Illicit Substance Use among Adolescents: A Matrix of ...

2 downloads 110 Views 658KB Size Report
in the field, there is a long list of causes of ISU. However ... tention to areas where the results are an inconsistent mix of significant and non- ..... Milwaukee. 78-79.
Substance Use & Misuse, 33( 13), 2561-2604, 1998

Illicit Substance Use among Adolescents: A Matrix of Prospective Predictors John Petraitis,ll* Brian R. Flay,2Todd Q. Miller,3 Edward J. Torpy,2 and Brenda Greiner2 ‘Department of Psychology, University of Alaska Anchorage, Anchorage, Alaska, USA *Prevention Research Center, University of Illinois at Chicago, Chicago, Illinois, USA 3Prevention Medicine, University of Texas at Galveston, Galveston, Texas, USA

ABSTRACT This paper reviews findings from 58 prospective studies of illicit substance use (ISU) among adolescents. It arranges 384 findings according to three types of influence (viz., social, attitudinal, and intrapersonal) and four levels of influence (viz., ultimate, distal, proximal, and immediate). The bulk of evidence reconfirms the importance of several predictors of ISU (e.g., intentions and prior substance-related behavior, friendship patterns and peer behaviors, absence of supportive parents, psychological temperament), reveals that a few variables thought to be well-established predictors may not be (e.g., parental behaviors, parental permissiveness, depression, low self-esteem), and uncovers several variables where findings were either sparse or inconsistent (e.g., the role of public policies concerning ISU, mass media depictions of ISU, certain parenting styles, affective states, perceptions of parental disapproval for ISU, and substance-specific refusal skills). Directions for future research are discussed. *To whom correspondence should be addressed at Department of Psychology, University of Alaska Anchorage, 3211 Providence Drive, Anchorage, AK 99508. E-mail: [email protected] 2561 Copyright 0 1998 by Marcel Dekker, Inc.

www.dekker.com

2562

PETRAITIS ET AL.

Key words. Illicit substances; Longitudinal; Predictors

Why is it that some adolescents never use illicit substances (e.g., marijuana and cocaine, rather than cigarettes and alcohol), others experiment with illicit substance use (ISU), and still others regularly use and misuse these substances? This question has been the focus of presidential debate, hundreds of empirical studies, dozens of theories (see Lettieri et al., 1980), and numerous reviews (e.g., Hawkins et al., 1992; Moncher et al., 1991). According to the dominant theories in the field, there is a long list of causes of ISU. However, Petraitis et al. (1995) suggested that all of the potential causes of ISU fall into 10 categories. These categories are depicted in Table 1 and were based on a 3 x 3 + 1 matrix in which Petraitis et al. crossed three different types of influence with four different levels of influence. As shown in Table I, the three types of influence were (a) social influences on ISU, consisting of the “characteristics and behaviors of the people who make up adolescents’ most intimate support systems” (p. 79), (b) attitudinal influences, consisting of “adolescents’ own substance-specific attitudes and factors that affect those attitudes” (p. 80), and (c) intrapersonal influences, consisting of the basic personality traits (e.g., extroversion), affective states (e.g., low self-esteem), and behavioral skills (e.g., refusal skills) that lead adolescents to use illicit substances. The four levels of influence were called immediate, proximal, distal, and ultimate influences, and represented a progression from those variables that probably have direct effects on ISU (e.g., substance-related intentions) to other variables whose effects on ISU are probably mediated by many other variables (e.g., marital disruption among parents, to external locus of control among adolescents). Of course, knowing whether the variables in Table 1 actually influence ISU requires empirical evidence. Moreover, given the unethical nature of experimental studies designed to induce ISU, the strongest empirical evidence comes from nonexperimental prospective (or longitudinal) studies that can demonstrate whether a variable is, at least, an antecedent (or predictor) of ISU. Although prospective studies have their own special set of problems (see Flay and Petraitis, 199 l), they are generally well-suited for testing theories regarding the predictors of ISU. It is not surprising, therefore, that the current decade has brought a rapid expansion in the number of prospective studies of ISU. In fact, at least 23 such studies have been published since 1990 alone. In this paper, 58 prospective studies of ISU by adolescents and young adults are brought together on one stage. Although other notable reviews of substance use exist (e.g., Hawkins et al., 1992), the current review is different in several ways. First, it focuses exclusively on longitudinal or prospective predictors and

L

ILLICIT SUBSTANCE USE AMONG ADOLESCENTS

2563

does not consider cross-sectional correlates of ISU. Second, it does not disregard nonsignificant findings. Although past reviews summarized hundreds of findings, they often omitted nonsignificant findings and, as a consequence, might have led to some overstated conclusions. The current review, by contrast, calls some attention to areas where the results are an inconsistent mix of significant and nonsignificant findings. Finally, not only does this review organize predictors into different types of influence (as prior reviews invariably did in one form or another), it also uses Table 1 to organize predictors into different levels of influence. That is, it first summarizes prospective predictors from three levels of social influences (viz., ultimate, distal, and proximal influences), then summarizes research on three levels of attitudinal and intrapersonal influences, and eventually describes research on the immediate influences of ISU (viz., substance-specific decisions, intentions, and related behavior). Certainly, some readers might disagree with Petraitis et al. (1995) about the location of some variables in Table 1, contending, for example, that peer drug use is an immediate influence rather than a distal one. However, for the purposes of our paper, the location of specific variables in Table 1 only affected the order in which we discussed findings and had no effect on the conclusions we drew. The relationship between peer drug use and ISU, for example, is the same whether peer drug use is considered an immediate influence or a distal influence. Nonetheless, Table 1 provided a useful way of ordering our reviews of dozens of variables and hundreds of findings, showing what variables are and are not related to ISU, and revealing areas where more research is needed. METHOD Selection of Prospective Studies We used several procedures recommended by meta-analysts (e.g., Rosenthal, 1984) to locate English-language publications and manuscripts that described prospective predictors of ISU among adolescents. The search relied on (a) computerized data bases (e.g., Medicus Zndicus and PsychLit), (b) the inspection of journals known to publish longitudinal studies of drug use, (c) careful examination of the citations and references of known studies and reviews of research on drug use, (d) personal knowledge of unpublished prospective studies, and (e) Social Science Citation Index to find recent studies that had cited known studies. The search located 58 prospective studies where predictor variables were measured at least 5 months before the dependent variables, where dependent variables specifically targeted the use of illicit substances (e.g., marijuana or cocaine) rather than licit substances (i.e., cigarettes and alcohol), and where dependent variables were assessed while participants were still adolescents or young adults (defined as younger than 27 years old). These criteria eliminated (a) all

Table 1.

A Matrix of Types and Levels of Influence on Illicit Substance Use Types of influence Level of influence

Social

Ultimate

Definition: Characteristics of the people who make

Definition: Aspects of adolescents’ immediate

Definition: Personality traits and intrapersonal

up adolescents’ most intimate social support system. These characteristics are not specific to ISU and are beyond the personal control of adolescents but nonetheless put adolescents “at risk” for succumbing to social pressure to use substances Construers: Infrequent opportunities for rewards from family members; lack of parental warmth, support, or supervision; negative evaluations from parents; home strain; parental divorce or separation; unconventional values of parents; unconventional values among peers

surroundings, neighborhoods, social institutions, and culture that, although beyond the personal control of adolescents, put them at risk for developing positive attitudes toward ISU Constructs: Local crime and employment rates; inadequate schools; poor career and academic options; infrequent opportunities for rewards at school; negative evaluations from teachers; media depictions of ISU; availability of substances; weak public policies on ISU

characteristics that, although beyond the easy control of adolescents, might promote some internal motivation to use substances or make them susceptible to the .physiological effects of ISU Constructs: Genetic susceptibility to addiction; lack of impulse control; external locus of control; aggressiveness; extroversion; sociability; risk-taking; sensation-seeking; neuroticism or emotional instability; intelligence

Attitudinal

Intrapersonal

Distal

Definition: Emotional attachments of adolescents

and the substance-specific attitudes and behaviors of influential role models who encourage ISU Consrructs: Weak attachment to and weak desire to please family members; strong attachment to and strong desire to please peers; greater influence by peers than parents; substancespecific attitudes and behaviors of role models Proximal

Definition: Beliefs about the normative nature of

ISU and pressures to use substances Consrrucrs: Prevalence estimates; motivation to comply with other users; beliefs that important others (i.e., friends, parents, and other role models) encourage ISU

Definilion: Personal values and behaviors of

Defnifion: Affective states and general behavioral adolescents that contribute to their attitudes skills of adolescents that promote some internal toward ISU motivation for ISU and that undermine their Con.rtracrs: Weak commitment to conventional refusal skills values, school, and religion; social alienation Constructs: Low self-esteem; temporary anxiety, and criticism; weak desire for success and stress, or depressed mood, poor coping skills; achievement; hedonic values and short-term inadequate social skill; weak academic skills gratification; rebelliousness; desire for independence from parents; tolerance of deviance Definition: Beliefs and evaluations about the costs Definition: Beliefs about one’s ability to use or and benefits of ISU avoid substances Con.rtructs: Expected costs and benefits of ISU, Conrtnrcts: Refusal skills; determination to use evaluation of costs and benefits of ISU; attitudes substances; use self-efftcacy; refusal skilltoward ISU by others; attitudes toward ISU by efficacy self

Decision/intentions; trial behavior and related behaviors

2566

PETFCAITIS ET AL.

