JOURNAL OF ADOLESCENT HEALTH 2004;35:529.e7–529.e18
REVIEW ARTICLE
Adolescent Alcohol Initiation: A Review of
Psychosocial Risk Factors
JOHN E. DONOVAN, Ph.D.
Abstract: Longitudinal research studies focused on
alcohol use initiation in adolescence were reviewed to
determine which variables function as antecedent predictors or risk factors. Only studies that focused on time-1
abstainers were included. Classes of risk factors examined include sociodemographic, family, peer, personality, and behavioral variables. The most consistent antecedent risk factors for starting to drink in adolescence
were parental and peer approval and models for drinking
and drug use as well as adolescents’ own prior involvement in delinquent behavior. There was little evidence
for gender differences in risk factors for alcohol use
initiation. Secondary analyses of existing longitudinal
data sets are encouraged to examine whether there are
ethnic/racial differences in the risk factors for starting to
drink and to establish those factors that serve a protective
or buffering function, delaying onset of alcohol use in
adolescence. © Society for Adolescent Medicine, 2004
KEY WORDS:
Adolescents
Alcohol use
Gender differences
Longitudinal research
Risk factors
Alcohol use is one of the most common health risk
behaviors among adolescents. According to the most
recent nationwide data (2001), 51% of 8th graders,
70% of 10th graders, and 80% of 12th graders have
From the Western Psychiatric Institute and Clinic, University of
Pittsburgh Medical Center, Pittsburgh, Pennsylvania
Address correspondence to: Dr. John E. Donovan, Western Psychiatric Institute and Clinic, University of Pittsburgh Medical Center, 3811
O’Hara Street, Pittsburgh, Pennsylvania, 15213. E-mail:
[email protected]
Manuscript accepted February 4, 2004.
© Society for Adolescent Medicine, 2004
Published by Elsevier Inc., 360 Park Avenue South, New York, NY 10010
had at least some experience with alcohol [1]. There
are a number of reasons for our concern with these
levels of adolescent involvement with alcohol. First,
alcohol use in adolescence is associated with an array
of other health risk behaviors, including cigarette
smoking, illicit drug use, and sexual behavior [2– 4].
Second, alcohol involvement increases the risks for
adolescents of negative health and social outcomes
such as physical and sexual assault, motor vehicle
crashes, school dropout, pregnancy, and sexually
transmitted diseases [5– 8]. Third, alcohol use in early
or middle adolescence is associated with greater
likelihood of alcohol problems later in life [9 –11]. To
reduce these risks, it is critical to develop a better
understanding of the risk factors for initiating alcohol use in adolescence. Preliminary steps in this
process are to establish what is currently known
regarding these risk factors and to identify what we
need to learn.
Previous reviews of the literature on risk factors
for adolescent drinking have been unsatisfactory for
several reasons. First, they routinely fail to distinguish between cross-sectional correlates of drinking
and antecedent risk factors for drinking. Second,
they fail to distinguish between risk factors for
initiation and risk factors for other transitions along
the dimension of alcohol involvement in adolescence. Third, they assume that if a variable has been
shown to be a risk factor for adolescent illicit drug
use, it is probably a risk factor for adolescent alcohol
use as well (which may or may not be true). All of
these deficiencies have led to confusion regarding
what is or is not known with respect to the risk
factors influencing the initiation of adolescent drinking. The present review explicitly recognizes these
pitfalls and should as a consequence provide a more
1054-139X/04/$–see front matter
doi:10.1016/j.jadohealth.2004.02.003
529.e8
DONOVAN
useful basis for theory building and the design of
preventive interventions. Focusing on risk factors for
the initiation of alcohol use rather than risk factors
for any alcohol use should be more useful for the
design of interventions to prevent or delay the onset
of drinking.
The Definition of Alcohol Initiation
How one defines a drinker is by no means settled in
the larger literature on adolescent alcohol involvement. A variety of different operational definitions
have been used that emphasize different aspects of
the experience as defining characteristics, such as the
amount of alcohol consumed on a single occasion
(e.g., drinking more than a sip or a taste, or having a
full drink of one’s own) or drinking alcohol some
minimum number of times (e.g., more than two or
three times). In the research literature on alcohol
initiation, however, the definition of a drinker that
has been used most frequently is anyone who has
had any alcohol in his or her lifetime. Prenatal
exposure to alcohol is not included. For consistency,
all of the studies included in this review focus on the
transition from lifetime abstention (never use) into
any use of alcohol.
Studies defining initiation on the basis of questions about alcohol use in the past month [12,13] or
past week [14] were excluded because of the likelihood that some of those considered abstainers were
actually drinkers who simply had not had an opportunity to drink within these foreshortened time
frames. Other studies that used significantly higher
thresholds such as initiation of experimental drinking [15], regular drinking [16 –18], “significant”
drinking [19], or drunkenness [20] were also excluded because the sub-threshold group consists of
drinkers as well as abstainers. Lastly, studies by
Kandel [21,22] and Paternoster [23] were excluded
because they dealt with the initiation of use of
distilled spirits, which typically occurs after the
initiation of beer or wine use.
The Definition of a Risk Factor
To be clear, a risk factor is a variable that significantly predicts whether an individual will develop a
disorder or disease. According to Mrazek and Haggerty, to qualify as a risk factor, “a variable must be
associated with an increased probability of disorder
and must antedate the onset of disorder” ([24], p.
127). Similarly, Kraemer et al state that “a risk factor
JOURNAL OF ADOLESCENT HEALTH Vol. 35, No. 6
is a measurable characterization of each subject in a
specified population that precedes the outcome of
interest and which can be used to divide the population into two groups (the high-risk and the low-risk
groups that comprise the total population)” ([25], p.
338). Based on these definitions, in order for a
variable to be considered a risk factor for the initiation of alcohol use, there needs to be evidence of a
statistically significant relationship between the candidate explanatory variable and the onset of adolescent alcohol use (establishing association), and there
needs to be evidence that the explanatory variable
was present before the adolescent’s involvement in
alcohol use (establishing temporal priority).
Cross-sectional data, which have constituted
much of the content of previous reviews, can establish that a variable correlates with alcohol use but
cannot determine the direction of influence between
the two variables. To establish that a variable functions as a possible risk factor for alcohol use requires
the collection of longitudinal data. The present review therefore excludes variables for which longitudinal evidence of risk factor status is lacking. The
strategy used here is similar to that advocated by
Gorsuch and Butler [26]. According to these authors,
studies of initial drug use need to display the following design features: they should be prospective longitudinal studies in which candidate risk factors are
measured before the respondents used the drug and
in which later data are used to identify which individuals went on to try the drug in question.
Inclusion Criteria
Articles included in this review were identified
through searches of electronic databases (PsycINFO,
1974 –February 2003; Medline, 1966 –February 2003)
as well as by combing the reference lists of published
articles in the literature. Based on the preceding
discussion, inclusion criteria to qualify as a study of
the initiation of adolescent alcohol use were the
following: (a) the sample examined had to consist of
middle or high school aged students (studies of
college students were excluded because of the vastly
different social context of these late adolescents/
young adults); (b) the study must have two or more
waves of prospective longitudinal data; (c) lifetime
(ever) alcohol use must have been assessed at all
waves of data collection; (d) analyses had to be
restricted to lifetime abstainers at time-1, that is,
adolescents who had never had a drink of alcohol;
and (e) the analyses had to predict time-2 drinker
December 2004
status (abstainer versus drinker). Drinkers at time-1
should thus have been excluded from the analyses.
