiddle School"
% High School*
7.1
57.4
73.8
46.4
9.0
62.2
55.5
66.7
"n = 561; ages 11-14.
*n = 680; ages 1 4 1 8 .
to their diets. Only 7% and 9% of middle and high school students, respectively,
added salt to their diets. In the other three areas of food behaviors, greater than 45%
of adolescents in all age groups and categories of eating behaviors report unhealthy
eating behaviors (see TABLE
4). In the area of eating high-fat and low-fiber servings,
there was an increasing progression of unhealthy behaviors going from early to late
adolescence. In the area of high sugar, there was a slight decrement in the behavior
going from early to late adolescence. With regard to the food behaviors, there were
no statistical differences between males and females. In the area of race and Hispanic
ethnicity, black and Hispanic adolescents report greater amounts of fat intake and
less fiber intake than whites.
As depicted in TABLE5 , there were surprisingly large numbers of adolescents reporting that they were engaging in or concerned about dieting and their weight.
Greater than 20% of adolescents in middle and high school were thinking about dieting and terrified about gaining weight. Nineteen percent of adolescents were preoccupied with being thinner, and 17% were worried about gaining one pound. Greater
than 30% of the adolescents were not satisfied with their current weight. A significant minority (7-1 1%) demonstrated abnormal eating patterns more commonly associated with eating disorders ( e g , cannot stop bingeing, think about bingeing,
eat/drink secretly, eat moderately public and stuff privately). As expected, there were
TABLE5. Percentage of Adolescents with Unhealthy Dieting/Eating Attitudes and
Behaviors
Think about dieting
Terrified about gaining weight
Worry about gaining 1 Ib
Not satisfied with weight
Preoccupied with being thinner
Cannot stop bingeing
Think about bingeing
Eatldnnk secretly
Eat moderately public and stuff private
an = 561; ages 11-14.
"n = 680; ages 14-18.
% Middle School"
% High School*
21.7
25.3
17.3
33.5
19.0
9.2
7.2
6.7
8.8
26.8
28.7
17.8
39.5
19.3
10.4
11.0
4.8
10.8
IRWIN et al.: ADOLESCENT RISK-TAKING BEHAVIOR
23
significant gender differences in the area of attitudes and behaviors about dieting. Females were much more likely than males to be concerned about the behaviors except
in the category of eating behaviors. Of interest in these categories of dieting behaviors is the fact that a significant number of males did report the following: 13% reported that they often thought about dieting (as compared to 34% of girls), 17% were
terrified about gaining weight (as compared with 36% of girls), 13% were preoccupied with thinness (as compared with 25% of girls), and 8% were worried about gaining one pound (as compared with 26% of girls). The self-reported frequencies of eating behaviors looked remarkably similar by gender, with no significant differences
by gender.
Frequency of Injuw-related Behaviors and Covariation
The frequency of adolescents engaging in injury-related behaviors is identified in
TABLE6. Behaviors reported by more than 50% of the sample include drivingriding
over the speed limit, not using a seat belt in a car, and taking chances on a bike or
skateboard. Forty-eight percent of the sample report riding in a car with a driver impaired by drugs or alcohol. A significant minority (14%) report using a bike or skateboard while under the influence of alcohol or drugs. Males were more likely than
girls to report behaviors that placed them at risk of injury (Fc = 77.16, P < 0.001).
Behaviors increasing the risk of injury include greater frequency of use of recreational vehicles, use while under the influence of alcohol or other drugs, and taking
chances while using recreational vehicles. Males also reported higher accident rates
(41.9%) than girls (17.5%; 2 = 39.20, P < 0.001).
Within the early adolescent (middle school population) sample, 3 5% reported riding with a driver who was under the influence of alcohol or some other drug. Among
these adolescents (n = 198), 68% had ridden with an impaired driver on more than
one occasion and 17% had done so on many occasions.
Within this early-adolescence sample, covariation is already well established.
TABLE6. Percentage of Middle- and High-School-age Adolescents Engaging in Injury-related Behaviors"
Taking chances on bikekkateboard
Using bikeiskateboard under influence
of alcohol/drugs
Not using seat belt in car
Drivingiriding in car over speed limit
Riding in car with alcohol- or
drug-impaired driver
Driving car/motorcycle under influence
of alcohol/drugs
"Adapted from Millstein and Irwin.
= 561; ages 11-14.
cn = 680; ages 14-18.
YOMiddle Schoolb
YOHigh SchoolC
67.2
7.7
59.5
19.5
56.6
31.8
35.4
71.9
73.4
58.3
2.9
9.4
ANNALS NEW YORK ACADEMY OF SCIENCES
24
Those adolescents engaging in any given risk behavior were more likely to engage in
other risk behaviors as well. Adolescents who were engaging in dangerous vehicle
risks were more likely to also be using tobacco and/or alcohol or illicit substances. In
fact, only 3.8% of the early adolescent sample who reported dangerous vehicle use
did this as his or her sole behavior.
Covariation: Multiple-risk Behaviors
In an effort to understand the covariation among the risk and eating behaviors in
the population of adolescents we studied, we conducted a series of analyses. First
(TABLE7), the relationship among well-established risky behaviors (sexual activity,
substance use, and dangerous vehicle use), is reported. Second (TABLES), the relationship among unhealthy eating behaviors is reported. Third (TABLE9 and lo), the
relationship among eating behaviors and the traditional risky behaviors is reported.
In the analyses of risk behaviors and eating behaviors, we first looked at the relationship using the eating behavior as the marker (e.g., of those adolescents who are eating poorly and asked the question about engagement in sexual activity substance use
and vehicle use) (TABLE9); then we looked at the relationship using the traditional
risky behavior as the marker (e.g., of those adolescents who are sexually active, and
asked the question about engagement in an unhealthy eating behavior) (TABLE10).