studies where predictor variables were measured retrospectively at the time ISU was measured (e.g., Robins and Pmybeck, 1985), (b) all prospective studies where dependent variables were based on a combination of licit and illicit substance use (e.g., Friedman et al., 1987), and (c) all prospective studies where the analyses addressed the consequences of ISU (e.g., Hansel1 and White, 1991) rather than the causes. In Table 2 we have organized these 58 publications and manuscripts into 40 different projects, combining papers that share a common set of participants. Two of these papers were conducted by Smith and Fogg ( 1978, 1979), four came from the Woodlawn Project conducted in Chicago (Ensminger et al., 1982; Fleming et al., 1982; Kellam et al., 1982, 1983), three came from the New York Longitudinal Survey (Lerner and Vicary, 1984; Vicary and Lerner, 1983, 1986), five came from a project at Rutgers University (Johnson, 1988; Johnson and Pandina, 1991; Pandina and Johnson, 1990; White, 1992; White et al., 1987), and nine came from a project based at the University of California Los Angeles (Huba et al., 1980, 1981; Maddahian et al., 1988; Newcomb and Bentler, 1986, 1987; Stein et al., 1986, 1987a, 1987b; Weng and Newcomb, 1989). Table 2 groups these publications into the Smith-Fogg, Chicago, New York, Rutgers, and UCLA projects, respectively. At the same time, however, in Table 2 we separate any publications that, although having common participants, had dependent variables that were measured at different times. Among these were the publications from the UCLA project. In particular, three UCLA publications (which we called the UCLA- 15 studies) reported predictors of marijuana use when participants were 15 years old or younger, four reported predictors when the same participants were 19 years old or younger (UCLA- 19), and five reported predictors when the same participants were 22 years old or younger (UCLA-22). Because the patterns of results were not consistent across waves, we organized the UCLA studies into three separate projects. Also, because Flay et al. (1989) have not yet published their findings, they are included as an Appendix. Characteristics of Prospective Projects Beyond just grouping prospective studies together, Table 2 also summarizes characteristics of each project’s sample, dependent variable, and most basic analysis method. As can be seen, these prospective studies varied widely in age ranges, geographic location, sample sizes, usage rates, and analysis methods. They also varied widely in their measures of ISU, covering distinctly different time frames (ranging from ISU during the last month to ISU during any point in an adolescent’s lifetime), substantially different levels of ISU (ranging from experimental use, through regular use, to misuse and addiction), and entirely different

ILLICIT SUBSTANCE USE AMONG ADOLESCENTS

2561

substances (ranging from the use of a single illicit substance to measures of polysubstance use). In fact, multiple measures of ISU in one project were common, and several projects (e.g., Simcha-Fagan et al., 1986) reported separate analyses for individual substances. Although these longitudinal studies provide perhaps the best evidence about the causes of ISU, they are not without limitations and potential threats to statistical inference validity, construct validity, and external validity. Concerning potential threats to statistical inference validity, Table 3 shows that projects regularly reported the statistical relationship between ISU and dozens of predictors without adjusting their alpha-level or cross-validating their findings, leaving open the possibility that some of the significant relationships they reported were simply Type I errors. Another concern involves the accuracy of the measures of ISU. As Table 3 demonstrates, (a) all but one study relied exclusively on self-reported measures of ISU, (b) most studies collected data from adolescents while they were in classrooms, and (c) only 25% of the projects reported using any extraordinary procedures to reduce underreporting of ISU (e.g., doing more than just assuring participants of confidentiality), leaving open the possibility that underreporting of ISU suppressed the statistical relationships between predictors and ISU. A last concern involves the generalizability of the results. As Table 3 demonstrates, however, this happens to be a strength of many studies because they used some method of probability sampling (viz., random, stratetied, or cluster sampling). Moreover, many of the studies that did not use probability sampling still attempted to collected data from all of the students present in a school or school district, thereby decreasing the likelihood that the results generalize only to a subset of adolescents who happened to be convenient sources of data. Strategy for This Review While reviewing the findings from these studies, we had two general goals: (a) to review, whenever possible, those findings that were most directly comparable, rather than focus on findings where differences in operational definitions and analysis strategies made comparisons difficult, and (b) to provide a review that was broad in scope and covered the full range of predictors from longitudinal studies of ISU, rather than focusing only on a small subset of such predictors. With these goals in mind, we adopted a four-part strategy for summarizing past findings. First, we limited our review to potential precursors of ISU that were investigated in two or more projects and thereby had the opportunity to be crossvalidated independently. As a result, our review did not include all findings from prospective studies, but instead targeted only findings from variables that were included in more than one study. Second, we limited our review, whenever possible, to bivariate rather than multivariate predictions of ISU. Although multivari-

. Table 2.

Characteristics of Prospective Studies of Adolescent Illicit Substance Usei T Project 1) Bailey et al. (1992) 2) Bailey and Hubbard (1990) 3) Barkley et al. (1990) 4) Baumrind (1985) 5) Block et al. (1988) 6) Brook et al. (1990) 7) Chicago: (a) Ensminger et al. (1982) (b) Fleming et al. (1982) (c) Kellam et al. (1982, 1983) 8) Dembo et al. (1990) 9) Ellickson et al. (1992) IO) Elliott et al. (1985) 11) Farrell (1993) 12) Fergusson et al. ( 1993) 13) Flay et al. (1989) 14) Hammer (1992) 15) Hansel1 and Mechanic ( 1990) 16) Jessor et al. (1991) 17) lessor and Jessor (1977) 18) Johnson et al. (1990) 19) Johnston et al. (1978) 20) Kandel et al. (1978) 21) Kandel et al. (1986) 22) Kaplan et al. (1986)

Age at time 1.

Age of ISUb

Year of ISU

Location of sample

12-14 1 l-13 5-7 4-5 3 S-10 6 6 6 15 12-13 11-17 12 8 12 17-20 12-14 15-17 15-17 I I-12 15 l&l7 15-16 12

14-16 12-14 13-15 14 14 13-18 16-17 1617 16-17 16 15-16 I&20 13-14 15 14 2 1-24 14-16 25-27 1618 14-15 19 15-18 24-25 23-25

8990 87-88 87-88 78-79 78-79 83 75-76 7576 75-76 87-88 Late 80s 79 9&91 ? 88 87-88 86 81 72 87 14 72 80 81-83

SE USA SE USA Milwaukee Metro Oakland Metro Oakland Upstate NY Chicago Chicago Chicago Tampa Bay CA and OR USA SE USA New Zealand S. CA Norway NJ Boulder, CO Boulder. CO Kansas City USA NY State NY State Houston

Final NE 2428 1,659 189 136 104 429 496 702 632 2294 4,145 1,655 1,261 875 22,290 126 2606 384 308 1,105 2 1,248 2460 1,004 I.229

Measure of ISU” Continued marijuana use Lifetime marijuana use Marijuana use Lifetime marijuana use Lifetime marijuana use Lifetime marijuana use Heavy marijuana use Age of tirst marijuana use Lifetime of marijuana use Yearly marijuana use Lifetime marijuana use Yearly marijuana use Monthly marijuana use Lifetime marijuana use Onset of marijuana use Yearly marijuana use Lifetime marijuana use Monthly marijuana use Onset of marijuana use Monthly marijuana use Stages of lSUh Onset of marijuana use Yearly marijuana use Daily marijuana use

Base rate End rate (%))’ (%)f Analysis methods 0

14 0 0 0 ? 0 0 0 ? 14 ? 5 ? 0 ? ? ? 0 2 ? 0 16 24

38 28 13 64 51 55 26 58 61 64 44 ? 9 9 20 ? ? ? I5 16 36 17 43 52

Pearson r Logistic regression Chi-square Pearson r Pearson r Pearson r Chi-square Survival analysis Log-linear Pearson r Scalogram analysis Path analysis Pearson r Chi-square Pearson r Multiple regression Multiple regression Pearson r t-tests Logistic regression One-way F Pearson r Pearson r Pearson r

Table 2. Continued Age at

. Project 35) UCLA-20: (a) Newcomb and Bender (I 986) (b) Maddahian et al. (1988) (c) Stein et al. (1987a) 36) UCLA-24: (a) Newcomb and Bender (1986) (b) Newcomb and Bender (1987) (c) Stein et al. (1986) (d) Stein et al. (1987a) (e) Stein et al. (1987b) 37) VanRammen and Locber (1994) 38) Wills et al. (1989) 39) Windel(l989, 1990) 40) Winefield et al. (1993)

time 1’

Age of ISUb

Year of ISU

12-14 12-14 12-14

17-20 17-20 17-20

17-20 17-20 17-20 i7-20 17-20 12-13 12-14 14-15 15-17

Base rate End rate (%) (o/o)’ Analysis methods

Location of sample

Final NC

80 80 80

Metro LA Metro LA Metro LA

654 847 654

6 month cannabis use 6 month cannabis use 6 month cannabis use

? ? ?

49 ? 49

EQS

2 l-24

84

Metro LA

654

6 month cannabis use

?

42

EQS

I+23 2 1-24 2 1-24 21-24 15-16 l&16 18-19 23-25

84 84 84 84 8%91 ? 84 88

Metro LA Metro LA Metro LA Metro LA Pittsburgh Suburban NYC USA Australia

6 months cannabis use 6 month cannabis use 6 month cannabis use 6 month cannabis use onset of ISU Lifetime marijuana use Lifetime marijuana use Regular ISU

? ? ? ? 0 ? ? ?

43 42 42 42 15 ? 23 7

Pearson r LISREL Regression

479 654 654 654 506 900 2,370 2472

Measure of ISUd

Partial r Regression

EQS Logistic regression Multiple regression Pearson r ?

aAge at Time 1 represents participants’ ages at the time predictor variables were measured.

bAge of ISU represents participants ages at the time ISU was measured. cFinal N represents the number of subjects included in statistical analyses. dMeasure of KU represents the operational definition of ISU. eBase rates represent the percent of participants included in the analyses who acknowledge ISU at the time predictors were measured. ‘End rates represent the percent of participants included in the analyses who acknowledge ISU at the time ISU was measured. sin many studies, several analyses were reported We note here the level of analysis that we focused on, usually the lowest order of analysis. hAlthough not defined identically by Johnston et al. (1978) and McBride et al. (1991), stages of ISU ranged from no drug use, through marijuana use, up to various degrees of narcotic use. ‘Unlike Johnson (1988) and White et al. (1987) who combined data from three cohorts, Johnson and Pandina (1991) analyzed data separately for three cohorts, reporting age 12, 15, and 18 predictors for age 15, 18, and 2 1 marijuana use, respectively. jA question mark denotes information that was not reported in published sources.

Yearly marijuana use Onset of marijuana use Stages of ISIJ Marijuana use Marijuana use Marijuana use

2 0 ? 0 0 0

Moderate marijuana use Marijuana use Marijuana use Problematic drug use Problematic marij. use Heavy drug use Monthly marijuana use Experimental marijuana use Onset of marijuana use

? ? ? ? ? ? ? 0 0

55 53

Pearson r Logistic regression Pearson r Multiple regression Multiple regression Discriminant functions Logistic regression Pearson r Multiple regression Chi-square Multiple regression ANCOVA Pearson r r-tests Biserial r

651

Lifetime marijuana use

?