Statistically significant time-1 predictors of time-2
drinker status could then be considered potential
risk factors for the initiation of drinking.
Based on these criteria, studies that merely predict
time-2 alcohol use intensity from time-1 explanatory
variables without deleting time-1 alcohol users were
excluded from this review because they included
both time-1 abstainers and drinkers [27–31]. Also
excluded were the numerous longitudinal studies
that predicted time-2 alcohol use while statistically
controlling for time-1 alcohol use. These latter studies are relevant to change in alcohol use or escalation
and do not discriminate between initiation and other
progressions along the dimension of alcohol use
involvement (e.g., from experimental to regular
drinker, or from regular to heavy drinker). They
uncritically assume that any risk factor is relevant for
change at any point along the dimension. The approach taken here is to establish the risk factors for
initiation and to determine their utility for other
alcohol transitions later.
A second type of studies meeting the above inclusion criteria consists of studies using event history
analyses or survival analyses to predict the length of
time from baseline assessment to the initiation of
alcohol use [32]. As above, these analyses should
have been restricted to those adolescents who had
not yet started drinking at baseline, and the variables
being evaluated as potential risk factors should have
been assessed at baseline or before the onset of
drinking.
Classes of Risk Factors to be Examined
Succeeding sections of this review report on the risk
factors identified in the qualifying research articles
and are organized to reflect the following domains of
explanatory variables: sociodemographic, family,
peer, personality, and behavioral risk factors. These
variables reflect or are relevant to a variety of different theoretical approaches to the prediction of alcohol use in adolescence including Jessor and Jessor’s
Problem Behavior Theory [33], Kandel’s Socialization Theory [21], Huba and Bentler’s DOMAIN Theory [34], Catalano and Hawkin’s Social Development
Model [35], Brook and Brook’s Family Interactional
Theory [36], Aker’s Social Learning Theory [37],
Patterson’s Social Context Model [38], Oetting and
Donnermeyer’s Primary Socialization Theory [39],
and Kaplan’s Self-Derogation Theory [40].
ADOLESCENT ALCOHOL INITIATION RISK FACTORS
529.e9
Common theoretical expectations deriving from
these frameworks are that the likelihood of adolescent alcohol initiation is greater if parents drink or
use illicit drugs, if adolescents do not have a close
relationship with their parents, if parents do not
monitor their children’s behavior, if adolescents affiliate with deviant or drinking peers, if they are not
attached and involved with prosocial peers, if adolescents have tolerant attitudes toward deviance or
approve of alcohol and drug use, if they are depressed, anxious, or stressed, if they already engage
in other problem behaviors like delinquent behavior
or smoking, and if they have little involvement in the
conventional institutions of school and religion.
Although the theories cited above are the more
important theoretical frameworks examined in the
field of adolescent drug and alcohol use, few of them
have been put directly to the test of accounting for
the transition from abstinence into alcohol use.
Sociodemographic Risk Factors
Most studies of alcohol use initiation have neither
examined sociodemographic variables as risk factors
nor statistically controlled for them when examining
other potential risk factors. There are thus only a few
studies relevant to gender, age, racial/ethnic background, and socioeconomic status as potential risk
factors for starting to drink. For example, only four
studies have examined gender as a potential risk
factor [41– 44]. Contrary to general expectation, gender is not predictive of the initiation of drinking: only
one of the four studies found that 12–14-year-old
males were significantly more likely than females to
start drinking over the 2-year interval studied [41].
Only two studies examined age as a predictor of
alcohol initiation. In a study of 9-, 11-, and 13-yearolds followed up 3 years later, age related significantly to starting to drink [43]. In another study that
followed up abstaining 12–14-year-old students 2
years later, however, age was not a significant predictor [41]. This difference in results is likely attributable to the fact that at follow-up, all of the adolescents in the second study were in the period of
highest risk for starting to drink (ages 14 –16).
With reference to ethnic/racial background, Asian
American and African-American adolescents tend to
be less likely to drink than white (European American) or Hispanic American adolescents [1]. The longitudinal literature, however, is equivocal about the
utility of ethnic/racial background as a predictor of
alcohol use initiation. In one study of 12–14-year-old
529.e10
DONOVAN
adolescents in 10 southeastern U.S. cities [41], white
adolescents were more likely to start drinking over a
2-year period than were nonwhite adolescents. A
study of fifth graders in Seattle followed up through
age 17–18 years also showed that white adolescents
started drinking at a younger age than nonwhite
adolescents [42]. Later survival analyses of this same
data set in which the nonwhite category was more
finely differentiated found that Asian American
youth had a more delayed onset of alcohol use than
white youth, but African-American, Native American, and Other youth did not [45]. An additional
study of 9-, 11-, and 13-year-olds found that minority
adolescents were as likely as white adolescents to
initiate alcohol use by follow-up [43], a result probably owing to the composition of the minority category in this sample (two-thirds Native American
who are generally more likely to drink than are white
adolescents, and one-third African-American who
are less likely to drink than white adolescents).
Lastly, in the only study to examine socioeconomic background (assessed by perceived parental
education) as a risk factor, it was found to bear no
relation to likelihood of alcohol use onset when
family structure was also included in the model [41].
Family Domain Risk Factors
The family variables that have been found to predict
starting to drink fall into three general categories:
family composition; parental or sibling modeling
and approval of drinking; and parenting and parentchild relationship. Family structure has been examined as a risk factor in only a single longitudinal
study of 12–14-year-olds, in which it was found that
there is a greater risk of alcohol use initiation for
adolescents living with a stepparent than for those
living in intact families [41].
Adolescents’ perceptions of family approval and
modeling of alcohol and drug use have been shown
to predict the initiation of drinking. In a study of 9th
and 10th grade students followed up 2 years later
[46], adolescents were more likely to start drinking if
they reported that their father drank, if their mother
used prescription drugs, or if a sibling used drugs,
but perceived mother’s drinking, father’s and mother’s smoking, and father’s and mother’s marijuana
use did not relate to starting to drink. Perceived
drinking by a parent also related to drinking onset in
a study of seventh grade students followed up
within a year [47]. Perceived parental approval of
teen drinking predicted ninth grade alcohol use
JOURNAL OF ADOLESCENT HEALTH Vol. 35, No. 6
among sixth grade female abstainers [48]. When
parental reports of their own attitudes and behavior
were assessed, parental attitudes remain important
but their alcohol use does not. Among 11–15-yearold adolescents, mother’s heavier smoking, more
positive mother’s and father’s attitudes toward alcohol and drug use, and more positive mother’s attitudes toward smoking predicted alcohol use initiation 1 year later [49]. In this study, neither parent’s
frequency of drinking nor their frequency of marijuana use related to adolescents’ starting to drink.
The likelihood of starting to drink was also found
to relate to adolescent perceptions of greater mother’s and father’s permissiveness and to lower levels
of adolescent identification with the mother in a
study of ninth and tenth grade students followed up
2 years later [46].