Covariation of Risk Behaviors
Adolescents engaging in any given risk behavior were more likely to be engaging
in other risk behaviors as well (see TABLE7). Adolescents who were sexually active
were also more likely to be using alcohol or marijuana and to be drivinghding under
the influence of alcohol/drugs. Adolescents who used alcohol or tobacco were more
likely to use marijuana.
Covariation of Eating Behaviors Among Adolescents
The covariation among the eating behaviors is reported in TABLE8. There is significant covariation among these eating behaviors.
TABLE7. Covariation of Risk Behaviors Among Adolescents (Percentage of Subjects, N = 1241)
Sex active
Tobacco
Alcohol
Marijuana
Dangerous vehicle use“
Sex Active
Tobacco
Alcohol
Marijuana
Dangerous
Vehicle Use”
100.0
44.7
40.0
56.2
58.0
83.0
100.0
77.8
93.8
82.3
88.0
91.7
100.0
95.3
90.8
74.2
66.5
57.4
IOO.0
71.9
16.7
12.8
12.0
15.6
100.0
“Dangerous vehicle users include subjects who reported driving or riding in vehicles when the
driver is under the influence of alcohol or other substances.
IRWIN er al.: ADOLESCENT RISK-TAKING BEHAVIOR
25
TABLE8. Covariation of Eating Behaviors Among Adolescents (Percentage of Subjects, N = 1241)
Salt
Fat
Sugar
Lack of fiber
Bingeing
Salt
Fat
Sugar
Lack of Fiber
Bingeing
100.0
73.4
100.0
64.8
60.0
66.7
83.1
77.1
23.4
18.6
19.5
100.0
17.5
15.0
11.1
9.1
9.4
100.0
11.3
10.8
11.6
13.2
100.0
75.1
69.3
Covariation of Eating/Risk Behaviors
There was a significant association among the eating and risk behaviors. Those
adolescents who used high salt, high fat, high sugar, little fiber, and binged were
more likely to engage in all the risk behaviors with the exception of dangerous vehicle use. The relationships are remarkably similar independent of which eating behavior is used as the marker. See TABLE9 for the results.
Covariation of RisWEating Behaviors
There was also significant covariation among the risk and eating behaviors. The
strongest association was in the area of the dangerous vehicle use, with the majority
of these adolescents engaging in high fat, high sugar, lack of fiber, and bingeing behavior. Those adolescents who have sex, drink, and smoke had a similar pattern of
high fat and high sugar intake, with approximately 20% of them engaging in lowfiber diets. See TABLE10 for results.
Covariation of Behaviors by Race, Gender andAge
Since many of the risk behaviors addressed in our study begin in adolescence, we
were particularly interested in how the assessments of health risk covaried with gen-
TABLE9. Covariation of Eating/Risk Behaviors Among Adolescents (Percentage of
Subjects, N = 1241)
Salt
Fat
sugar
Lack of fiber
Bingeing
Sex
Tobacco
Alcohol
Marijuana
Dangerous
Vehicle Use"
46.2
40.2
39.2
42. I
38.4
82.9
71.1
70.4
71.9
69.6
83.7
82.0
80.9
78.4
84.3
56.1
50.2
50.2
58.0
51.3
12.9
11.6
9.5
9.7
10.0
"Dangerous vehicle users include subjects who reported driving or riding in vehicles when the
driver is under the influence of alcohol or other substances.
ANNALS NEW YORK ACADEMY OF SCIENCES
26
TABLE10. Covariation of RiskEating Behaviors Among Adolescents (Percentage of
Subjects, N = 1241)
Sex
Tobacco
Alcohol
Marijuana
Dangerous vehicle use"
Salt
Fat
Sugar
Lack of Fiber
Bingeing
11.5
11.3
9.7
10.7
11.3
66.3
63.3
62.1
71.4
74.6
72.8
75.1
65.0
21.8
19.8
18.4
22.6
76.6
10.5
10.3
63.1
66.7
10.5
10.3
85.4
"Dangerous vehicle users include subjects who reported driving or riding in vehicles when the
driver is under the influence of alcohol or other substances.
der and ethnic or racial background. Accordingly, we subjected the assessments of
the following I7 behaviors to factor analysis with varimax rotation:
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Smoking cigarettes
Drinking beer
Drinking wine
Drinking mixed drinks
Using controlled substances (e.g., marijuana, cocaine, LSD)
Motorcycle driving or riding
Using seat belts
Taking physical risks
Having broken bones
Being knocked unconscious
Engaging in dangerous behaviors (e.g., carry a knife, argue with strangers,
seek entertainment in high-crime area)
Eating high-sugar foods
Eating cholesterolhigh-fat foods
Going for dental checkups
Sleeping 8 hours per night
Exercising
Getting into a fight
TABLES11 and 12 highlight the results. For both males and females, five factors
emerged that accounted for 53% of the variance among males and 49% of the variance among females.'38 Each factor had a loading of 0.30 or above.
For males (see TABLE1 I), the first factor to emerge consisted of three items relating to alcohol use (frequency of the use of mixed drinks, wine, and beer) and an item
regarding use of controlled substances (which mostly reflects marijuana use). The
second factor represents a broader range of risk behaviors; in addition to beer and
substance use, it includes cigarette use, hours riding a motorcycle, and frequency of
participating in fights that result in being knocked unconscious. The third factor consists of the amount of fat and sugar consumed each day. The fourth factor includes
seat-belt use, consumption of high-fiber foods, and frequency of dental checkups. Finally, the fifth factor reflects potential for interpersonal violence, including fights
and the number of dangerous behaviors in which the respondent usually engages.