20

Biserial r

Metro Boston Israel Metro LA Metro LA

2,249 >I,200 2584 1,177

Lifetime marijuana use Lifetime cannabis use 6 month marijuana use 6 month marijuana use

8 ? ? ?

33 ? 24

t-tests One-way F Bivariate r Bivariate r

Metro LA

1,177

6 month cannabis use

?

26

Bivariate r

I2 23) fiDk3D et al. (1984) 14-19 24) Le;y and Pierce (1990) 13-17 25) McBride et al. (199 1) 26) New York: (a) Lemer and Vicary (1984) 1 (b) Vicary and Lemer (I 983) 3 (c) Vicary and Lemer (1986) 3

13 15-20 17-21 220 220 220

12 86 85-89 t75 275 215

Houston Syndey, Aust. Austin, TX Metro NYC Metro NYC Metro NYC

22,709 703 110 133 133 133

16-18 27) Pedersen (1991) 18 28) Rutgers: (a) Johnson (1988) (b) Johnson and Pandins (1991)’ 12-18 (c) Pandina and Johnson (1990) 12-18 15 (d) White (1992) 12-18 (e) White et al. (1987) 17 29) Schulenberg et al. (1994) Sand11 30) Shedler and Block (1990) 31) Simcha-Fagan et al. (1986) 618 32) Smith and Fogg: (a) Smith and II-12 Fogg (1978) (b) Smith and Fogg (1979) 12-13 14-17 33) Teichman et al. (I 989) 34) UCLA-15: (a) Huba et al. (1980) 12-14 (b) Huba et al. (1981) 12-14 (c) Weng and Newcomb 12-14 (1989)

I MO 21 15-21 18-24 21 15-21 21-22 18 I &22

a9 82-83 82-84 85-87 85-87 82-83 81-85 82-83 71-72

Oslo, Norway NJ NJ NJ NJ NJ USA Metro Oakland NY city

553 121 1,308 1,380 1411 882 3,399 65 193

1617

73

Metro Boston

14-17 15-18 13-15 13-15

71 83 77 77

13-15

77

12 6 41 47 41

?

(continued)

,

ILLICIT SUBSTANCE USE AMONG ADOLESCENTS

2571

ate analyses in different studies occasionally shared a common predictor variable, multivariate analyses typically included widely dissimilar covariates, thereby leaving bivariate analyses as the most comparable analyses across studies. Third, in studies with multiple measures of ISU, we focused on the least extreme measure of ISU (e.g., having ever used an illicit substance, rather than having used illicit substances on a regular basis), again because such measures were the most comparable across studies. In 44 of the 58 studies, the least extreme and most comparable measure was a measure of experimental (i.e., less than regular) marijuana use. Finally, and perhaps most important, we opted for a traditional qualitative review rather than a quantitative meta-analysis. The reason for adopting a traditional review came down to a choice between the depth provided by a meta-analysis and the breadth provided by a more traditional review. Me&analyses are (a) yell-suited for summarizing the relationship between some dependent variables (e.g., ISU) and a small set of predictors, and (b) effective at providing detailed information about average effect sizes and moderators of effect sizes for a handful of predictors. Certainly, such focused investigations are worth conducting. At the same time, however, because metaanalyses are best suited to focusing on a small set of predictors, they tend to be “a mile deep but only an inch wide” and are less well-suited for summarizing relationships between an outcome variable and dozens of predictors. As shown below, past research on ISU has produced more than 50 different longitudinal predictors, making a meta-analysis of this field infeasible for any single paper. Moreover, with so many predictors and so many studies, the field still needs a systematic and comprehensive review of findings. One final methodological point is in order. Reviewers of published studies must be concerned that their conclusions might be distorted by a publication bias that screens out nonsignificant findings and favors significant findings. However, we believe that a publication bias is not especially strong in this review because longitudinal studies of ISU have a strong record of publishing both significant and nonsignificant findings. In fact, researchers in this area have typically collected data for a large set of variables and then routinely presented the findings for all variables. In fact, 75% of the findings we reviewed were from studies or projects that began with a large set of variables and ended with an equally large table that reported both the significant and nonsignificant predictors of ISU. That is, most of the findings in this review were from “kitchen sink” studies that reported on everything in the sink.

RESULTS FROM PROSPECTIVE STUDIES OF SOCIAL INFLUENCES Social influences represent characteristics in adolescents’ immediate social settings (e.g., schools, neighborhoods, families, friendship groups) that contrib-

Table 3. Selected Methodological Features of Prospective Studies of Adolescent illicit Substance Usea

Project 1) Bailey et al. (1992) 2) Bailey and Hubbard (1990) 3) Barkley et al. (1990) 4) Baumrind (1985) 5) Block et al. (1988) 6) Brook et al. (1990) 7) Chicago: (a) Ensminger et al. ( 1982) (b) Fleming et al. ( 1982) (c) Kellam et al. (1982, 1983) 8) Dembo et al. (1990) 9) Ellickson et al. (1992) 10) Elliott et al. (1985) 11) Farrell (I 993) 12) Fergusson et al. (1993) 13) Flay et al. (1989) 14) Hammer (1992) 15) Hansell and Mechanic (1990) 16) Jessor et al. (1991) 17) Jessor and Jessor ( 1977) IS) Johnson et al. (1990) 19) Johnston et al. (1978) 20) Kandel et al. (1978) 2 1) Kandel et al. ( 1986) 22) Kaplan et al. (1986) 23) Kaplan et al. (1984) 24) Levy and Pierce (1990)

Analyses without adjusting a 45 72 1 50 682 27 4 12 10 4 I 6 2 2 1 1 4 30 54 II I 100 ? 22 9 I

Crossvalidation

Source for measure of ISUb

Testing location where ISU data werecollected

Extraordinary efforts to reduce underreporting

Probability sampling method

No No No No No No No No No No No Yes No No No No No No No No No No No No No No

SR SR SR SR SR SR SR SR SR SR SR SR SR Parents; SR SR SR SR SR SR SR SR SR SR SR SR SR

School ? Clinic ? ? ? ? ? ? Home or custody ? ? School ? School Mail 7 Mail School School Home School School School School Home

None None None None None None None None None None Saliva sample None None Parental reports None None None None None Breath test Return via mail Self-generated ID Self-generated ID None None None

? ? Purposive ? ? Random ? ? ? Random ? Cluster Schoolwide ? Schoolwide Stratefied Convenience Stratetied Stratefied ? Stratefied Stratefied Stratetied Schoolwide Schoolwide Random

10 25) McBride et al. (1991) 1 26) NY: (a) Lemer and Vicary (1984) 4 (b) Vicary and Lemer (1983) 1 (c) Vicary and Lemer (1986) 2 27) Pedersen (1991) 6 28) Rutgers: (a) Johnson (1988) 12 (b) Johnson and Pandina ( 199 1) 1 (c) Pandina and Johnson (1990) 4 (d) White (1992) (e) White et al. (1987) 25 1 29) Schulenberg et al. (1994) 149 30) Shedler and Block (1990) 27 3 1) Simcha-Fagan et al. ( 1986) 5 32) Smith and Fogg (1978) 26 Smith and Fogg (1979) 8 33) Teichman et al. (1989) 1 34) UCLA- 15: (a) Huba et al. (1980) 1 (b) Huba et al. (1981) 4 (c) Weng and Newcomb (1989) 35) UCLA-20: (a) Newcomb and Bentler (1986) 3 4 (b) Maddahian et al. (1988) 36 (c) Stein et al. (1987a) 36) UCLA-24: (a) Newcomb and Bentier (1986) 3 (b) Newcomb and Bentler ( 1987) 1 3 (c) Stein et al. (1986) (d) Stein et al. (1987a) 48 (e) Stein et al. (1987b) 4 37) VanKammen and Loeber (1994) I 38) Wills et al. (1989) I 39) Windel(1989, 1990) 4 9 40) Winefield et al. (1993)

No No No No No No No No No No No No No No No No Yes No No No No No No No No No No No No No No

SR SR SR SR SR SR SR SR SR SR SR SR SR SR SR SR SR SR SR SR SR SR SR SR SR SR SR SR SR SR SR

BA question mark denotes information about this feature was not published in the study. bSR = self-report by adolescents or young adults.

? ? School Clinic Clinic Clinic Clinic Clinic Mail ? School School School School School School School School School Mail Mail Mail Mail Mail Home School Home Mail

None None None None None None None None None None Return via mail None None None None None None None None None None None Return via mail Return via mail Return via mail Return via mail Return via mail None Saliva samples None Return via mail

Purposive ? ? ? ? Random Random Random Random Random Stratetied ? Random Schoolwide Schoolwide ? Schoolwide Schoolwide Schoolwide Schoolwide Schoolwide Schoolwide Schoolwide Schoolwide Schoolwide Schoolwide Schoolwide Random Schoolwide Stratefied Random

2574

PETRAITIS ET AL.

ute to ISU by shaping adolescents’ perceptions of the social norms and social pressures concerning ISU. They also represent a mainstay of prospective studies of ISU. In fact, 37.5% of theprospective findings we reviewed fell into the three cells of Table 1 that make up the social influences. The bivariate relationships between these influences and ISU are summarized in Table 4. Ultimate-Level Social Influences: The Home Environment As Table 1 suggests, the ultimate level among social influences represents aspects of adolescents’ immediate social settings that, although beyond their personal control, put them at long-term risk for believing that ISU is socially normative and for succumbing to social pressure to use illicit substances. Numerous theorists (e.g., Elliott et al., 1985) have speculated about such influences, and 56 findings from prospective studies focused on ultimate-level social influences, including socioeconomic status (SES), signs of family disruption, marital styles, and signs of strained relationships at home. SES. Six studies found that marijuana use was related to the SES of an adolescent’s family (see Table 4). In particular, adolescents from higher SES families are at greater risk for marijuana use than adolescents from families with lower SES, perhaps because the former have the financial ability to obtain marijuana when it is available. However, five other studies reported no significant relationships between SES and ISU, and one study (Jessor and Jessor, 1977) found that the relationship did not hold for females. Family Disruptions and Conflicts. Other ultimate-level social influences represent different forms of social disorganization. Elliott et al. (1985) reasoned that adolescents might be at risk for ISU when social institutions are weak or breaking down. If this is true, ISU should be higher among adolescents from disrupted families where parents are absent, family members lack cohesion, adolescents are exposed to psychological and emotional problems among family members, and schools are inadequate. Several studies addressed the role of family disruption in ISU. However, as shown in Table 4, the evidence has been mixed. Contrary to predictions, and contrary to Hawkins et al.‘s (1992) conclusion, four studies concluded that there was no relationship between marital disruptions and subsequent ISU, finding that adolescents who lived with one parent or divorced parents were not more likely to use marijuana than adolescents who live with both natural parents. Moreover, two studies (Johnson and Pandina, 1991; McBride et al., 1991) found no evidence that the frequency of fighting among family members or the cohesiveness of relationships among family members affected adolescents’ subsequent ISU. By contrast, however, five studies in Table 4 did find that children and young ado-