Earlier timing of the initiation of alcohol use (soon
after baseline versus in later waves) was found to
relate to lower initial levels of parental support and
to perceptions of less parental disapproval of teen
drinking in a four-wave study of seventh to ninth
graders [50]. Similarly, survival analyses have shown
that the greater the level of perceived parental alcohol and drug use [42,51] and the more permissive the
parents’ norms regarding teen drinking [45], the
younger the age of alcohol use onset. These results
parallel those derived from the studies predicting
onset versus no-onset above.
Peer Domain Risk Factors
Most peer variables examined as potential risk factors for drinking onset reflect peer involvement in
delinquent or drug-using behaviors or perceived
peer attitudes toward drug use, rather than more
qualitative aspects of adolescent-peer relationships.
In a 2-year study of high school students, onset of
drinking was associated with greater peer deviance,
greater friends’ use of legal and illegal drugs, and
greater identification with friends [46]. In a study
involving two waves of assessment within a single
school year, high school seniors who started drinking in the spring reported greater drinking among
their friends the previous fall [52]. Social assertiveness in childhood, which can be interpreted as independence from peer influence, predicts delayed onset of alcohol use, at least up through age 14 [53].
Peer alcohol use and peer marijuana use were two of
the three significant predictors of the initiation of
alcohol use in a study of seventh grade students [47].
High estimates of the prevalence of alcohol use
December 2004
among other students have also been shown to
predict the onset of alcohol use among seventh and
eighth grade abstainers [54,55]. Fisher and Bauman
[56] found that seventh grade abstainers who had a
friend who drank at time-1 were more likely to onset
drinking by time-2. Urberg et al [57] found that it
was best friend’s alcohol use and not the alcohol use
of the larger friendship group that predicted initiation of alcohol use among 6th, 8th, and 10th grade
students.
In a study of the predictors of time to onset of
drinking among seventh to ninth grade abstainers,
significant time-1 predictors included perceptions of
the number of friends who drank and their perceived
approval for drinking [50]. Similarly, age of first
drink was found to be predicted in survival analyses
by level of friends’ alcohol involvement [42,45] and
by friends’ use of tobacco and alcohol [51].
In a study that clustered seventh and eighth grade
adolescents on the basis of their similarity on both
family and peer variables, abstinent teens described
as “problem behavior prone” (who saw their parents
as unconcerned about their alcohol use, who saw
alcohol as available, and whose friends were unconcerned about their alcohol use or who used alcohol)
were more likely to start alcohol use a year later than
were other teens [58].
Personality Domain Risk Factors
In the present review, a broad definition of personality was employed. Thus, the domain of personality
factors includes personal attributes such as values,
beliefs, and expectancies in addition to temperament
factors and affective disorder factors (depression,
anxiety, internalizing disorders). The onset of drinking among abstainers is signaled in this literature by
antecedent personality attributes reflecting lower
levels of conventionality, greater negative affectivity,
greater behavioral under-control, and higher alcohol
expectancies.
The personality risk factors reflecting psychosocial unconventionality that have been linked to the
initiation of drinking are the following. In relation to
the conventional institution of school, lower values
on academic achievement [46,50,59], lower expectations of academic achievement [50,59], and lower
school motivation [60] increase the likelihood of
starting to drink in samples of 9 –10th graders, 7–9th
graders, 7–11th graders, and 9th graders, respec-
ADOLESCENT ALCOHOL INITIATION RISK FACTORS
529.e11
tively. Lower levels of bonding to school also related
to a younger age of alcohol use initiation [42]. More
general indicators of psychosocial unconventionality
were also associated with starting to drink. These
include more tolerant attitudes toward deviant behavior [46,50,59], lower levels of religiosity [44,50],
less of an orientation to hard work [46], greater
rebelliousness [46], and greater rejection of parental
authority [44].
Although alcohol is often used by drinkers as a
means of coping with stress, anxiety, or depression,
there have been few studies of adolescents that have
focused on negative affectivity as a motivation to
start drinking. In a study of 9-, 11-, and 13-year-olds
followed up for 3 years, the number of depressive
symptoms at baseline predicted who would start to
drink [43]. Similarly, alcohol use initiation 2 years
later was significantly predicted by depressed mood
among 9 –10th grade adolescents [46]. With respect
to anxiety, alcohol use initiation over a 3-year interval was found to relate positively to the number of
generalized anxiety symptoms and negatively to the
number of separation anxiety symptoms at baseline
(at ages 9, 11, 13 years), but not to relate to overall
anxiety symptoms at baseline [43].
The evidence linking negative affectivity to initiation of drinking is not univocal, however. One
study found that abstaining high school seniors who
exhibited greater negative affectivity and internalizing symptoms in the fall semester were less (not
more) likely to initiate drinking by spring semester
[52], a contrary finding possibly attributable to their
lower level of involvement with peers. A second
study reports no significant relation between a diagnosis of major depression at age 11 years and the
onset of drinking 3 years later [61]. The nonsignificant odds ratios for both genders in this study are
likely affected, however, by the low prevalence of
major depression at baseline (3.4% for boys, 2.9% for
girls).
As a result of our stringent inclusion criteria, there
are few longitudinal studies that have examined
temperament variables such as emotionality or sociability as potential risk factors for alcohol use initiation in adolescence. In the single qualifying study,
Fleming et al [62] found that teacher ratings of
shyness or lack of social contact in first grade were
associated with delayed onset of beer/wine use in
adolescence among African-American girls and, to a
lesser extent, boys.
Whereas the construct of behavioral under-control
[63] is important in studies of the onset of alcohol use
disorders [64 – 67], its components (risk-taking, im-
529.e12
DONOVAN
pulsivity, sensation-seeking) have not been examined to any degree as antecedent risk factors for
alcohol use onset in adolescence. Only one qualifying
study has demonstrated the longitudinal predictive
utility of sensation-seeking for alcohol use initiation
among seventh grade students [44], and only a single
study has shown that greater impulsivity in 11-yearolds serves as an antecedent risk factor for starting to
drink by age 14 years [61].
Lastly, a number of alcohol-related personality
domain variables have been shown to function as
risk factors for the initiation of drinking. Adolescents
with more reasons not to drink are less likely to start
drinking a year later [59], or they started drinking
later than those with fewer reasons not to drink [50].
Abstainers in seventh grade who agreed more
strongly that drinking enhances social interaction
were more likely to initiate drinking a year later than
those who did not agree with these expectancies
regarding the effects of alcohol use [68]. Starting to
drink was also more common over a 1-year period
among seventh grade students who rated the effects
or consequences of alcohol use as being more likely
and more positive [69].
Behavioral Domain Risk Factors
One of the most consistent behavioral risk factors for
starting to drink in adolescence is prior involvement
in delinquent behavior [44,46,50,52,59]. Abstinent
adolescents reporting more frequent antecedent involvement in delinquent behavior were more likely
to initiate drinking. Kellam et al found in an AfricanAmerican sample that teacher-rated aggressiveness
in first grade, particularly aggressiveness in the
context of having few friends (characterized as “shyness”), was predictive of male drinking as adolescents [70]. McGue et al found that psychiatric diagnoses of conduct disorder, oppositional defiant
disorder, or of any externalizing disorder at 11 years
of age significantly increased the likelihood of starting to drink by age 14 years [61].