IRWIN et al,: ADOLESCENT RISK-TAKING BEHAVIOR
27
TABLE11. Factor Structure for Males ( N = 72 I )”
I
I1
Mixed drinks
Wine
Beer
Substance use
Fat servings
Sugar servings
Beer
Riding motorcycle
Substance use
Knocked unconscious
v
IV
111
Seat belt
Fiber servings
Dental checkups
Fights
Dangerous
behaviors
Broken bones
“Adapted from Irwin.’3
For females (see TABLE12), a slightly different structure emerged. Alcohol use is
more differentiated for females than for males. For females, beer ingestion is associated with cigarette and controlled substance use. Use of wine and mixed drinks falls
on separate factors, but is also associated with cigarette use. The third factor reflects
interpersonal and physical risk; it includes frequency of fights, number of dangerous
behaviors, enjoyment of physical risk, and broken bones. The fourth factor represents
the same factor seen with males: daily intake of fat and sugar. The final factor for females is a health factor that includes two of the three items in the comparable factor
for males; it includes seat-belt use, exercise, and dental checkups.
We examined ethnic-group differences on these factors separately by sex. For
both males and females we found significant differences in the dietary factor scores
(fat and sugar intake) across ethnic groups. Blacks scored significantly higher on this
factor (showing greater sugar and fat intake) than Asians or whites. In addition,
among males, Hispanics scored higher than whites. Among females, ethnic group
differences were also evident on the health-behavior factor, with whites showing
higher scores than Asians (on seat-belt use, exercise, and dental checkups).
CONCLUSION
Adolescents are engaging in behaviors in early adolescence with a high probability of negative outcomes throughout adolescence and young adulthood. Behaviors
cluster together in specific patterns for females and males. Early in adolescence,
teenagers are capable of making differential assessments of risk behaviors. In spite of
TABLE12. Factor Structure for Females ( N = 691)”
1
~~
11
III
IV
Fights
Dangerous
behaviors
Physical risk
Broken bones
Fat servings
Sugar servings
V
~
Substance use
Beer
Smoking
Wine
Smoking
Mixed drinks
“Adapted from Irwin.’3
Seat-belt use
Exercise
Dental checkups
28
ANNALS NEW YORK ACADEMY OF SCIENCES
this, as adolescents mature both in chronological and physiological age, there appears
to be a tendency to reduce the estimate of risk associated with the behavior.13This
may be associated with experience, advanced cognitive abilities, or with a better understanding of risk. A more likely explanation, consistent with adult literature, is that
as adolescents participate in risky behaviors, they tend to underestimate the risks associated with these behaviors, or certain environmental contexts may enhance or decrease the likelihood of engaging in risky behavior^.^' This change in perception, environment, and context for older adolescents may help explain the increasing fiequency-of-risk behaviors in late adolescence with more negative health outcomes.
The various theoretical models presented in this paper are helpful in explaining
the process of initiation, the covariation, and the trajectory of risk. What is missing
from these models is the fact that decisions about engagement are made in the context of real life experiences and the situational context of the adolescent. Adolescents
make decisions about eating, sexual behavior, substance use, and driving vehicles
based on biological, social, and psychologic phenomena. Understanding these
processes will enable us to develop more effective health-promoting programs for
adolescents.
ACKNOWLEDGMENTS
The authors thank Ms. Duke Mohler and Mr. Jeremy Schiller for their assistance
with the preparation of the manuscript for publication.
REFERENCES
1, IRWIN,C. E., JR. 1990. The theoretical concept of at-risk adolescents. Adolesc. Med. State
Art Rev. 1: 1-14.
2. IRWIN,
C. E., JR., C. BRINDIS,
S. BRODT,
T. BENNETT
& R. RODRIGUEZ.
199 1. The Health of
America’s Youth: Current Trends in Adolescent Health Status and Utilization of Health
Services. Maternal and Child Health Bureau, Health and Human Services Administration, Public Health Service, U.S. Department of Health and Human Services,
Rockville, MD.
3. BLUM,R. W. 1991. Global trends in adolescent health. JAMA 265: 271 1-2719.
4. NEWACHECK, P. W. 1989. Adolescents with special health needs: Prevalence, seventy, and
access to health services. Pediatrics 84: 872-881.
5. ELLIOTT,
D. S. 1993. Health-enhancing and health compromising lifestyles. In Promoting
the Health of Adolescents: New Directions for the Twenty-first Century, S. C. Millstein, A. C. Petersen, and E. 0. Nightingale, Eds.: 119-145. Oxford Univ. Press. New
York.
6. HALPERIN,
S. F., J. L. BASS,K. A. MEHTA,et al. Unintentional injuries among adolescents
and young adults: A review and analysis. J. Adolesc. Health Care. 4: 275-281.
7. KANDEI.,
D. B. & J. A. LOGAN.1984. Patterns of drug use from adolescence to early adulthood: 1. Periods of risk for initiation, continued use and discontinuation. Am. J. Public
Health 74: 660-665.
J. L. COLI-INS,
B. J. WILLIAMS,
J. G . ROSS & J.
8. KANN,L., C. W. WARREN,
W. A. HARRIS,
KOLBEL.
1996. Youth risk behavior surveillance. United States. 1995. In CDC Surveil-
IRWIN et al.: ADOLESCENT RISK-TAKING BEHAVIOR
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
2 1.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
29
lance Summaries, September 27, 1996. Morbidity and Mortality Weekly Report 45(ss4): entire.
IRWIN,
C.E., JR.,Ed. 1987. Adolescent social behavior and health. New Dir. Child Develop. 37: 1-12.