Table 4. Social Predictors of Illicit Substance Use Organized by Level of Influence, Project, and Findings Project0 Levels and predictorsb Ultimatepredictors:Socialsettings: Parents’ SES Family disruptions and conflict (+) Family history of psychological problems (+) Parental permissiveness (+) Lack of parental support (+) Home strain (+) Distal predictors: Social bonds and role models: Family bonds (+) Importance of peers vs family (+) General peer bonds (+) Deviant peer bonds (+) Adult cigarette use (+) , Adult alcohol use (+) Adults’ attitudes re ISU (+) Adult ISU (+) Peer cigarette use (+) Peer alcohol use (+) Peer marijuana use (+) ; Peer narcotic use (+) i Peers’ attitudes toward ISU (+) Peer offers of drugs (+) Proximalpredictors:Normativeperceptions: Perceived adult approval (+) pbrceived peer approval (+)

Significant relationshipsC

Nonsignificant relationships

4, 15, 17,20,22,31 4, 13, 15,20,26c, 36d 21 4,21,26c 4,6,8, 13, 15, 17,26c, 28e, 30(-) 10, 13,21-23

5, 6, 16,21,40 1, 6, 25,28be, 3 1 6,31 6, 13, 17,20,25,28be, 31 25,28be, 3 1 2,25,28b, 3 1

2, IO, 13,20 2, 13, 16, 17,20 6, 15,20,21,23,25,31 10, 16, 17,25,28e 13 13,21,28bc, 34a 20 18,21,34a 13, 18,20,34a 20,34ac 20,34ac, 36d 23,28e, 34a 2, 20, 28e 20,34ac

6,21,28b

I,20 13, 16,20,28ae

2, 13, 17,22 18,28b 15, 18,20, 35c, 36d 2, 17,28e 20,35c, 36d 2, 18, 28e 1, 18,35c 22,35c, 36d

2, 16,28be, 36d

192

aProject numbers correspond to the numbers used in the first column of Table 2. bThe theoretically expected relationship between each predictor and ISU is noted in parentheses. cWhenever possible, findings are based on reported bivariate relationships between each predictor andISU, p < .OS. Unless otherwise denoted with (-), all significant relationships am in the positive direction.

2576

PETRAITIS ET AL.

lescents from homes with single or divorced parents were at risk for subsequent marijuana use. Interestingly, Vicary and Lemer (1996) found that children were even at risk for marijuana use as young adults if their parents had conflicts about child-rearing practices. One study (Kandel et al., 1986) also found that adolescents were at risk if there was a history of psychological problems among the family; however, this finding has not stood the test of replication in longitudinal studies. Parental Permissiveness and Support. Parenting style is another factor that is beyond the control of adolescents but might make them succumb to social pressures to use illicit substances. In fact, Brook et al. (1990) and Simon et al. (1988) hypothesized that child-rearing practices might directly affect adolescents’ initial ISU. Prospective studies focused on two parenting styles in particular: parental permissiveness or control; and parental support, encouragement, and warmth. As Table 4 shows, there is some evidence, albeit inconsistent, that adolescents are more likely to use illicit drugs if their parents treat them permissively and do not exercise control over them at an earlier age. Specifically, Baumrind (1985) found that 4-year-old children whose parents were firm and demanded household help from children were less likely to use marijuana by age 14 than children with more permissive and less demanding parents. Conversely, Kandel et al. (1986) found that ISU was more common among male adolescents (but not females) whose mothers were relatively permissive. Similarly, Vicary and Lemer (1986) found that marijuana use was more common among young adults if their parents were neither disciplinary nor restrictive with them. Nevertheless, most studies found that parental permissiveness had no direct effect on ISU, a conclusion that stands in contrast to Hawkins et al’s (1992) suggestion that permissiveness does precede alcohol and substance use. For instance, studies found that marijuana use was not directly affected by parents (a) having been strict (Jessor and Jessor, 1977), (b) having exerted influence over an adolescent’s activities and choice of friends (Flay et al., 1989; Kandel et al., 1978), (c) having allowed adolescents to participate in family decisions (Flay et al., 1989), (d) having been hostile toward adolescents or having punished their misdeeds (Brook et al., 1990; Johnson and Pandina, 1991), (e) having been intolerant of delinquency (White et al., 1987), (f) having been punitive (SimchaFagan et al., 1986), or (g) having exerted control over their children (McBride et al., 1991). By contrast, Table 4 shows a more consistent relationship between ISU and having parents who are nonsupportive or abusive. Specifically, six prospective studies (Baumrind, 1985; Brook et al., 1990; Flay et al., 1989; Hansel1 and Mechanic, 1990 [older students only]; Jessor and Jessor, 1977; Vicary and Lemer, 1986) found that youths who felt that their parents were unresponsive to their needs, were not nurturing, and discouraged their personal interests, were at least

ILLICIT SUBSTANCE USE AMONG ADOLESCENTS

2577

marginally more likely to use marijuana as adolescents than were youths who felt more supported by their parents. Similarly, White et al. (1987) found that “heavy users” of illicit substances reported less nurturing from their parents at an earlier age than less frequent users. Moreover, one study (Dembo et al., 1990) found that male adolescents who reported being physically or sexually abused in their lives were at risk of subsequent marijuana use. It should be noted, however, that (a) one study (Shedler and Block, 1990) found that adolescents who experimented with marijuana had significantly more parental support than adolescents who had abstained from marijuana throughout their lives, and (b) separate projects found no relationship between parental warmth or support and ISU. Home Strain. Finally, Elliott et al. (1985) proposed that another ultimatelevel cause of ISU is strain (i.e., the discrepancy between adolescents’ aspirations and their opportunities to achieve those aspirations). One form of strain pertains to families and the discrepancies between adolescents’ desired and obtained relationships with family members. As Table 4 shows, results are mixed and three studies (Bailey and Hubbard, 1990; McBride et al., 1991; Simcha-Fagan et al., 1986) found no relationship between ISU and strained contact with parents. Furthermore, Johnson and Pandina (1991) found that frequent fights with siblings was not related to ISU. However, four other studies found that family-related strain can affect subsequent ISU. In particular, Elliott et al. (1985) found that adolescents who wanted (but did not have) closer relationships with their families were at risk for marijuana use during later adolescence and early adulthood. Similarly, 12 year olds who felt rejected by their parents were more likely to use marijuana as 14 and 22 year olds (Kaplan et al., 1984, 1986). Finally, there is some evidence that adolescents who report troubled or emotionally distant relationships with their parents are at risk for later marijuana use (Flay et al., 1989; Kandel et al., 1986). Distal-Level Social Influences: Social Bonds and Models The second level of social influences has, by far, received the most attention in prospective studies. In fact, although Table 1 has 10 cells, nearly 20% of all prospective findings fell into the one cell represented by distal-level social influences. Unlike ultimate-level social influences that are beyond the easy control of adolescents and are thought to put adolescents at long-term risk for ISU, distallevel social influences are more intermediate causes of ISU, some (although not all) of which are within the control of adolescents. At the heart of these influences lies (a) adolescents’ emotional attachments with both conventional and deviant role models, and (b) the substance-specific behaviors and attitudes of those role models. A summary of findings concerning distal-level social influences is given in Table 4.

2578

PETRAITIS ET AL.

Family Bonds. Assuming that parents and family members generally oppose ISU, numerous theorists (e.g., Elliott et al., 1985; Jessor et al., 1991) have argued that adolescents who feel emotionally distant, alienated, or detached from their families will be more susceptible to social pressures to use illicit substances. In fact, a popular notion asserts that ISU can be decreased by strengthening adolescents’ attachments to their families. As can be seen in Table 4, however, this notion has received only inconsistent support from prospective studies. In fact, whereas four studies found that close bonds between adolescents and parents or other family members apparently deter marijuana use, three studies found no relationship between weak family bonds and subsequent ISU. General and Deviant Peer Bonds. Presumably, peers are more likely than parents to encourage unconventional behaviors among adolescents. Consequently, Jessor et al. (199 1) argued that adolescents are at risk for ISU and other problem behaviors whenever they hold closer ties to peers than to parents. In line with this, four studies (Bailey and Hubbard, 1990; Flay et al., 1989; Jessor et al., 1991; Jessor and Jessor, 1977; Kandel et al., 1978) consistently found that adolescents who felt more attached to their peers than their parents were at risk for subsequent ISU. Further, the simple strength of bonds to peers (as opposed to the relative strength of peer versus parental bonds) also predicted ISU. In particular, although Table 4 shows that four studies found no relationship between bonds with peers and subsequent ISU, seven studies found that close ties to peers in general precede marijuana use. For instance, McBride et al. (1991) found that the amount of activities engaged in with peers was positively related to subsequent marijuana use, and Simcha-Fagan et al. (1986) found that marijuana use was more common among adolescents who, at an earlier age, frequently played with other children and maintained friends for a long time. Moreover, involvement with deviant peers who have histories of problem behaviors was especially important in five studies of ISU in Table 4. Adult Smoking, Drinking, Attitudes, and ISU. The comparative importance of peers was echoed in the effects of the substance-specific behaviors of adult role models. Table 4 shows generally inconsistent relationships between ISU by adolescents and licit or illicit substance use by parents or close adults. Several studies found that ISU was more common among adolescents whose parents used alcohol and illicit substances. By contrast, however, many studies have found that ISU among adolescents was not significantly related to (a) cigarette use by parents, (b) controlled levels of alcohol use by adult role models, (c) parental or adult endorsement of ISU, and (d) ISU by parents or close adult role models. The mix of findings does not mean, of course, that adolescents are unaffected by any level of parental substance use. In fact, adolescents are at risk for developing serious drug problems if they have a family history of problematic