Also implicated in longitudinal research as behavioral risk factors for starting to drink are lower
grades in school [59,61], more frequent cigarette
smoking and dieting in sixth grade girls [48], and
greater levels of television viewing and exposure to
music videos among ninth grade students [71].
Lower grades in school also predict starting to drink
sooner rather than later [50].
JOURNAL OF ADOLESCENT HEALTH Vol. 35, No. 6
Gender Differences in Risk Factors
for Initiation
Less than half of the studies reviewed above examined whether there were differences between male
and female adolescents in the risk factors for initiation of drinking. Of the 11 studies that did examine
gender differences, seven performed tests for gender
by risk factor interactions and found them to be
statistically nonsignificant [41,46,49,54,57,58,72]. In
an eighth study, the Jessor and Jessor study predicting the timing of onset of drinking, within-gender
analyses found that 11 of the 12 variables that were
statistically significant in the combined-gender analyses were significant predictors within each gender
group as well [50].
Of the remaining three studies comparing results
by gender, two report only modest differences. Pedersen and Skrondal found that three predictors (parental alcohol consequences, friends’ norms, and
friends’ tobacco and alcohol use) displayed identical
regression coefficients for both genders, but a fourth
variable (parental norms) entered for girls but not for
boys [51]. Marks et al found that perceived prevalence of drinking predicted onset of drinking somewhat more strongly for girls than for boys: of those
adolescents who were above the median in perceived
prevalence of alcohol use at time-1, 32% of the girls
and 29% of the boys started drinking at time-2 (in
contrast to 15% of the girls and 16% of the boys who
were below the median) [55]. In the final study that
performed within-gender analyses, correlation coefficients were reported only if significant, making it
impossible to determine whether the effect sizes
were actually different for the genders [53]. Overall,
then, the evidence suggests that the risk factors for
starting to drink are generally similar for male and
female adolescents.
Age Differences in Risk Factors for Initiation
Given that the prevalence of alcohol use increases
with age throughout adolescence, one might expect
that the risk factors for early adolescent initiation of
use would differ in nature and intensity from those
that predict later adolescent initiation. However,
none of the research studies reviewed here explicitly
addressed the issue of whether the risk factors for
alcohol use initiation differ as a function of the age of
the adolescents. That would require parallel follow-up studies of cohorts of early and middle adolescents assessed with the same measures at baseline,
or following up adolescents of varying ages and
December 2004
looking for age by risk factor interactions in predicting initiation by the follow-up assessment.
Most of the prospective studies examined in this
review started with cohorts of early adolescents
(6 – 8th graders) and followed them up later in the
same school year [47], a year or so later [44,54 –
56,58,68,69], or several years later [41,43,48]. Only
three of the studies started with cohorts of middle
adolescents (9 –12th graders) who were followed up
while still middle adolescents [46,52,71]. Other studies combined early and middle adolescents but did
not examine the interaction of age at baseline with
the other risk factors [49,50,57,59].
It is not really possible to address this issue by just
sorting the reprints into piles based on the age of the
samples at baseline. The main obstacle to this is the
lack of consistency in the variables assessed in the
different studies. Only a few variables were assessed
in studies at two different substages of adolescence.
Among these, peer alcohol use was consistently
important in all studies in which it was included.
Delinquent behavior was a significant risk factor in
both early adolescent samples [44] and late adolescent samples [52], whereas behavioral undercontrol
was significant in early adolescence [44] but not in
late adolescence [52]. Beyond this, definitive statements concerning age differences in adolescent risk
factors must await more systematic examination.
Risk Factors and Theories of Adolescent
Alcohol Use
In an earlier section, a number of theoretical expectations that derive from a variety of frameworks
were listed regarding the factors that should increase
the likelihood of making the transition from abstinence into alcohol use. The preceding sections of this
review provide support for the great majority of
these expectations. The likelihood of adolescent alcohol initiation is greater if parents drink or use illicit
drugs, if adolescents do not have a close relationship
with their parents, if adolescents affiliate with deviant or drinking peers, if adolescents have tolerant
attitudes toward deviance or approve of alcohol and
drug use, if they are depressed or anxious, if they
already engage in other problem behaviors like delinquent behavior or smoking, and if they have less
involvement in the conventional institution of
school. There was no support for the expectation that
those who would start drinking would have less
involvement in the conventional institution of religion [50,59], a result possibly ascribable to the use of
ADOLESCENT ALCOHOL INITIATION RISK FACTORS
529.e13
church attendance as a measure of this. Other expectations simply were not addressed in the qualifying
articles: these include the expectations that the likelihood of alcohol initiation is increased if parents do
not monitor their children’s behavior, if adolescents
are not attached and involved with prosocial peers,
or if the adolescents are stressed.
Although these findings support the relevance of
these factors for the prediction of alcohol use initiation, they cannot be used to determine which of the
theories are supported most strongly by this review.
First, most of the research articles reviewed were not
designed to test specific theories of alcohol use.
Second, there is considerable overlap among these
theories in their components. What distinguishes one
theory from another is often whether variables such
as parent-child relationships have a hypothesized
direct influence on adolescent alcohol use or whether
their influence on initiation is indirect or mediated
through other variables such as affiliation with deviant peers (where a distant parent-child relationship
increases affiliation with deviant peers, which relates
to alcohol use initiation). In the absence of studies
explicitly designed to compare and contrast the
utility of specific theories for the prediction of the
initiation of alcohol use in adolescence, it is impossible for a review such as this to say much about their
relative merits.
Antecedent psychiatric disorders, despite their
low prevalence in the population, also appear to play
a role in the initiation of alcohol use among affected
youth. Although this result seems to contradict the
hypothesis that mental disorders are more relevant
to movement into heavier or problematic drinking
than to initiation [73], it is not necessarily dissonant.
The relationship may be a consequence of the relatively long interval in these studies between baseline
and the assessment of drinking onset. By the time of
the 2–3-year follow-up, those adolescents with antecedent psychiatric disorders may well have been
involved not only in drinking but also in problematic
drinking (which was not measured) as well.
Gaps in the Literature
Risk factor research with respect to the initiation of
alcohol use needs to be instituted in younger, that is,
elementary-school-aged, children. Almost all of the
longitudinal studies reviewed above have been carried out in cohorts of adolescents who were in grade
seven (age 13 years) or older. Yet we know from
epidemiological research that 50% of eighth grade
529.e14
DONOVAN
students in the United States have already started
drinking [1]. Most of what we know about alcohol
initiation in adolescence therefore describes only
“normative onset,” that is, the initiation of alcohol
use at the ages when most adolescents start to drink.
The risk factors for childhood/preadolescent onset of
drinking may well differ from those predicting normative adolescent onset.