1986. Biopsychosocial correlates of risk-taking beIRWIN,
C. E., JR. & S. G. MILLSIEIN.
haviors during adolescence: Can the physician intervene? J. Adolesc. Health Care
~ ( S U P P ~82s-96s.
.):
TONKIN,
R. S. 1987. Adolescent risk-taking behavior. J. Adolesc. Health Care, 8:
21 3-220.
MILLSTEIN, S. G. & L! IGRA.1995. Theoretical models of adolescent risk-taking behavior.
In Adolescent Health Problems: Behavioral Perspectives, J. L. Wallander and L. J.
Siegel, Eds.: 52-71. Guilford. New York.
IRWIN,
C. E., JR. 1996. Adolescent development and driving. In Reducing the Risks for
Young Motorists, H. Simpson, Ed.: 5 1-70. Youth Enhancement Service. Los Angeles.
&PLAN,H. B. 1980. Deviant Behavior in Defense of Self. Academic Press. New York.
KAPLAN,
H. B., R. J. JOHNSON
& C. A. BAILEY.
1987. Deviant peers and deviant behaviors.
Further elaboration of a model. SOC.Psychol. Q. 50: 277-284.
PETERSEN,
A. C., B. E. COMPAS,
J. BROOKS-GUNN,
M. STEMMLER,
S. EY & K. E. GRANT.
1993. Depression in adolescence. Am. Psychol. 48: 155-168.
BorviN, G. J. 1986. Substance abuse prevention research: Recent developments and hture directions. J. Sch. Health 56: 369-374.
MCCORD,J. 1990. Problem behaviors. In At the Threshold: The Developing Adolescent,
G. S. Feldman and G. R. Elliott, Eds.: 414430. Harvard Univ. Press. Cambridge,
Mass.
DONOVAN,
J. E. & R. JESSOR.1985. Structure of problem behavior in adolescence and
young adulthood. J. Consult. Clin. Psychol. 53: 89&904.
OSGOOD,
D. W., JOHNSTON,
L. D., P. M. O’MALLEY
& J. G. BACHMAN.
1988. The generality of deviance in late adolescence and early adulthood. Am. Sociol. Rev. 53: 81-93.
JESSOR,
R., J. E. DONOVAN
& F. M. COSTA.1991. Beyond Adolescence: Problem Behavior
in Young Adult Development. Cambridge Univ. Press. New York.
ORR,D. P., M. BEITER
& G. INGERSOLL.
1991. Premature sexual activity as an indicator of
psychosocial risk. Pediatrics 87: 141-147.
ZUCKERMAN,
M. 1979. Sensation Seeking: Beyond the Optimal Level of Arousal. Erlbaum. Hillsdale, N.J.
ZUCKERMAN,
M., S. BALL& J. BLACK.
1990. Influences of sensation seeking, gender, risk
appraisal, and situational motivation on smoking. Addict. Behav. 15: 209-220.
ZUCKERMAN,
M. 1990. The psychophysiology of sensation seeking. J. Pers. 58: 313-345.
FARLEY,
F. H. 1991. The type T pcrsonality. In Risk Taking: Causes and Consequences, L.
F! Lipsitt and L. L. Mitnick, Eds.: 371-382.
LIANG,
S. W., J. JEMERIN,
J. TSCHANN,
C. E. IRWIN,JR., D. WARA& W. T. BOYCE.1995. Interaction of cardiovascular reactivity to stress and life events on risk-taking behavior in
adolescent boys. Pediatrics 96: 1101-1 105.
BRONFENBRENNER,
M. 1979. The Ecology of Human Development: Experiments by Nature and Design. Harvard Univ. Press. Cambridge, Mass.
BROWN,B. B. 1989. The role of peer groups in adolescents’ adjustment to secondary
school. In Peer Relationships in Child Development, T. J. Berndt and G. W. Lad4 Eds.:
188-2 16. Wiley. New York.
BROWN,
B. B. 1990. Peer groups and peer culture. In At the Threshold: The Developing
Adolescent, S. S. Feldman and G. R. Elliott, Eds.: 171-196. Harvard Univ. Press.
Cambridge, Mass.
DRYFOOS,
J. G. 1990. Adolescents at Risk: 34. Oxford Univ. Press. Oxford.
30
ANNALS NEW YORK ACADEMY OF SCIENCES
32. COOK,T. D., A. R. ANSON& S. B. WALCHLI.
1993. In Promoting the Health of Adolescents: New Directions for the Twenty-first Century, S. G. Millstein, A. C. Petersen. and
E. 0. Nightingale, Eds.: 339-374. Oxford Univ. Press. New York.
33. MILLSrEIN, S. G., A. C. PETERSEN
& E. 0. N1~;mNmi.E.1993. Promoting the Health of
Adolescents: New Directions for the Twenty-First Century. Oxford Univ. Press. New
York.
34. B ~ R N DT.T ,1979. Developmental changes in conformity to peers and parents. Dev. Psychol. 15: 608416.
35. NEW('OMB, M. D. & F! M. BENTLER.
1986. Risk factors for drug use among adolescents:
Concurrent and longitudinal analysis. Am. J. Public Health 76: 525-53 1.
36. BILLY,
J. 0. G. & J. R. UDRY.1985. The influence of male and female best friends on adolescent sexual behavior. Adolescence 20: 2 1-32.
37. LEVENTIIAL,
H. & F! KEESHAN.
1993. Promoting healthy alternatives to substance abuse.
In Promoting the Health of Adolescents: New Directions for the Twenty-first Century,
S. G. Millstein, A. C. Petersen, and E. 0. Nightingale, Eds.: 26C284. Oxford Univ.
Press. New York.