ILLICIT SUBSTANCE USE AMONG ADOLESCENTS

2579

alcohol use (see Pandina and Johnson, 1990). Rather, the mix of findings suggests that more prospective research is needed to clarify the size and nature of the effects of adult smoking, drinking, and ISU. Peer Smoking, Drinking, Attitudes, and ISU. The prospective links between peer role models and ISU were, by comparison, far more consistent (see Table 4). There was considerable evidence that ISU among adolescents is preceded by alcohol, marijuana, and narcotic use by their close friends, and there was consistent evidence that ISU among adolescents is preceded by their friends’ use of tobacco. Moreover, studies consistently showed that adolescents were at risk for ISU in the future if they had friends who endorse ISU (Bailey and Hubbard, 1990; Kandel et al., 1978; White et al., 1987), had friends who talked to them about ISU (Kandel et al., 1978) or had peers who offered them illicit substances (Huba et al., 1980; Kandel et al., 1978; Weng and Newcomb, 1989). These findings, along with findings regarding the comparative importance of family and friends (discussed above), suggest that peer role models seem especially influential, and more influential than adult role models. Proximal-Level Social Influences: Perceptions of ISU The final level of social influences concerns the most proximal and immediate causes of ISU. Unlike the previous level (distal-level influences), proximallevel influences are entirely under the control of adolescents and reside not in the substance-specific behaviors of role models, but reside in adolescents’ subjective perceptions that adults and peers either endorse, tolerate, or condone ISU. And, unlike distal-level influences, proximal-level social influences received considerably less attention in prospective studies. In fact, as Table 4 shows, only 10 findings from prospective studies could be considered as proximal-level social influences. Perceptions of Peer and Adult Approval. Supporting the role of adolescents’ subjective perceptions of substance-specific social norms, three separate projects (Jessor et al., 1991; Johnson, 1988; Kandel et al., 1978; White et al., 1987) found that perceptions of peer approval for marijuana use precede an adolescent’s own marijuana use. Even thinking that friends’ approved of cigarette smoking put adolescents at risk for marijuana use in one study (Flay et al., 1989). Interestingly, whereas two studies (Bailey et al., 1992; Kandel et al., 1978) found that perceptions of parental disapproval of marijuana use discouraged later marijuana use, four projects found that adolescent marijuana use was not preceded by perceptions of approval or disapproval for either licit or illicit substance use from parents, teachers, neighbors, and the community in general (see Table 4). Although these results were not entirely consistent, taken together, they suggest

2580

PETRAITIS ET AL.

that perceptions of peer approval might be more important in ISU than perceptions of adult approval.

RESULTS FROM PROSPECTIVE STUDIES OF ATTITUDINAL INFLUENCES Social influences (described above) represent characteristics in adolescents’ immediate social setting (e.g., schools, families, and friendship groups) that shape subjective beliefs that ISU is socially encouraged, condoned, or tolerated. Social influences also have been the main focus of prospective studies of ISU. By contrast, attitudinal influences have received far less attention among prospective studies (representing only 22% of all prospective findings). Nonetheless, attitudinal influences consist of (a) broad sociocultural factors and (b) adolescents’ personal values and beliefs that might contribute to their feelings about their own ISU. The three levels of attitudinal influences are described next and summarized in Table 5. Ultimate-Level Attitudinal Influences: The Community Environment As defined in Table 1, the ultimate level among attitudinal influences represents aspects of the general sociocultural environment that, although beyond the personal control of adolescents, put adolescents at long-term risk for developing positive attitudes toward ISU. Such influences include “disorganized” neighborhoods where drug use and crime rates are high, media depictions that glamorize ISU, ISU by public role models (e.g., athletes or musicians), laws and government policies that might unintentionally promote ISU (e.g., decriminalizing ISU or cutting funds for drug interdiction), and social or political climates that make ISU more likely. Regarding social and political climates, Johnston ( 199 1) argued persuasively that the broad sociocultural environment surrounding adolescents indirectly contributes to ISU by fostering personal values on which ISU is based. As an example, Johnston suggested that the social and political climate during the Vietnam War and the Watergate scandal might have fostered the political alienation of the 1970s and indirectly fed the drug-use epidemic of the 1980s. Furthermore, Johnston speculated that the instantaneous lifestyles of the 1980s and 1990s with rapidly-paced television, video games, and fast foods might be promoting a general need for instant gratification which is manifested in ISU by today’s adolescents. Despite the cogency of Johnston’s (199 1) ideas, few prospective studies have looked at ultimate-level attitudinal influences. In fact, only 10 out of 384 findings (2.6%) could be classified into this cell of Table 1, focusing only on the

P R 3 m

Table 5.

.

Attitudinal Predictors of Illicit Substance Use Organized by Level of Influence, Project and Findings Projects0 Levels and predictorsb Ultimate predictors: The community environment: Availability (+) Employment or academic strain (+) Distal predictors: Personal values and deviant behaviors: Nonconforming, independent from parents (+) Nontraditional or unconventional values (+) Socially critical, tolerant of deviance (+) Politically detached or liberal (+) Low religious commitment (+) Low school commitment (+) School absenteeism (+) Rebelliousness (+) Socially alienated (+) Disobedience, prior deviant behaviors (+) Proximal predictors: Drug-specific attitudes: Positive attitudes toward ISU (+)

Significant relationships’

20,33 10, 14,22,23

% w Nonsignificant relationships ,= CA

I 16,28e, 29 5, 16,30, 35c, 36e 6

G P

35e

F?

6, 17,35e, 36d 2,6, 13,25

c? ;i

6,17,20,28e, 30,3 1,32b(-) 21,31 16,17 20,36e 13, 15, 16,20,21,29 10, 16, 17,20,21,28e, 32a 20,29 13,32b 17 5, 13, 16, 17, l9-21,28d, 32ab, 37,39

28e 23 l6,40 6, 30,31,35c, 37, 38

1, 16, 17,20,22,24,28a

2

aProject numbers correspond to the numbers used in the first column of Table 2. bathe theoretically expected relationship between each predictor and ISIJ is noted in parentheses. ‘Whenever possible, findings are based on reported bivariate relationships between each predictor and ISU, p < .05. Unless otherwise denoted with (-), all significant relationships are in the positive direction.

in

2582

PETRAITIS ET AL.

availability of illicit substances, school-related strain, and employment strain. For instance, we found no prospective studies of how ISU is affected by media depictions of ISU, ISU by national role models, laws and public policies, or changes in social and political climates. Availability. Consistent with expectations, two studies in Table 5 found that lifetime levels of marijuana use were positively related to earlier availability of marijuana, although one study found no such relationship (see Table 5). Employment and School Strain. Previously we described home strain as occurring when adolescents want but do not have closer relationships with their families. Two other forms of strain pertain to employment and schools, occurring whenever employment or academic goals are not being reached. One study (Hammer, 1992) found that being unemployed is a risk factor for subsequent marijuana use. Similarly, Elliott et al. (1985) suggested that adolescents are at risk of ISU when their academic aspirations are being frustrated by limits on their educational opportunities. Three studies show that adolescents’ aspirations might be frustrated by inadequate schools and uncaring teachers (see Table 5). In particular, Elliott et al. reported that young adolescents who felt their schooling was inadequate were at greater risk for subsequent marijuana use. Furthermore, two studies by Kaplan (Kaplan et al., 1984, 1986) found that young adolescents who felt rejected by teachers were more likely to use marijuana by middle adolescence. However, another three studies found no relationship between ISU and having low educational expectations. Distal-Level Attitudinal Influences: Personal Values and Deviant Behaviors Unlike the preceding level of attitudinal influences which centered on the broad cultural environment surrounding adolescents, the second level of attitudinal influences centers on the general values of adolescents that shape their current attitudes toward ISU. And, unlike ultimate-level attitudinal influences, distal-level attitudinal influences have captured substantial attention and represent 17% of all prospective findings. This attention aimed at assessing the relationship between ISU and particular values (including religiosity, social alienation, rebelliousness, independence, and attachment to school). The findings from these studies are summarized in Table 5. Unconventional Social Values of Adolescents. Among the most influential theories of ISU has been Jessor and Jessor’s (1977) problem behavior theory which, in part, argues that adolescents are at risk for ISU and other problem behaviors (e.g, vandalism) if they hold certain personal values, especially unconventional social values. In line with this, the bulk of evidence from prospective

ILLICIT SUBSTANCE USE AMONG ADOLESCENTS

2583

studies found that ISU was more common among adolescents who (a) were nonconforming and sought independence from their parents, (b) held untraditional or unconventional values, (c) were critical of and tolerated deviance from mainstream society, (d) were detached from politics or lacked conservative ideals, (e) were detached from religion, (t) lacked commitment to school, and (g) were rebellious. The one exception seems to be social alienation (see Table 5). Unlike other personal values of adolescents, the bulk of evidence suggests that social alienation does not promote ISU. Disobedience and Prior Deviant Behaviors. As demonstrated in Table 5, ISU might reflect an underlying tendency to engage in various problem behaviors. In particular, about two-thirds of relevant projects found that marijuana use among adolescents was preceded by school absenteeism and deviant behaviors, ranging from minor acts of deviance (e.g., lying and disobedience) to relatively major behavioral problems (e.g., fleeing home, committing crimes, and dealing drugs). For instance, Smith and Fogg (1978) found that marijuana use was more common among adolescents who previously lacked obedience and did not abide by the law. Similarly, Johnston et al. (1978) found that theft and vandalism at age 15 was significantly related to the use of marijuana and other illicit drugs at age19. However, results were not entirely consistent, and about one-third of relevant studies found that marijuana use was not related to problem behaviors, law abidance, or serious delinquency during early ages. For instance, VanKammen and Loeber (1994) found that the initiation of illicit drug use was not preceded by either carrying a concealed weapon or selling drugs. Proximal-Level Attitudinal Influences: Attitudes toward ISU Among the distal-level attitudinal influences were adolescents’ personal values and attitudes toward society in general. By contrast, the proximal-level attitudinal influences focus on adolescents’ attitudes toward ISU in particular, and target their subjective perceptions regarding the physiological, legal, and psychological costs and benefits of ISU. Lending support to the role of attitudes, Kandel et al. (1978) found that 16 year olds were more likely to use marijuana if, as 15 year olds, they had some desire to use marijuana, did not consider its use to be personally harmful, did not fear any negative consequences of its use, or thought marijuana should be legalized. Furthermore, Johnson (1988) found that marijuana use was related to expecting positive consequences, whereas marijuana abstinence was related to expecting negative consequences Corn marijuana. Similarly, Bailey et al. (1992) found that adolescents were more likely to use marijuana two years later if they had perceived relatively few risks to its use, and Levy and Pierce (1990) found that 12-year-old adolescents who had never used marijuana but who held relatively positive attitudes toward licit and illicit substances were at risk for