Aside from the single report by Zimmerman and
Schmeelk-Cone on school motivation as a risk factor
among African-American adolescents [60], there
have been no longitudinal studies establishing the
risk factors for alcohol initiation in minority (nonEuropean American) adolescents. First, most studies
of minority adolescents have been cross-sectional in
design (or have not yet appropriately analyzed their
longitudinal data), and thus only establish the psychosocial correlates of adolescent alcohol use [74 –
81]. Second, where longitudinal data have been collected on minority adolescents, the data have been
analyzed predicting variation in time-2 alcohol use
without limiting the analyses to time-1 abstainers
[29,82,83], or the analyses predicted a dependent
variable that reflected multiple substance use onset
rather than just alcohol use initiation [84,85]. Studies
are needed that test whether those variables that
serve as risk factors for the initiation of alcohol use
among white adolescents also serve as risk factors for
adolescents from African-American, Hispanic American, or other ethnic/racial groups. Research is also
called for that examines the predictive utility across
cultures of a broad array of potential risk factors.
This information could be used to adapt preventive
interventions for use in environments where it is
feasible to present interventions tailored for adolescents of specific ethnic/racial, cultural, or national
backgrounds.
A further issue that needs to be addressed is
whether the variables identified here as risk factors
for initiation also function as risk factors for other
transitions along the dimension of adolescent alcohol
involvement. Although a number of theories focus
on the explanation of adolescent involvement in
alcohol and drug use, few have paid attention to
developing models of risk factors for specific transitions along the dimension of alcohol involvement
(e.g., for initiation, or from experimentation into
regular drinking). Problem Behavior Theory suggests that the same variables would predict all transitions along this dimension [33]. According to this
framework, it is “transition-proneness,” as reflected
in more unconventional or problem-behavior prone
scores on the personality, perceived social environ-
JOURNAL OF ADOLESCENT HEALTH Vol. 35, No. 6
ment, and behavior variables at time-1, that increases
the likelihood that an adolescent will engage in
transition-marking behavior by time-2. According to
Socialization Theory, on the other hand, variables
such as parental drinking and parental attitudes
toward adolescent drinking should function as predictors of the transition into drinking, whereas peer
influences would predict higher-level transitions
such as that from regular drinking into problem
drinking [21]. Newcomb and Bentler suggested that
social influences are most important in the explanation of drug use, but that drug abuse is more strongly
tied to internal, psychological processes such as
self-medication to combat emotional distress [73].
At the present time, it is difficult to determine
whether the same or different variables operate as
risk factors at different levels of transition along the
dimension of adolescent involvement with alcohol.
First, there are only a few studies of the risk factors
for specific transitions in adolescent alcohol involvement beyond initiation [15–19,52,86 –93]. Second,
comparison between studies of initiation and studies
of other transitions is difficult because of sample
differences (e.g., ages, grades included), study design differences (e.g., intervals between data waves),
and the lack of assessment of a consistent array of
candidate risk factors for examination. Third, there is
no consensus in the field regarding the operational
definitions of these higher stages of alcohol involvement in adolescence. To date, stages of alcohol involvement have been defined arbitrarily and nonempirically. Establishment of replicated stages along the
dimension of alcohol involvement is a prerequisite
for determination of similarities and differences in
risk factors along this dimension.
All of the variables identified in this review as risk
factors had to demonstrate at least a significant
bivariate relation to later onset of alcohol use. It
should be recognized, however, that there are substantial correlations among many of these antecedent
variables that may influence whether they retain
their utility as longitudinal predictors in larger, more
extensive multivariate models or in tests of etiological theories. When a variable with a significant
bivariate relationship to initiation becomes nonsignificant when it is considered as part of a larger
regression model (assuming other variables measuring the same construct have already been removed),
its relation to initiation is said to be “mediated”
through other variables in the model. Analyses to
clarify which variables mediate the impact of other
variables are rarely explored in this literature despite
their relevance to the explanation of the mechanisms
December 2004
or processes through which risk factors affect initiation.
Not much is currently known about variables that
serve as protective or buffering factors against the
adolescent initiation of alcohol use. According to one
definition of protective factors, as variables that
relate negatively to alcohol use, many of the variables described above could be considered protective
factors. The problem with calling variables protective
factors just because they correlate negatively with
alcohol use is that they can be converted to risk
factors simply by reverse scoring them. Rutter [94]
proposed a second, more restrictive definition,
widely used in the field of developmental psychopathology [24,95–98], that states for a variable to qualify
as a protective factor, it must significantly reduce the
likelihood that an individual who is at risk for an
outcome will experience that outcome. In statistical
terms, this type of protective factor “moderates” the
relationship between a risk factor and an outcome or
“buffers” the impact of risk factors on the individual
(as shown by a significant multiplicative interaction
term between the candidate protective factor and a
risk factor in a regression model). When the protective factor is present, there should be considerably
less alcohol use or a lower likelihood of alcohol use
than would be expected given the risk factors that
are also present.
Only one of the studies reviewed here performed
the interaction analyses to determine whether selected variables moderate the impact of the other risk
factors examined. Brook et al used two-way analyses
of variance to examine personality and family variables as potential protective factors against peer risk
factors [46]. In this study, a higher achievement
orientation, less maternal rejection and permissiveness, greater identification with parents, and higher
family expectations buffered the impact of more
deviant or drug-using friends on alcohol use initiation at time-2. The problem in this area is not that
potentially protective variables have not been assessed, it is that analyses have generally not been
undertaken to establish whether they perform this
important buffering function.
Concern with establishing protective factors folds
into the larger issue of encouraging the development
and testing of more complex models of risk that
include mediating as well as moderating variables in
multiple-wave longitudinal models. Mediating variables would reveal the pathways by which some risk
factors indirectly affect initiation of alcohol use,
whereas moderating variables would indicate where
relations between risk factors and initiation differ in
ADOLESCENT ALCOHOL INITIATION RISK FACTORS
529.e15
magnitude, direction, or form as a function of third
variables (including sociodemographic factors or potentially protective factors). To date, there have been
few such longitudinal models focused on adolescent
drinking initiation. With the continuing development and refinement of statistical analysis programs
to perform latent transition analysis [99], hierarchical
linear modeling [100], structural equation modeling
and latent growth curve modeling [101–103], and
latent variable mixture modeling [104], the possibilities for capturing the complexity of adolescent development are enhanced. Use of these methods
should permit determination of whether a variable is
a causal risk factor, a correlate, or a mediator of other
risk factors [105,106].
Although several of the concerns expressed here
do require the design and collection of new data,
most can be ameliorated through re-analyses of
existing longitudinal data sets. Secondary analyses
could profitably be pursued in the larger of these
longitudinal data sets to develop a better understanding of the natural history of adolescent drinking, to determine the risk factors for transition into
alcohol use, and to evaluate the cross-ethnic or
cross-national utility of various frameworks of variables for the prediction of transition into alcohol use.
This may require greater federal support for the
archiving of longitudinal data sets collected in earlier
years.
A major limitation of the present review of the
literature is that many of the variables examined as
risk factors were assessed in only a single study.
Further research is needed to determine whether
these risk factors demonstrate the same antecedent
predictive function when measured in other ways in
other samples in other geographic areas at other
times. There are, nevertheless, a number of variables
that have been shown to be risk factors for starting to
drink in several different longitudinal studies. Variables with the best support for their status as risk
factors include the following: perceived parental
drinking, parental approval for adolescent drinking,
perceived friends’ alcohol and drug use, and selfreported delinquent behavior.