38. BAIJMRIND,
D. 1987. A developmental perspective on adolescent risk taking in contemporary America. In Adolescent Social Behavior and Health. New Directions for Child
Development, C. E. Irwin, Jr., Ed.: 91-126. Jossey Bass. San Francisco.
D. 1991. The influence of parenting style on adolescent competence and sub39. BAUMRIND,
stance abuse. J. Early Adolesc. 1I: 56-95.
40. HAYI~S,
C. D., Ed. 1987. Risking the Future: Adolescent Sexuality, Pregnancy and Childbearing. National Academy Press. Washington, D.C.
41. MOSIIFR,
W. D. & J. W. MCNALLY.
1991. Contraceptive use at first premarital intercourse:
United States, 1965-85. Fam. Plann. Perspect. 23: 108-128.
42. TURNER,
R., C. E. IRWIN,
JR. & S. G. MILLSTEIN.
1991. Family structure, family processes
and experimenting with substances during adolescence. J. Res. Adolesc. 1: 93-106.
43. Fi.EwELI.INc;, R. L. & K. E. BAUMAN.
1990. Family structure as a predictor of initial substance use and sexual intercourse in early adolescence. J. Marriage Fam. 52: 171-1 81.
44. RICHARDSON,
J. L., K. DWYER,
K. MCG~JIGAN,
W. B. HANSEN,
C. DENT,C. A. JOHNSON, S.
Y. SIISSMAN,
B. BRANNON
& B. FLAY.1989. Substance use among eighth-grade students who take care of themselves after school. Pediatrics 84: 556-566.
S., C. E. IRWIN,JR. & R. TURNER.
1990. Family structure, emotional distancing
45. FORMAN,
and risk-taking behavior in adolescents. Pediatr. Res. 27: 5A.
46. BRINDIS,
C., C. E. IRWIN,JR.& S. G. MILLSTEIN. 1992. The demography of adolescents in
the U.S.: A national profile. In Textbook ofAdolescent Medicine, E. R. McAnarney, R.
E. Kreipe, D. P. Orr, and G. D. Commerci, Eds.: 12-27. Saunders. Philadelphia.
47. HILL,J. 1987. Research on adolescents and their families: Past and prospect. In Adolescent Social Behavior and Health: New Directions for Child Development, C. E. Irwin,
Jr., Ed.: 13-32. Jossey Bass. San Francisco.
48. MII.L.STEIN,
S. G., C. E. IRWIN,
JR. & C. BRINDIS.
1991. Sociodemographic trends and projections in the adolescent population. I n The Health of Adolescents, w. R. Hendee,
Ed.: 1-15. Jossey Bass. San Francisco.
49. FUKSI
ENBERG, F. F. 1990. Coming of age in a changing family system. In At the Threshold: The Developing Adolescent, S. S. Feldman and G. R. Elliott, Eds.: 147-170. Harvard Univ. Press. Cambridge, Mass.
50. N a I l o N ~ lRESEARCH
COIINCIL.
1993. Losing Generations: Adolescents in High Risk Settings. National Academy Press. Washington, D.C.
R. 1. & B. E. RAINES. 1982. Control and prevention of smoking in adolescents: A
5 1. EVANS,
psychosocial perspective. In Promoting Adolescent Health: A Dialog on Research and
Practice, T. J. Coates, A. C. Petersen, and C. Perry, Eds.: 101-136. Academic Press.
New York.
IRWIN er al.: ADOLESCENT RISK-TAKING BEHAVIOR
31
52. WERNER,
M. J. 1991. Adolescent substance abuse. Maternal and Child Health Technical
Information Bulletin. National Center for Education in Maternal and Child Health.
Cincinnati, Ohio.
53. CLONINGER,
C. R. 1987. Neurogenetic adaptive mechanisms in alcoholism. Science 236:
410416.
54. JESSOR,R. & S. L. JESSOR.1977. Problem Behavior and Psychosocial Development: A
Longitudinal Study of Youth. Academic Press. New York.
55. JESSOR,R., C. A. CHASE
& J. E. DONOVAN.
1980. Psychosocial correlates of marijuana use
and problem drinking in a national sample of adolescents. Am. J. Public Health 70:
6 0 4 6 13.
56. JESSOR,
R., F. M. COSTA,L. JESSOR,
et al. 1983. Time of first intercourse: A perspective
study. J. Pers. SOC.Psychol. 44: 608426.
57. PERRY,
C. L., S. H. KELDER
& K. A. KOMRO.
1993. The social world of adolescents: Families, peers, schools, and the community. In Promoting the Health of Adolescents: New
Directions for the Twenty-first Century, S. G. Millstein, A. C. Petersen, and E. 0.
Nightingale, Eds.: 73-96. Oxford Univ. Press. New York.
58. MCGUIRE,
W. 1986. The myth of massive media impact. Savages and salvagings. In Public Communication Behavior, Vol. 1, G. Comstock, Ed.: 173-257. Academic Press.
Orlando, Fla.
59. GADOW,
K. & J. SPRAiwN. 1989. Field experiments of television with children: Evidence
for an environmental hazard? Pediatrics 83: 399405.
60. COHEN,
M., N. E. ADLER,
A. BECK& C. E. IRWIN,JR. 1986. Parental reactions to the onset
ofadolescence. I. Adolesc. Health Care 7: 101-106.
61. CROCKETT,
L. J. & A. C. PETERSEN.
1993. Adolescent development: Health risks and opportunities for health promotion. In Promoting the Health of Adolescents: New Directions for the Twenty-first Century, S. G. Millstein, A. C. Petersen, and E. 0. Nightingale, Eds.: 13-37. Oxford Univ. Press. New York.
62. SPENCER,
M. B. & S. M. DORNBUSCH.
1990. Challenges in studying minority youth. In At
the Threshold: The Developing Adolescent, S. S. Feldman and G. R. Elliott, Eds.:
123-146. Harvard Univ. Press. Cambridge, Mass.