2584

PETRAITIS ET AL.

initiating marijuana use by age 15. Moreover, Kaplan et al. (1986) found that attitudes which are based on prior experiences with marijuana use put adolescents at significant risk for continued marijuana use in the future. Finally, two studies (Jessor et al. 199 1; Jessor and Jessor, 1977) in Table 5 concluded that adolescents who thought the benefits of marijuana use exceeded its costs were more likely to use marijuana as young adults. In fact, only one of eight prospective studies reported a nonsignificant relationship between adolescents’ attitudes toward substance use and subsequent marijuana use.

RESULTS FROM PROSPECTIVE STUDIES OF INTRAPERSONAL INFLUENCES Whereas attitudinal influences focus in part on adolescents’ general values, intrapersonal influences focus on adolescents’ fundamental demographic characteristics, stable personality traits, and more transitory affective states. Demographic characteristics, personality traits (e.g., aggressiveness), and affective states (e.g., low self-esteem) have played important roles in several theories of ISU (see Brook et al., 1990; Kaplan et al., 1982, 1984; Simons et al., 1988), and appeared in 27% of prospective analyses. The findings from these analyses (described below) are grouped into ultimate-, distal-, and proximal-level influences, and are summarized in Table 6. Ultimate-Level Intrapersonal Influences: Demographics and Personality Traits Seventy-one findings focused on the relationship between ISU and either demographic characteristics or stable personality traits. Eighteen of those findings targeted adolescents’ basic demographics and suggest that ISU is not evenly distributed across demographic categories (see Table 6). In particular, five studies found that ISU was more common among males than females, and five studies found that ISU was more common among Whites than among Blacks, Hispanics, or non-Whites in general. The remaining 53 findings among ultimate-level intrapersonal influences focused on adolescents’ stable personality traits. Although there are potentially dozens of intrapersonal characteristics that might contribute to ISU, leading personality theorists (e.g., Goldberg, 1993) have argued that personality consists of numerous subtraits and a smaller set of supertraits. Specifically, these theorists have argued that most, if not all, personality traits can be arranged into a hierarchy that is capped by five supertraits, often called ihe bigfive. Although theorists have disagreed on the exact names and natures of supertraits, they have generally agreed that an individual’s personality can be adequately characterized

. Table 6.

Intrapersonal Predictors of Illicit Substance Use Organized by Level of Influence, Findings, and Projects Project.9

Predictorsb Ultimate level: Demographics and personality: Male White Low persistence and will to achieve (+) Weak emotional stability & control (+) Disinhibition; assertiveness; thrill seeking (+) Aggressiveness;hostility(+) Low intelligence; academic problems Distalinfluences:Affectivestates: Anxiety (+) Poor self-esteem and self-concept(+) Depression or depressed mood (+)

Significant relationshipsc

Nonsignificant relationships

7b, 14,22,24,29

17, 18,20,31,35c, 36cd

z

F

% 13,20,22,31,37 5, 12,28e, 32a, 6(-) 4,5, 13,26ab, 28d

6, ‘labc, 27,28e, 30, 33,38 5,7c, 12, 19,28de, 32a

7ab, 13, 15, 16,20,28de, 29 7b(-), 28e (-), 32b(-) 8, 13,30(-)

32a 20,28d,30(-)

15,40 2-5, 28e, 30,3 1,40 6,30 5,36e 6, 7b, 30,3 1,37 7c, 17,22

5,7c,22,28e,31,33 4, 5, 16, 17,20,23,25,28e,30,36e,40 5,6, 7c,21,27,28e,31, 33,40

Vroject numbers correspond to the numbers used in the first column of Table 2. bathe theoretically expected relationship between each predictor and ISU is noted in parentheses. cWhenever possible, findings are based on reported bivariate relationships between each predictor andKU, p < .05. Unless otherwise denoted with (-), all significant relationships are in the positive direction.

2

>

8 E x

z

d

2586

PETRAITIS ET AL.

by his or her (a) task persistence, behavioral constraint, impulsivity, and will to achieve; (b) emotional stability, emotional well-being, neuroticism, and psychological adjustment, (c) extroversion/introversion, assertiveness, and social activity; (d) friendliness, hostility, and level of socialization; and (e) intellect or general intelligence. All five dimensions were represented in prospective studies of ISU. Task Persistence and Will to Achieve. The first dimension represents the degree to which a person is able to persist on tasks, focus on achieving long-range goals, and control ones’ impulses. Some theorists (e.g., Jessor and Jessor, 1977) have suggested that adolescents who have little interest in long-range goals are at risk for ISU. However, as Table 6 shows, prospective studies generally have not supported this contention. In fact, although four studies reported that children were at risk for marijuana use if they were impulsive, unable to delay gratitication, or had attention deficit disorders, a fifth study (Brook et al., 1990) found that children who were unable to persist on a task were actually at lower risk for marijuana use than children who showed stronger willingness to achieve and persevere. Moreover, nine studies found that ISU among adolescents could not be predicted from earlier task persistence (Block et al., 1988; Shedler and Block, 1990; White et al., 1987), purposiveness (Baumrind, 1985), compulsiveness (Simcha-Fagan et al., 1986), importance placed on future plans and goals (Bailey and Hubbard, 1990), need for achievement (Winetield et al., 1993), or problems with attention deficits (Barkley et al., 1990). Emotional Stability and Control. The second dimension of personality represents adolescents’ long-term emotional stability and ability to control their emotions. Some theorists (e.g., Brook et al., 1990; Simons et al., 1988) have contended that adolescents who are emotional and chronically volatile might be motivated to use illicit substances as a means of coping. Moreover, the bulk of prospective evidence has supported this contention, showing that the ability to control one’s emotions might deter ISU (see Table 6). In particular, marijuana use is more common among adolescents who, at earlier times, had problems with anger (Flay et al., 1989), temperament (Lerner and Vicar-y, 1984), emotional disabilities (Baumrind, 1985), tantrums and inappropriate negative emotions (Block et al., 1990), emotional maladjustment (Vicary and Lemer, 1983), and who showed signs of global psychological problems (White, 1992). Extroversion, Assertiveness, Social Disinhibition, and Thrill Seeking. The third personality dimension focuses on people’s introversion and shyness on one pole, and their extroversion, assertiveness, and disinhibition on the other. As shown in Table 6, this dimension might contribute to ISU. In particular, the bulk of relevant studies found that children and young adolescents who were socially

ILLICIT SUBSTANCE USE AMONG ADOLESCENTS

2587

extraverted, assertive and disinhibited were at risk for subsequent ISU. Furthermore, two studies (Pedersen, 199 1; Teichman et al., 1989) concluded that ISU was more common among adolescents who sought thrills and new sensations. Aggressiveness and Hostility. The fourth dimension concerns the levels of aggressiveness and hostility with which an adolescent treats other people. As shown in Table 6, although findings were not uniformly consistent, several of the relevant prospective studies did find that aggressiveness and hostility at earlier ages raised the chances of ISU at later ages. For instance, Fergusson et al. (1993) found conduct problems among 8 year olds predicted the onset of marijuana use among 15 year olds. These findings are in line with theoretical assertions that inadequate social interaction skills contribute to ISU (Simons et al., 1988). Intelligence. The fifth and final intrapersonal characteristic is an adolescent’s overall intelligence. Although several prospective studies found that intelligence was related to ISU, conflicting results make it hard to determine whether the relationship is positive or negative. On one hand, six studies in Table 6 found that marijuana use was more common among students who performed poorly in school. Furthermore, Fleming et al. (1982) found that marijuana use was more common among males who previously had learning problems. On the other hand, however, Fleming et al. (1982) found that first-grade children who scored high on tests of intelligence and school-readiness were at risk for marijuana use as adolescents. Similarly, White et al. (1987) and Smith and Fogg (1979) reported that ISU was positively related to grade point averages in school. Distal-Level Intrapersonal Influences: Affective States On the distal level, intrapersonal influences represent affective states that might provide the motivation for ISU and might undermine adolescents’ refusal skills. Three affective states---low self-esteem, elevated anxiety, and depressed mood-captured considerable attention in prospective research, representing nearly 9% of all prospective findings. However, as shown in Table 6, there was dramatically little evidence that these affective states are related to ISU. The lack of evidence was most dramatic with regards to low self-esteem where 11 of 12 studies found no significant relationship betvveen low self-esteem and subsequent ISU. The evidence was also quite powerful with regards to anxiety and depressed mood. In particular, six studies found no significant relationship between anxiety and subsequent ISU, only two found a positive relationship, and one actually found a negative relationship such that adolescents who experimented with marijuana had less anxiety as children than adolescents who abstained from marijuana use. Similarly, the vast majority of studies found no significant relationship between depressed mood and subsequent ISU, and those studies that found signiti-

2588

PETRAITIS ET AL.

cant relationships reported a mix of positive and negative effects. Taken together, these findings suggest that inadequate emotional adjustment might have little effect on subsequent ISU. In fact, Hansel1 and White (1991) provided evidence that inadequate emotional adjustment is more likely a consequence than a cause of alcohol consumption and ISU. Proximal-Level Intrapersonal Influences: Self-Efficacy Proximal influences focus on substance-specific beliefs. In the case of proximal-level intrapersonal influences, these beliefs take the form of adolescents’ substance-specific self-efficacy. According to Ajzen (1988) and Bandura (1982, 1986), self-efficacy is among the most proximal causes of any behavior and represents a person’s belief that he or she has the ability to perform the behavior. In the case of ISU, self-efficacy represents adolescents’ beliefs that they are personally capable of either obtaining illicit substances and engaging in ISU, or refusing all pressures to use illicit substances. Moreover, self-efficacy has been linked to ISU in one prospective study. Although this finding has not yet been replicated in another prospective study of ISU, Wills et al. (1989) found that 14year-old adolescents were more likely to abstain from marijuana use if they felt capable of avoiding drugs when they were 12 years old.