With regard to the role of aspects of the larger
social environment as risk factors for alcohol initiation, efforts need to be made to incorporate the
influence of community norms, the neighborhood,
the school context, the influence of the mass media,
as well as the structures of economic opportunity,
alcohol access and availability, and of legal sanctions
and enforcement into research on alcohol use initiation. Despite their obvious importance, these poten-
529.e16
DONOVAN
tial domains of risk have not figured thus far in
longitudinal investigations of the transition from
abstention into alcohol involvement among adolescents.
It is to be hoped that future research in the field
will begin to develop the knowledge that is currently
missing regarding the etiology of alcohol involvement among adolescents. Only through the development of a greater understanding of these issues will
there be improvements in the effectiveness of preventive interventions for use with adolescents.
This work was supported by NIAAA Grants R01 AA-12342 and
K02 AA-00181.
JOURNAL OF ADOLESCENT HEALTH Vol. 35, No. 6
14. Kaplan HB, Pokorny AD. Self-attitudes and alcohol use
among adolescents. In: Seixas FA (ed). Currents in Alcoholism, Volume II. New York, NY: Grune & Stratton, 1977:285–
98.
15. Dishion TJ, Capaldi DM, Yoerger K. Middle childhood antecedents to progressions in male adolescent substance use: An
ecological analysis of risk and protection. J Adolesc Res
1999;14:175–205.
16. Hill SY, Shen S, Lowers L, Locke J. Factors predicting the
onset of adolescent drinking in families at high risk for
developing alcoholism. Biol Psychiatry 2000;48:265–75.
17. Hill SY, Yuan H. Familial density of alcoholism and onset of
adolescent drinking. J Stud Alcohol 1999;60:7–17.
18. Shen S, Locke-Wellman J, Hill SY. Adolescent alcohol expectancies in offspring from families at high risk for developing
alcoholism. J Stud Alcohol 2001;62:763–72.
19. Smith GT, Goldman MS. Alcohol expectancy theory and the
identification of high-risk adolescents. J Res Adolesc 1994;4:
229 –47.
References
1. Johnston LD, O’Malley PM, Bachman JG. Monitoring the
Future National Survey Results on Drug Use, 1975–2001:
Volume I: Secondary School Students (NIH Publication No.
02-5106). Bethesda, MD: National Institute on Drug Abuse,
2002.
2. Donovan JE, Jessor R. Structure of problem behavior in
adolescence and young adulthood. J Consult Clin Psychol
1985;53:890 –904.
20. Dobkin PL, Tremblay RE, Masse LC, Vitaro F. Individual and
peer characteristics in predicting boys’ early onset of substance abuse: A seven-year longitudinal study. Child Dev
1995;66:1198 –214.
21. Kandel DB, Andrews K. Processes of adolescent socialization
by parents and peers. Int J Addict 1987;22:319 –42.
22. Margulies RZ, Kessler RC, Kandel DB. A longitudinal study
of onset of drinking among high-school students. J Stud
Alcohol 1977;38:897–912.
3. Kulbok PA, Cox CL. Dimensions of adolescent health behavior. J Adolesc Health 2002;31:393–400.
23. Paternoster R. Decisions to participate in and desist from four
types of common delinquency: Deterrence and the rational
choice perspective. Law Soc Rev 1989;23:7–40.
4. Valois RF, Oeltmann JE, Waller J, Hussey JR. Relationship
between number of sexual intercourse partners and selected
health risk behaviors among public high school adolescents.
J Adolesc Health 1999;25:328 –35.
24. Mrazek PJ, Haggerty RJ (eds). Reducing Risks for Mental
Disorders: Frontiers for Preventive Intervention Research.
Washington, DC: Institute of Medicine, National Academy
Press, 1994.
5. Bachanas PJ, Morris MK, Lewis-Gess JK, et al. Predictors of
risky sexual behavior in African American adolescent girls:
Implications for preventive interventions. J Pediatr Psychol
2002;27:519 –30.
25. Kraemer HC, Kazdin AE, Offord DR, et al. Coming to terms
with terms of risk. Arch Gen Psychiatry 1997;54:337–43.
6. Bonomo Y, Coffey C, Wolfe R, et al. Adverse outcomes of
alcohol use in adolescents. Addiction 2001;96:1485–96.
7. Lang SW, Waller PF, Shope JT. Adolescent driving: Characteristics associated with single-vehicle and injury crashes. J
Safety Res 1996;27:241–57.
8. Meropol SB, Moscati RM, Lillis KA, et al. Alcohol-related
injuries among adolescents in the emergency department.
Ann Emerg Med 1995;26:180 –6.
9. DeWit DJ, Adlaf EM, Offord DR, Ogborne AC. Age of first
alcohol use: A risk factor for the development of alcohol
disorders. Am J Psychiatry 2000;157:745–50.
10. Grant BF, Dawson DA. Age at onset of alcohol use and its
association with DSM-IV alcohol abuse and dependence:
Results from the National Longitudinal Alcohol Epidemiologic Survey. J Subst Abuse 1997;9:103–10.
26. Gorsuch RL, Butler MC. Initial drug abuse: A review of
predisposing social psychological factors. Psychol Bull 1976;
83:120 –37.
27. Barnes GM, Farrell MP, Dintcheff BA. Family socialization
effects on alcohol abuse and related problem behaviors
among female and male adolescents. In: Wilsnack RW,
Wilsnack SC (eds). Gender and Alcohol: Individual and
Social Perspectives. New Brunswick, NJ: Rutgers Center of
Alcohol Studies, 1997:156 –75.
28. Boyle MH, Offord DR, Racine YA, et al. Predicting substance
use in early adolescence based on parent and teacher assessments of childhood psychiatric disorder: Results from the
Ontario Child Health Study Follow-up. J Child Psychol
Psychiatry 1993;34:535–44.
29. Gil AG, Wagner EF, Vega WA. Acculturation, familism, and
alcohol use among Latino adolescent males: Longitudinal
relations. J Community Psychol 2000;28:443–58.
11. Hingson R, Heeren T, Levenson S, et al. Age of drinking
onset, driving after drinking, and involvement in alcohol
related motor-vehicle crashes. Accid Anal Prev 2002;34:85–
92.
30. Lambert NM. Adolescent outcomes for hyperactive children:
Perspectives on general and specific patterns of childhood
risk for adolescent educational, social, and mental health
problems. Am Psychol 1988;43:786 –99.
12. Cohen DA, Richardson J, LaBree L. Parenting behaviors and
the onset of smoking and alcohol use: A longitudinal study.
Pediatrics 1994;94:368 –75.
31. Peterson PL, Hawkins JD, Abbott RD, Catalano RF. Disentangling the effects of parental drinking, family management,
and parental alcohol norms on current drinking by black and
white adolescents. J Res Adolesc 1994;4:203–27.
13. Killen JD, Hayward C, Wilson DM, et al. Predicting onset of
drinking in a community sample of adolescents: The role of
expectancy and temperament. Addict Behav 1996;21:473–80.