J. M., R. KOHN,G. BOLAN,S. SHiBosKi & N. KRIEGER.
1995. Socioeconomic dif63. ELLEN,
ferences in the STD morbidity rates among black and white adolescents. Am. I. Public
Health 85: 1546-1548.
64. KLERMAN,
L. V. 1993. The influence of economic factors on health related behaviors. In
Promoting the Health of Adolescents: New Directions for the Twenty-first Century, S.
G. Millstein, A. C. Petersen, and E. 0. Nightingale, Eds.: 38-57. Oxford Univ. Press.
New York.
S. G. 1993. A view ofhealth from the adolescent's perspective. In Promoting
65. MILLSTEIN,
the Health of Adolescents: New Directions for the Twenty-first Century, S. G. Millstein, A. C. Petersen, and E. 0. Nightingale, Eds.: 97-108. Oxford Univ. Press. New
York.
66. UDRY,I. R. 1988. Biological predispositions and social control in adolescent sexual behavior. Am. Sociolog. Rev. 53: 709-722.
67. UDRY,J. R., L. M. TALBERT
& N. M. MORRIS.
1986. Biosocial foundations for adolescent
female sexuality. Demography 23: 21 7-227.
68. UDRY,J. R. & J. 0. G. BILLY.1987. Initiation of coitus in early adolescence. Am. Sociolog.
Rev. 52: 841-855.
69. UDRY,J. R. 1994. Integrating biological and sociological models of adolescent problem
behaviors. In Adolescent Problem Behaviors: Issues and Research, R. D. Ketterlinus
and M. E. Lamb, Eds.: 58-172. Erlbaum. Hillsdale, N.J.
70. IRWIN, C. E., JR. & S. G. MILLSTEIN.
1992. Risk taking behavior and biopsychosocial development during adolescence. In Emotion, Cognition, Health and Development in
32
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
ANNALS NEW YORK ACADEMY OF SCIENCES
Children and Adolescents, E. Susman, L. V Feagans, and W. J. Ray, Eds.: 75-102. Erlbaum. Hillsdale, N.J.
STEINBERG,
L. 1981. Transformation in family relations at puberty. Dev. Psychol. 17:
833-840.
SAVIN-WILI.IAMS,
R. 1979. Dominance hierarchies in groups of early adolescents. Child
Dev. 50: 923-935.
PETERSEN,
A. C., B. E. CAMPAS
& J. BROOKS-GUNN,
et al. 1993. Depression in adolescence. Am. Psychol. 48(2): 155-168.
RUTTI~R,
M. 1995. Resilience: Some conceptual considerations. J. Adolesc. Health 14:
6 2 6 6 3 1.
SIMMONS,
R. G., D. BLYTH& E. G . VANCLEAVE.
1979. Entry in early adolescence: The
impact of school structure, puberty and early dating on self-esteem. Am. Sociolog.
Rev. 44: 948-967.
BROOKS-GUNN,
J. 1988. Antecedents and consequences of variations in girls’ maturational
timing. J. Adolesc. Health. 9: 365-373.
BROOKS-GUNN,
J. & E. 0. REITER.
1990. The role of pubertal processes. In At the Threshold: The Developing Adolescent, S. S. Feldman and G. R. Elliott, Eds.: 16-53. Harvard Univ. Press. Cambridge, Mass.
BROOKS-GUNN,
J. & J. A. GRADER.
1994. Puberty as a biological & social event: Implications for research on pharmacology. J. Adolesc. Health 15: 6 6 3 4 7 1 .
BROOKS-G~JNN,
I. & R. L. PAIKOFE
1993. “Sex is a gamble, kissing is a game”; Adolescent
sexuality and health promotion. In Promoting the Health of Adolescents: New Directions for the Twenty-first Century, S. G. Millstein, A. C. Petersen, and E. 0. Nightingale, Eds.: I8&208. Oxford Univ. Press. New York.
FAUS-T, M. S. 1983. Alternative construction of adolescent growth. In Girls at Puberty: Biological and Psychological Perspectives, J. Brooks-Gunn and A. C. Petersen, Eds.:
105-126. Plenum Press. New York.
MARSHALL, W. A. & J. M. TANNER.
1969. Vanations in the pattern of pubertal changes in
girls. Arch. Dis. Child. 44: 291-303.
MARstiAt.L, W. A. & J. M. TANNER.
1970. Vanations in the pattern of pubertal changes in
boys. Arch. Dis. Child. 45: 13-23.
GRUMRACH,
M. M., l? C. SIZONENKO
& M. L. AUBERT.
1990. Control of the Onset of Puberty. Williams & Wilkins. Baltimore, Md.
SHORE,L. 1984. Experiences ofpuberty development. SOC.Sci. Med. 19: 461465.
RYAN,S., S. G. MILLSTEIN
& C. E. IRWIN,JR. 1996. Puberty concerns of early adolescents.
J. Adolesc. Health 19: 145-152.
SLJSMAN,
E. J., L. D. DORN,G. INOFF-GERMAIN,
E. D. NOTTELMANN
& G. P. CHROUSOS.
1997. Cortisol reactivity, distress behavior, and behavioral and psychological problems
in young adolescents: A longitudinal perspective. J. Res. Adolesc. 7: 81-105.
BROOKS-GUNN, J., A. C. PETERSEN
& D. EICHORN.
1985. The timing of maturation and
psychosocial functioning in adolescence. J. Youth Adolesc. 14: 3 4 .
SUSMAN,
E. J., E. D. NOITELMA”, G. INOFF-GERMAIN, D. LORIAUX,
G. CUTLER,
JR. & G.