RESULTS FROM PROSPECTIVE STUDIES OF INTENTIONS AND PRIOR BEHAVIOR Fishbein and Ajzen (1975; Ajzen and Fishbein, 1980) argued convincingly that all causes of behavior act through three proximal-level cognitive variables (attitudes, social normative beliefs, and self-efficacy), and that these, in turn, act through intentions. Empirical studies (described below) have supported these notions and have supported the truism that the best predictor of future behavior is past related-behavior. In this section (and in Table 7) we present results from 5 1 analyses concerning the relationship between substance-specific intentions, prior substance use behaviors, and ISU. Substance-Specific Intentions Table 7 presents adolescents’ substance-specific intentions which, according to Ajzen and Fishbein (1980), are the most immediate precursors of ISU. In line with this, the few prospective studies that included measures of intentions invariably found that adolescents who had some intentions to use marijuana in their lifetimes were more likely to try marijuana before becoming young adults (see Table 7).

C

Table 7.

Ki

Immediate Precursors of Illicit Substance Use Projectsa Predictorsb Intentions regarding ISU (+) Prior behaviors: Prior cigarette use (+) Prior alcohol use (+) Prior marijuana use (+) Prior narcotic use (+)

Positive flndingsC

h 4 0 Nonsignificant findings $

13,34bc, 35b 9, 11, 13,20,21,24,29,32b, 34a 9, II, 13, 16, 18,20,21,23,24,29,34ac, 35b, 36d 1,8, 10, 11, 1416, 18,21-23,28a, 29,34ac, 35ab, 36d 16,21,28d, 29,34ac, 39

18 9,35abc, 36d

aProject numbers correspond to the numbers used in the tirst column of Table 2. “The theoretically expected relationship between each predictor and ISU is noted in parentheses. cWhenever possible, findings are based on reported bivariate relationships behveen each predictor and ISU, p C .05.

g E c, 2 ;i

2590

PETRAITIS ET AL.

Prior Substance Use In another influential model of ISU, Kandel (1980, 1989) suggested that involvement with drugs follows a predictable sequence that begins with cigarettes and alcohol, progresses to marijuana, and for some people culminates in use of “harder” drugs like cocaine. Although there is reason to doubt the causal nature of this sequence (see Baumrind, 1985; Miller, 1994; Miller and Flay, 1993), these prospective studies suggest that adolescents who have used a variety of substances in the past are at risk for using marijuana in the future. This is especially true of adolescents who have used cigarettes, alcohol, or marijuana. In fact, Table 7 shows that prior cigarette smoking, earlier alcohol use, and prior marijuana use preceded subsequent levels of marijuana use in at least 90% of relevant studies.

CONCLUSIONS AND RECOMMENDATIONS The simple goal of this paper was to summarize what is known about prospective predictors of ISU among adolescents. Achieving this goal, however, was complicated by the existence of 384 significant and nonsignificant findings from 58 different studies conducted by researchers who had vastly different theoretical interests and methodological approaches. Nonetheless, our organization of findings into different types and different levels of influence pointed to some general conclusions about the predictors of ISU. Some conclusions pertain to findings that have been repeated throughout prospective studies, others pertain to gaps in existing research, and still others pertain to directions for future research. Well-Established Findings Despite wide methodological differences in prospective studies, many of them led to some common conclusions. First, ISU is rarely the first sign of trouble for adolescents. More often, ISU follows other problem-behaviors, some of which involve the use of licit substances (e.g., cigarettes and alcohol) whereas others involve acts of delinquency (ranging from truancy to serious crimes). Second, studies suggested that adolescents are usually prepared cognitively for ISU. Rarely do adolescents use illicit substances without first believing that the potential benefits of ISU exceed the potential costs, anticipating that other people support ISU, or consciously intending to use illicit drugs. Third, ISU usually occurs after exposure to other substance users. Although it is less clear whether the substance-specific attitudes and behaviors of adults affect an adolescent’s own ISU, it is very clear that adolescents who use illicit substances usually have friends who have used and approved of ISU. Fourth, ISU is closely related to deviant peer

ILLICIT SUBSTANCE USE AMONG ADOLESCENTS

2591

bonding such that attachment to deviant peers regularly precedes ISU. To a lesser extent, ISU is also related to detachment from and rebellion against religion, school, and family. Fifth, ISU is more common when adolescents do not control their emotions (e.g., when they are temperamental), when they are extraverted and socially disinhibited, and when they are aggressive around other people. By contrast, and counter to conventional wisdom, ISU is not more common when adolescents are anxious, depressed, or have poor self-esteem. Finally, although parental permissiveness seems to have little direct effect on ISU, the absence of parents who offer emotional support for their children or the absence of an intact family might lay the long-term foundations for ISU among adolescents. Gaps in the Existing Research Despite these common conclusions, there are important gaps among prospective studies. Some of these gaps can be identified by comparing the list of constructs from the cells in Table 1 to the list of variables in Tables 4-7. When this is done, gaps can be seen around ultimate-level social influences in general, and around the effects of parents in particular, Despite the fact that many theories of ISU point toward the home as the breeding ground for ISU, prospective studies have not addressed the role of infrequent opportunities for rewards from family members, negative evaluations from parents, or unconventional values among parents. Consequently, these constructs should be the focus of future prospective studies. Other gaps can be identified by looking at the number of findings that fall into each cell of Table 1. Table 8 provides this breakdown and shows that the attentions of researchers have not been evenly distributed across the different types and levels of influence. Most notably, there have been relatively few studies of ultimate-level attitudinal influences and factors in adolescents’ broad cultural milieu that might affect ISU. As we noted, only 10 (or 2.6%) of the tindings targeted such influences on ISU. Consequently, prospective studies have told very little about whether or how ISU is affected by local crime and employment rates, poor career opportunities, infrequent opportunities for rewards at school, negative evaluations from teachers, media depictions of ISU, and weak public policies on ISU. This is unfortunate because all of these factors have been incorporated in the dominant theories of ISU. As Table 8 also shows, distal-level intrapersonal influences represent another area that needs more research. Although nearly 9% of all findings targeted just three variables-anxiety, low self-esteem, and depression--surprisingly few studies found that any of these affective states precede ISU. Consequently, a need might exist for more prospective studies to clarify the influence of affective states on ISU. As researchers, we need to understand whether the absence of signiti-

2592

PETRAITIS ET AL. Table 8.

Number of Potentially Replicated Findings by Types and Levels of Influence Types of influence Level of influence Ultimate Distal Proximal

Social

Attitudinal

Intrapersonal

56 76 12

10 67 8

71 33 0

Immediatea

51

OImmediate influences are not subdivided into social, attitudinal and intrapersonal predictors because they are seen as the product of all three types of influence.

cant findings reflects problems with prospective designs or with theories about how affective states contribute to KU. For instance, we do not know whether selfesteem was unrelated to ISU in past studies because self-esteem (a) simply does not affect ISU-something that is a distinct possibility, or (b) does affect ISU but it is so transient that prospective studies with only one or two waves of data cannot capture its effects. Further, Table 8 calls attention to proximal-level social predictors. This cell consisted of adolescents’ perceptions that adults, parents, peers, and close friends approve of ISU. Somewhat surprisingly, the link between such perceptions and subsequent ISU was far from consistent. However, closer inspection of the studies in this cell suggests that ISU is generally related to perceptions that peers and friends approve of ISU, but is not related to similar perceptions of parents and other adults. Clearly, more research is needed to understand why this is the case. Finally, despite recent theoretical interest in self-efficacy (e.g., Ajzen, 1988), there has been surprisingly little attention in prospective studies to the effects of substance-specific self-efficacy. In fact, only one study (Wills et al., 1989) showed how ISU was related to adolescents’ beliefs about their abilities to refuse illicit substances. Thus, the link between self-efficacy and ISU has not been replicated in prospective research. Directions for Future Research Our review has six clear implications for future research and theory on the causes of ISU, implications that overlap and lead into each other. 1.

More research is needed on ultimate-level social influences (especially on rewards from, evaluations from, and values of parents), ultimate-level

.

ILLICIT SUBSTANCE USE AMONG ADOLESCENTS

4.