32. Willett JB, Singer JD. How long did it take? Using survival
analysis in educational and psychological research. In: Col-
December 2004
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
lins LM, Horn JL (eds). Best Methods for the Analysis of
Change. Washington, DC: American Psychological Association, 1991:310 –23.
Jessor R, Jessor SL. Problem Behavior and Psychosocial
Development: A Longitudinal Study of Youth. New York,
NY: Academic Press, 1977.
Huba GJ, Bentler PM. A developmental theory of drug use:
Derivation and assessment of a causal modeling approach.
In: Baltes BP, Brim OG Jr (eds). Life-span Development and
Behavior, Volume 4. New York, NY: Academic Press, 1982:
147–203.
Catalano RF, Kosterman R, Hawkins JD, et al. Modeling the
etiology of adolescent substance use: A test of the Social
Development Model. J Drug Issues 1996;26:429 –55.
Brook JS, Brook DW, Gordon AS, et al. The psychosocial
etiology of adolescent drug use: A family interactional approach. Genet Soc Gen Psychol Monogr 1990;116:111–267.
Akers RL, Krohn MD, Lanza-Kaduce L, Radosevich M. Social
learning and deviant behavior: A specific test of a general
theory. Am Sociol Rev 1979;44:636 –55.
Patterson GR, DeBaryshe BD, Ramsey E. A developmental
perspective on antisocial behavior. Am Psychol 1989;44:329 –
35.
Oetting ER, Donnermeyer JF. Primary socialization theory:
The etiology of drug use and deviance. I. Subst Use Misuse
1998;33:995–1026.
Kaplan HB. Testing a general theory of drug abuse and other
deviant adaptations. J Drug Issues 1985;15:477–92.
Flewelling RL, Bauman KE. Family structure as a predictor of
initial substance use and sexual intercourse in early adolescence. J Marriage Fam 1990;52:171–81.
Hawkins JD, Graham JW, Maguin E, et al. Exploring the
effects of age of alcohol use initiation and psychosocial risk
factors on subsequent alcohol misuse. J Stud Alcohol 1997;
58:280 –90.
Kaplow JB, Curran PJ, Angold A, Costello EJ. The prospective relation between dimensions of anxiety and the initiation
of adolescent alcohol use. J Clin Child Psychol 2001;30:316 –
26.
Webb JA, Baer PE, McLaughlin RJ, et al. Risk factors and their
relation to initiation of alcohol use among early adolescents.
J Am Acad Child Adolesc Psychiatry 1991;30:563–8.
Kosterman R, Hawkins JD, Guo J, et al. The dynamics of
alcohol and marijuana initiation: Patterns and predictors of
first use in adolescence. Am J Public Health 2000;90:360 –6.
Brook JS, Whiteman M, Gordon AS, et al. Onset of adolescent
drinking: A longitudinal study of intrapersonal and interpersonal antecedents. Adv Alcohol Subst Abuse 1985;5:91–110.
Ellickson PL, Hays RD. Antecedents of drinking among
young adolescents with different alcohol use histories. J Stud
Alcohol 1991;52:398 –408.
Krahn D, Piper D, King M, et al. Dieting in sixth grade
predicts alcohol use in ninth grade. J Subst Abuse 1996;8:293–
301.
Andrews JA, Hops H, Ary D, et al. Parental influences on
early adolescent substance use: Specific and nonspecific
effects. J Early Adolesc 1993;13:285–310.
Jessor R, Jessor SL. Adolescent development and the onset of
drinking: A longitudinal study. J Stud Alcohol 1975;36:27–51.
Pedersen W, Skrondal A. Alcohol consumption debut: Predictors and consequences. J Stud Alcohol 1998;59:32–42.
Stice E, Myers MG, Brown SA. A longitudinal grouping
analysis of adolescent substance use escalation and deescalation. Psychol Addict Behav 1998;12:14 –27.
ADOLESCENT ALCOHOL INITIATION RISK FACTORS
529.e17
53. Baumrind D. Familial antecedents of adolescent drug use: A
developmental perspective. In: Jones CL, Battjes RJ (eds).
Etiology of Drug Abuse: Implications for Prevention. (NIDA
Research Monograph 56, DHHS Publication No. (ADM)
90-1335). Rockville, MD: National Institute on Drug Abuse,
1985:13–44.
54. Graham JW, Marks G, Hansen WB. Social influence processes
affecting adolescent substance use. J Appl Psychol 1991;76:
291–8.
55. Marks G, Graham JW, Hansen WB. Social projection and
social conformity in adolescent alcohol use. Pers Soc Psychol
Bull 1992;18:96 –101.
56. Fisher LA, Bauman KE. Influence and selection in the friendadolescent relationship: Findings from studies of adolescent
smoking and drinking. J Appl Soc Psychol 1988;18:289 –314.
57. Urberg KA, Degirmencioglu SM, Pilgram C. Close friend and
group influence on adolescent cigarette smoking and alcohol
use. Dev Psychol 1997;33:834 –44.
58. Weber MD, Graham JW, Hansen WB, et al. Evidence for two
paths of alcohol onset in adolescents. Addict Behav 1989;14:
399 –408.
59. Jessor R, Collins MI, Jessor SL. On becoming a drinker:
Social-psychological aspects of an adolescent transition. Ann
NY Acad Sci 1972;197:199 –213.
60. Zimmerman MA, Schmeelk-Cone KH. A longitudinal analysis of adolescent substance use and school motivation among
African American youth. J Res Adolesc 2003;13:185–210.
61. McGue M, Iacono WG, Legrand LN, et al. Origins and
consequences of age at first drink. I. Associations with
substance-use disorders, disinhibitory behavior and psychopathology, and P3 amplitude. Alcohol Clin Exp Res 2001;25:
1156 –65.
62. Fleming JP, Kellam SG, Brown CH. Early predictors of age at
first use of alcohol, marijuana, and cigarettes. Drug Alcohol
Depend 1982;9:285–303.
63. Sher KJ. Children of Alcoholics: A Critical Appraisal of
Theory and Research. Chicago, IL: The University of Chicago
Press, 1991.
64. Jacob T, Windle M, Seilhamer RA, Bost J. Adult children of
alcoholics: Drinking, psychiatric, and psychosocial status.
Psychol Addict Behav 1999;13:3–21.
65. Martin CS, Lynch KG, Pollock NK, Clark DB. Gender differences and similarities in the personality correlates of adolescent alcohol problems. Psychol Addict Behav 2000;14:121–33.
66. Schuckit M. Biological, psychological and environmental
predictors of alcoholism risk: A longitudinal study. J Stud
Alcohol 1998;59:485–94.
67. Slutske WS, Heath AC, Madden PAF, et al. Personality and
the genetic risk for alcohol dependence. J Abnorm Psychol
2002;111:124 –33.
68. Aas HN, Leigh BC, Anderssen N, Jakobsen R. Two-year
longitudinal study of alcohol expectancies and drinking
among Norwegian adolescents. Addiction 1998;93:373–84.
69. Bauman KE, Fisher LA, Bryan ES, Chenoweth RL. Relationship between subjective expected utility and behavior: A
longitudinal study of adolescent drinking behavior. J Stud
Alcohol 1985;46:32–8.