CHROUSOS.
1987. Hormones, emotional dispositions and aggressive attributes in early
adolescents. Child Dev. 58: 1 1 14-1 134.
PETERSEN,
A. C., M. TOBIN-RICHARDS
& A. BOXER.1982. Puberty: Its measurement and
meaning. I. Early Adolesc. 3: 47-62.
PESKIN,
H. 1973. Influence of the development schedule of puberty on learning and ego
development. J. Youth Adolesc. 14: 191-206.
PEERSEN,A. C. & L. J. C R O C K E .1985.
~ . Pubertal timing and grade effects on adjustment. J. Youth Adolesc. 14: 191-205.
PETERSEN,
A. C., l? A. SARICIANI
& R. E. KENNEDY.
1991. Adolescent depression: Why
more girls? J. Youth Adolesc. 20: 247-27 1.
IRWIN ef al.: ADOLESCENT RISK-TAKING BEHAVIOR
33
93. SctIi.ossmRcim, N., R. TURNER
& C. E. IRWIN, JR. 1992. Validity of self report ofpubertal
maturation in early adolescents. J. Adolesc. Health 13: 109-1 13.
I.
The influence of older adolescents on
94. Bi.yrii, D. A,, J. P. Hi1.1. & C. K. S M Y T I1981.
younger adolescents: Do grade-level arrangements make a difference in behaviors, attitudes, and experience. J. Early Adolesc. 1: 85-1 10.
95. EWHORN,
D. H., J. A. CLAIJSEN,
N. HAAN,M. P. HONZIK& P H. MUSSEN.
1981. Present
and Past in Middle Life. Academic Press. New York.
96. MCNEII.,D. & N. Livsm. 1963. Maturation rate and body build in men. Child Dev. 34:
25-32.
97. JONES,M. C. 1957. The later careers of boys who were early or late maturers. Child Dev.
28: 113-128.
98. IRWIN,C. E., JR. & S. G. MII.I.STI:IN.
1992. Risktaking & biopsychosocial development
during adolescence. In Emotion. Cognition, Health, Development in Children and
Adolescents, E. Susman, L. V. Feagans, and W. J. Ray, Eds.
99. OiwtxJs, D., A. MATISON,
D. SCHAI.I.IN(;
& H. Low. 1980. Testosterone, aggression, physical, and personality dimension in normal adolescent males. Psychosom. Med. 42:
253-269.
100. LEWIS,
C. E. & M. A. LEWIS.
1984. Peer pressure and risk-taking behaviors in children.
Am. J. Public Health 74: 58G.584.
101. ELKIND,
D. 1967. Egocentrism in adolescence. Child Dev. 38: 1025-1034.
102. MILLSTEIN,
S. G. 1993. Perceptual, attributional and affective processes in perceptions of
vulnerability through the life span. In Adolescent Risk Taking, N. J. Bell and R. W.
Bell, Eds. Sage. Newbury Park, Calif.
103. KEATING,
D. P 1990. Adolescent thinking. In At the Threshold: The Developing Adolescent, S. S. Feldman and G. R. Elliott, Eds.: 5&89. Harvard Univ. Press. Cambridge,
Mass.
104. Tumi., E. 1989. Domain specific social judgments and domain ambiguities. MerrillPalmer Q. 35: 89-1 14.
105. FiS(.tiOfT, B. 1991. Risk taking-A developmental perspective. In Risk-Taking Behavior,
J. F. Yates, Ed.: 133-162. Wiley. New York.
106. NEWACHECK, P. W. & M. A. MCMANIIS.
1989. Health insurance status of adolescents in
the United States. Pediatrics 84: 699-708.
Eds. 1993. HardTime, Healing Hands. National Center
107. THOMPSON,
L. S. & J. A. FAKROW,
for Education in Maternal and Child Health. Arlington, Va.
J. D. & S. S I N G H1990.
.
The sexual and reproductive behavior of American
108. FOKKESI,
women, 1982-1988. Fam. Plann. Perspect. 22: 2 0 6 2 1 4 .
109. CENTITRS
FOR DISEASE
CoNi ROI.. 1991. Current trends: Premarital sexual experience
among adolescent women-United States, 1970-1 988. Morb. Mortal. Wkly. Rep. 39:
9 19-9 32.
110. IRWIN,
C. E., JR. & M. A. SHAFER.
1992. Adolescent sexuality: The problem of negative outcomes of a normative behavior. In Adolescents at Risk: Medical and Social
Perspectives, D. E. Rogers and E. Ginzberg, Eds.: 35-79. Westview Press. Boulder,
Colo.
11 1. J O H N s r O N , L. D., P M. O’MAI-IEY
& J. G . BACHMAN. 1996. National survey results on
drug use from the monitoring the future study, 1975-1995. U.S. Department of Health
& Human Services, Public Health Service. Rockville, Md.
112. KANIXL,D. & K. YAMAGUCHI.
1993. From Beer to Crack: Developmental Patterns of
Drug Involvement. 83: 851-855.
113. BACIIMAN,
J. G., J. M. WALLACE,
P. M. O’MAi.i.i:Y, L. D. JOHNSION, C. L. KIIRTII
& H. W.
NElGllBORS. 1991. RaciaVethnic differences in smoking, drinking, and illicit drug use
among American high school seniors, 1976-1989. Am. J. Public Health 81: 372-377.
114. ELLIm, D. S., S. S. AGETON, D. A. HUIZINCA,
et al. 1983. The prevalence and incidence of
34
ANNALS NEW YORK ACADEMY OF SCIENCES
delinquent behavior: 1976-1980. National Youth Survey Report No. 26, Boulder Colorado Behavioral Research Institute.