70. Kellam SG, Brown CH, Fleming JP. Social adaptation to first
grade and teenage drug, alcohol and cigarette use. J Sch
Health 1982;52:301–6.
71. Robinson TN, Chen HL, Killen JD. Television and music
video exposure and risk of adolescent alcohol use. Pediatrics
1998;102:E54. Available at: http://www.pediatrics.org/cgi/
content/full/102/5/e54. Accessed December 20, 1998.
529.e18
DONOVAN
72. Ensminger ME, Brown CH, Kellam SG. Sex differences in
antecedents of substance use among adolescents. J Soc Issues
1982;38:25–42.
73. Newcomb MD, Bentler PM. Substance use and abuse among
children and teenagers. Am Psychol 1989;44:242–8.
74. Dusenbury L, Epstein JA, Botvin GJ, Diaz T. Social influence
predictors of alcohol use among New York Latino youth.
Addict Behav 1994;4:363–72.
75. Epstein JA, Botvin GJ, Baker E, Diaz T. Impact of social
influences and problem behavior on alcohol use among
inner-city Hispanic and Black adolescents. J Stud Alcohol
1999;60:595–604.
76. Epstein JA, Williams C, Botvin GJ. How universal are social
influences to drink and problem behaviors for alcohol use? A
test comparing urban African-American and CaribbeanAmerican adolescents. Addict Behav 2002;27:75–86.
77. Harachi TW, Catalano RF, Kim S, Choi Y. Etiology and
prevention of substance use among Asian American youth.
Prev Sci 2001;2:57–65.
78. Makini GK Jr, Hishinuma ES, Kim SP, et al. Risk and
protective factors related to native Hawaiian adolescent
alcohol use. Alcohol Alcohol 2001;36:235–42.
79. Nakashima J, Wong MM. Characteristics of alcohol consumption, correlates of alcohol misuse among Korean American adolescents. J Drug Educ 2000;30:343–59.
80. Parker KD, Calhoun TC, Weaver G. Variables associated with
adolescent alcohol use: A multiethnic comparison. J Soc
Psychol 2000;140:51–62.
81. Vega WA, Zimmerman RS, Warheit GJ, et al. Risk factors for
early adolescent drug use in four ethnic and racial groups.
Am J Public Health 1993;83:185–9.
82. Barnes GM, Farrell MP, Banerjee S. Family influences on
alcohol abuse and other problem behaviors among Black and
white adolescents in a general population sample. J Res
Adolesc 1994;4:183–201.
83. Griffin KW, Scheier LM, Botvin GJ, Diaz T. Ethnic and gender
differences in psychosocial risk, protection, and adolescent
alcohol use. Prev Sci 2000;1:199 –212.
84. Catalano RF, Morrison DM, Wells EA, et al. Ethnic differences in family factors related to early drug initiation. J Stud
Alcohol 1992;53:208 –17.
85. Gillmore MR, Catalano RF, Morrison DM, et al. Racial
differences in acceptability and availability of drugs and
early initiation of substance use. Am J Drug Alcohol Abuse
1990;16:185–206.
86. Adalbjarnardottir S, Rafnsson FD. Adolescent antisocial behavior and substance use: Longitudinal analyses. Addict
Behav 2002;27:227–40.
87. Bradizza CM, Reifman A, Barnes GM. Social and coping
reasons for drinking: Predicting alcohol misuse in adolescents. J Stud Alcohol 1999;60:491–9.
88. Chen L, Anthony JC, Crum RM. Perceived cognitive competence, depressive symptoms and the incidence of alcoholrelated problems in urban school children. J Child Adolesc
Subst Abuse 1999;8(4):37–53.
89. Christiansen BA, Smith GT, Roehling PV, Goldman MS.
Using alcohol expectancies to predict adolescent drinking
behavior after one year. J Consult Clin Psychol 1989;57:93–9.
JOURNAL OF ADOLESCENT HEALTH Vol. 35, No. 6
90. Costa FM, Jessor R, Turbin MS. Transition into adolescent
problem drinking: The role of psychosocial risk and protective factors. J Stud Alcohol 1999;60:480 –90.
91. Jessor R, Jessor SL. Problem drinking in youth: Personality,
social, and behavioral antecedents and correlates. In: Chafetz
ME (ed). Proceedings of the Second Annual Alcoholism
Conference of the National Institute on Alcohol Abuse and
Alcoholism: Psychological and Social Factors in Drinking
and Treatment and Treatment Evaluation. DHEW Publication No. (NIH) 74-676. Rockville, MD: National Institute on
Alcohol Abuse and Alcoholism, 1973:3–23.
92. Reifman A, Barnes GM, Dintcheff BA, et al. Parental and peer
influences on the onset of heavier drinking among adolescents. J Stud Alcohol 1998;59:311–7.
93. Stattin H, Gustafson SB, Magnusson D. Peer influences on
adolescent drinking: A social transition perspective. J Early
Adolesc 1989;9:227–46.
94. Rutter M. Psychosocial resilience and protective mechanisms.
Am J Orthopsychiatry 1987;57:316 –31.
95. Gore S, Eckenrode J. Context and process in research on risk
and resilience. In: Haggerty RJ, Sherrod LR, Garmezy N,
Rutter M (eds). Stress, Risk, and Resilience in Children and
Adolescents: Processes, Mechanism, and Interventions. New
York, NY: Cambridge University Press, 1996:19 –63.
96. Luthar SS, Zigler E. Vulnerability and competence: A review
of research on resilience in childhood. Am J Orthopsychiatry
1991;61:6 –22.
97. Seifer R, Sameroff AJ, Baldwin CP, Baldwin A. Child and
family factors that ameliorate risk between 4 and 13 years of
age. J Am Acad Child Adolesc Psychiatry 1992;31:893–903.
98. Werner EE. Resilience in development. Psychol Sci 1995;4(3):
81–5.
99. Collins LM, Flaherty BP, Hyatt SL, Schafer JL. WinLTA
User’s Guide, Version 2.0. State College, PA: The Methodology Center, Pennsylvania State University, 1999.
100. Raudenbush S, Bryk A, Cheong YF, Congdon R. HLM 5:
Hierarchical Linear and Nonlinear Modeling, 2nd edition.
Chicago, IL: Scientific Software International, 2001.
101. Arbuckle JL. Amos Users’ Guide, Version 4.0. Chicago, IL:
SPSS, Inc., 1999.
102. Bentler PM. EQS Structural Equations Program Manual.
Encino, CA: Multivariate Software, Inc., 1995.
103. Jöreskog KG, Sörbom D. LISREL 8: Structural Equation
Modeling with the SIMPLIS Command Language. Hillsdale,
NJ: Erlbaum, 1993.
104. Muthen LK, Muthen BO. Mplus, Statistical Analysis with
Latent Variables: User’s Guide. Los Angeles, CA: Author,
2001.
105. Kraemer HC, Stice E, Kazdin A, et al. How do risk factors
work together? Mediators, moderators, and independent,
overlapping, and proxy risk factors. Am J Psychiatry 2001;
158:848 –56.
106. MacKinnon DP, Krull JL, Lockwood CM. Equivalence of the
mediation, confounding and suppression effect. Prev Sci
2000;1:173–81.