115. U.S. CONGRESS.
1991. Office of Technology Assessment, Adolescent Health-Vol. 11,
Background and the Effectiveness of Selected Prevention and Treatment Services,
OTA-H-466, U.S. Government Printing Office. Washington, D.C.
A. P MATHENY
& L. PETERSON.
1991. Health conse116. IRWIN,
C. E., JR., M. F. CATALDO,
quences of behaviors: Injury as a model. Pediatrics 90: 798-807.
117. SLAP,G . B., D. F. VORTERS,
S. CHAUDHURI
& R. M. CENTOR.1989. Risk factors for attempted sicide during adolescence. Pediatrics 84: 762-772.
118. CENTERS
FOR DISEASE
CONTROL
A N D PREVENTION.
1994. Update: Alcohol-related traffic
fatalities-US., 1982-1993. Morb. Mortal. Wkly. Rep. 43: 861-872.
119. MAYHEW,
D. R.,A. C. DONELSON,
D. J. BEIRNESS
& H. SIMPSON.
1986. Youth, alcohol, and
relative risk of crash involvement. Accid. Anal. Prev. 18: 273-287.
120. MILLSTEIN,
S. G . , C. E. IRWIN,
JR., N. E. ADLER,L. D. COHN,S. M. KEGELES
& M. M.
DOLCINI.
1992. Health-risk behaviors and health concerns among young adolescents.
Pediatrics 89: 422-428.
121. BEIRNESS,
D. 1996. The relationship between lifestyle factors and collisions involving
young drivers. In New to the Road: Reducing the Risks for Young Motorists, H. Simpson, Ed.: 71-78. Youth Enhancement Service. Los Angeles.
122. BEIRNESS,
D. & H. SIMPSON.
1988. Lifestyle correlates of risky dnving and accident involvement among youth. Alcohol, Drugs Driv. 4: 193-204.
123. DICLEMENTE,
R. L., M. M. LANIER,
P. F. HORAN& M. LODICO.1991. Comparison of
AIDS knowledge, attitudes, and behaviors among incarcerated adolescents and a public school sample in San Francisco. Am. J. Public Health 81: 628430.
124. KANDEL,
D., 0. SIMCHA-FAGAN
& M. DAVIES.1986. Risk factors for delinquency and illicit drug use from adolescence to young adulthood. J. Drug Issues 16: 67-90.
125. MILLER,
H. G., C. F. TURNER
& L. E. MOSES.1990. AIDS, The Second Decade: 147-253.
National Academy Press. Washington, D.C.
126. ZABIN,L. S. 1984. The association between smoking and sexual behavior among teens in
U.S. contraceptive clinics. Am. J. Public Health 74: 261-263.
127. ZABIN,
L. S., J. B. HARDY,
E. A. SMITH
& M. B. HIRSCH.
1986. Substance use and its relation to sexual activity among inner-city adolescents. J. Adolesc. Health Care 7:
320-33 1.
128. Moir, E. L. & R. J. HAURIN.
1988. Linkages between sexual activity and alcohol and drug
use among American Adolescents. Fam. Plann. Perspect. 20: 108-128.
129. ROSENBAUM,
E. & D. B. KANDEL.1990. Early onset of adolescent sexual behavior and
drug involvement. J. Marriage Fam. 52: 783-798.
130. ELLIOTT,
D. S . & B. J. MORSE.1989. Delinquency and drug use as risk factors in teenage
sexual activity. Youth SOC.21: 3 2 4 0 .
13I . ENSMINGER,
M. D. 1987. Adolescent sexual behavior as it relates to other transition behaviors in youth. In hsking the Future: Adolescent Sexuality, Pregnancy, and Childbearing, Vol. 2, Working Papers and Statistical Appendixes, S. L. Hoffeth and C. D.
Hayes, Eds. National Academy Press. Washington, D.C.
132. MILLER,
P. Y. & W. SIMON.1974. Adolescent sexual behavior: Context and change. SOC.
Probl. 22: 58-76.
133. IRWM,C. E., JR.& S. G. MILLSTEIN.
1991. Correlates and predictors of risk taking behaviors during adolescence. In Self Regulatory Behavior and Risk Taking: Causes and
Consequences, L. P Lipsitt and L. L. Mitnick, Eds.: 3-22. Ablex. Norwood, N.J.
134. SHAFER,
M. A. & C. BOYER.1991. Psychosocial & behavioral factors associated with risk
of sexually transmitted diseases, including human immunodeficiency virus infection
among urgan high school students. J. Pediatr. 119: 826833.
IRWIN ef uI.: ADOLESCENT RISK-TAKING BEHAVIOR
3s
135. JESSOR,R. 1984. Adolescent development and behavioral health. In Behavioral Health: A
Handbook of Health Enhancement and Disease Prevention, J. D. Matarazzo, S. M.
Weiss, J. A. Herd, N. E. Miller, and S. M. Weiss, Eds. Wiley. New York.
136. JESSOR,R. 1987. Risky driving and problem adolescent behavior: An extension of problem behavior theory. Alcohol, Drugs Driv. 3: 1-1 1.
137. IRWIN,C. E., JR. & V. IGRA. 1995. Risk and risk-taking in adolescents. In Social Pedi-
atrics, B. Lindstrom and N. Spencer, Eds. Oxford Univ. Press. Oxford.
138. ADLER,N. E., S. G. MILLSTEIN,
C. E. IRWIN,
JR., et al. 1986. Risk factors and health behaviors in adolescents. Society for Behavioral Medicine Annu. Meet., San Francisco,
March.
139. WEINSTEIN,
N. D. 1984. Why it won’t happen to me: Perceptions of risk factors and susceptibility. Health Psychol. 3: 431457.