2593

attitudinal influences (especially on characteristics of neighborhoods, schools, teachers, the mass media, and public policies), and proximal-level intrapersonal influences (e.g., substance-related self-efficacy). More research is needed to clarify whether and how affective states (e.g., anxiety, low self-esteem, and depression) contribute to ISU. Although prospective studies rarely found significant relationships among ISU and affective states, such states might contribute to ISU in a manner that was not detected in previous studies. For instance, the effect of low self-esteem on ISU might have been mediated through or moderated by other variables that were not included in previous analyses involving self-esteem. We encourage meta-analytic reviews in this field. Our review aimed at providing breadth. In effect, it used a wide-angle lens to summarize prospective research on a wide variety of predictors. Given that no such picture had ever been provided, a broad approach was justified. However, subsequent reviews now ought to supplement this review by providing more depth. That is, subsequent reviews ought to use a telephoto lens and focus on a detailed analysis of the link between ISU and one or two predictors, especially for those predictors that show inconsistent relationships with ISU. For instance, they might target the inconsistent link between ISU and low religious commitment and determine if the inconsistency is because 1) those studies with nonsignificant findings showed positive relationships that fell just short of achieving the point of statistical significance, 2) the operational definitions of religious commitment of ISU varied from study to study, or 3) because some other moderating variable was at work that led some studies to find nonsignificant findings while leading other studies to significant findings. Of course, given the breadth of variables that are related to ISU, several me&analytic reviews will be necessary. Future studies should focus on testing more complex causal pathways that include mediating and moderating variables. Too frequently, past research was limited to looking for bivariate relationships between ISU and potential predictor variables. Although bivariate relationships described whether a variable is related to ISU, they told relatively little about (a) the mechanisms by which it contributed to ISU and (b) the extent to which those mechanisms generalize to adolescents at different ages (e.g., early adolescents vs early adults) in different generations (e.g., during the 1980s vs the 199Os), in different neighborhoods (e.g., rural vs urban), in different socioeconomic strata (e.g., poor vs affluent), in different ethnic groups (e.g., Latinos vs Asians) and in different countries (e.g., developing vs industrial). Consequently, future studies should rely more on theoretically derived hypotheses and multivariate analyses that test complex causal pathways and the factors that interact with or moderate the pathways. Moreover, where possible, future

2594

PETRAITIS ET AL.

reviews should focus on conducting meta-analyses of such causal pathways. Fortunately, there have been recent advances in conducting meta-analyses of causal paths (see Shadish, 1996). Unfortunately, however, such metaanalyses would require a critical mass of studies that quantify comparable causal pathways (i.e., dealing with comparable exogenous, mediating, and endogenous variables), something that does not yet exist in this field. 5. Future analyses should consider how different levels of influence work together to shape ISU. Although the primary intent of Table 1 was to organize the potential causes of ISU, it also suggests possible causal orderings among different levels of influence. In general, variables that are located on higher levels in Table 1 might exert their influence on ISU by contributing to variables on more proximal tiers. Specifically, variables that are among the ultimate influences might contribute to ISU by first contributing directly to distal influences, and then contributing indirectly to proximal and immediate influences. For instance, lack of parental support (an ultimate-level variable) might have its greatest influence on an adolescent’s emotional bonds to deviant peers (a distal-level variable) which, in turn, might contribute to perceptions of norms concerning ISU and decisions to use illicit substances. Similarly, distal-level variables (such as low self-esteem) might contribute indirectly to ISU by contributing directly to proximal-level variables (such as substance-specific self-efficacy). We encourage more prospective research on the complex causal pathways between different levels of influence. 6. We encourage more attention to the development of larger theoretical models that integrate all of the constructs from Table 1, constructs which came from several midlevel theories of ISU. Like a recipe that provides a list of ingredients and instructions for combining the ingredients, the next step in understanding ISU requires (a) a list of constructs, like those provided in Table 1, and (b) an integrative theoretical model that carefully details how those constructs combine or work together to promote ISU. Several theorists have accepted this challenge, most notably Jessor et al. (1991) and Simons et al. (1988), and more recently Flay and Petraitis (1994). We believe that attention to integrative theories of ISU provides a critical step in understanding ISU. Moreover, given the apparent complexity in the etiology of ISU and recent advances being made in artificial-science paradigms, another critical step might lie with the development of nonlinear models. Although the onset of ISU might come from a linear combination of different influences, different influences might also combine in more complex, less certain, less universal, and more chaotic ways---ways that might tell us why some adolescents never use illicit substances, why others experiment with ISU, and why a few adolescents regularly use and misuse these substances.

ILLICIT SUBSTANCE USE AMONG ADOLESCENTS

2595

APPENDIX Grade 7 Correlates of Marijuana Use in Grade 9 for Subjects Who Had Not Used Marjuana in Grade 7, as Reported by Flay et al. (1989) Pearson correlationsa Social influences: Father’s job level Mother works Mother’s job level Lives with both parents Familyconflicts Parents influence child’s choice of friends Reports being heavily influenced by adult role models Reports being heavily influenced by close friends Reports that friends have more influence than parents Closeness to parents is more important than good friends Has close friends who are older Frequency of attending parties Perceived severity of drug problems in school Perceived prevalence of smoking among adults and peers Alcohol use by favorite adults Number of close adults who smoke Mother smokes Father smokes Older sister smokes Older brother smokes Parental disapproval of smoking Number of close friends who have tried smoking Number of close friends who smoke every week Number of friends who approve of smoking Number of times friends have offered a cigarette Attitudinal influences: Educational aspirations Career aspirations Frequency of church attendance Believing that health is more important than fun Hoping to live a long life Rebelliousness Frequency of participation in sports Believing TV information about health Believing that TV influences teen smoking Feeling immune to the influence of cigarette ads Not wanting parents to smoke Concerned that smoking is dangerous to other people Concerned that smoking is dangerous to oneself

.

.030 .029 ,040 -.075* .092* ,078 ,046 .018 .091* -.105* .079* .134* .008 .079* .115* .139* .093* .042 .112* .I 14* -.002 .246* .129* .135* .253* ,053 -.035 -.090* -.I61 .014 .178* .046 .040 .017 .166* .124* -.090* -.143

PETRAITIS ET AL.

2596

Pearson correlations” Intrapersonal influences: Gender (male) Ethnicity (White) Easily angered (emotional control) Academicgrades Stress (anxiety) Perceived ability to refuse an offer of a cigarette Effort put into not smoking cigarettes Intentions and prior behaviors: Thinking that future smoking is personally likely Perceived probability of asking to try a cigarette Perceived probability of asking to try chewing tobacco Perceived probability of getting drunk in the future Perceived probability of smoking marijuana in the future Having smoked cigarettes during or before seventh grade Having drank alcohol during or before seventh grade

.030 .074* .145* .I 13* .094* ,030 -.013 .I84 .205* .144* .230* .238* .262* .210*

‘Because of missing data, the sample sizes varied from 1,541 to 3,412. *p < .OS using the Bonferroni criteria.

ACKNOWLEDGMENT The preparation of this paper was supported by Grant DA06307 from the National Institute on Drug Abuse.

REFERENCES AJZEN, 1. (I 988). Attitudes, Personality and Behavior. Chicago, IL: Dorsey Press. AJZEN, I., and FISHBEIN, M. (I 980). Understanding Attitudes and Predicting Social Behavior. Englewood Cliffs, NJ: Prentice-Hall. BAILEY, S., FLEWELLING, R. L., and RACHAL, J. V. (1992). Predicting continued use of marijuana among adolescents: The relative influence of drug-specific and social context factors. J. Health Sot. Behuv. 33: 5 l-66. BAILEY, S., and HUBBARD, R. L. (1990). Developmental variation in the context of marijuana initiation among adolescents. J. Health Sot. Behuv. 3 1: 58-70. BANDURA, A. (1982). Self-efficacy mechanism in human agency. Am. Psychol. 37: 122-147. BANDURA, A. (1986). Social Foundations of Thought and Action: A Social Cognitive Theory. New York, NY: Prentice-Hall. BARKLEY, R. A., FISCHER, M., EDELBROCK, C. S., and SMALLISH, L. (1990). The adolescent outcome of hyperactive children diagnosed by research criteria: I. An I-year prospective follow-up study. J. Am. Acud. Child Adolesc. Psychiatry 29: 546-557. BAUMRIND, D. (1985). Familial antecedents of adolescent drug use: A developmental perspective. In C. L. Jones and R. J. Battjes (Eds.), Etiology of Drug Abuse: Implications for Prevention (pp. 13-44). Rockville, MD: National Institute on Drug Abuse, Research Monograph 56.

ILLICIT SUBSTANCE USE AMONG ADOLESCENTS

2597

BLOCK, J, BLOCK, J. H., and KEYES, S. (1988). Longitudinally foretelling drug usage in adolescence: Early childhood personality and environmental precursors. Child Dev. 52: 336-355. BROOK, J. S., BROOK, D. W., GORDON, A. S., WHITEMAN, M., and COHEN, P. (1990). The psychosocial etiology of adolescent drug use: A family interactional approach. Genet. Sot. Gen. Psychol. Monogr. 116: 1 I l-267. DEMBO, R., WILLIAMS, L., LAVOIE, L., SCHMEIDLER, J., KERN, J., GETREU, A., BERRY, E., GENUNG, L., and WISH, E. D. (1990). A longitudinal study of the relationships among alcohol use, marijuanaihashish use, cocaine use, and emotional/psychological functioning problems in a cohort of high-risk youth. Int. J. Addict. 25: 1341-1382. ELLICKSON, P. L., HAYS, R. D., and BELL, R. M. (1992). Stepping through the drug use sequence: Longitudinal scalogram analysis of initiation and regular use. J. Abnorm. Psychol. 101: 441451. ELLIOTT, D. S., HUIZINGA, D., and AGETON, S. S. (1985). Explaining Delinquency and Drug Use. Beverly Hills, CA: Sage. ENSMINGER, M. E., BROWN, C. H., and KELLAM, S. G. (1982). Sex differences in antecedents of substance use among adolescents. J. Sot. Issues 38: 2542. FARRELL, A. D. (1993). Risk factors for drug use in urban adolescents: A three-wave longitudinal study. J. Drug Issues 23: 443462. FERGUSSON, D. M., LYNSKEY, M. T., and HORWOOD, L. J. (1993). Conduct problems and attention deficit behaviour in middle childhood and cannabis use by age 15. Aust. IV. Z. J. Psychiatty 27: 673682.

FISHBEIN, M., and AJZEN, I. (1975). Beliefs, Attitude, Intention and Behavior: An Introduction to Theory and Research. Reading, MA: Addison-Wesley. FLAY, B. R., MILLER, T. Q., and KOEPKE, D. (1989). Grade 7 Correlates of Onset of Substance Experimentation and Use by Grade 9. Unpublished manuscript, University of Illinois at Chicago. FLAY, B. R., and PETRAITIS, J. (1991). Methodological issues in drug abuse prevention research: Theoretical foundations. In C. Leukefeld and W. Bukoski (Eds.), Drug Abuse and Prevention Research: Methodological Issues. Washington, DC: National Institute of Drug Abuse Research Monographs. FLAY, B. R., and PETRAITIS, J. (1994). The theory of triadic influence: A new theory of health behavior with implications for preventive interventions. In G. Albrecht (Ed.), Advances in Medical Sociology, Vol. 4 (pp. lw4). Greenwich, CT: JAI Press. FLEMING, J. P., KELLAM, S. G., and BROWN, C. H. (1982). Early predictors of age of first use of alcohol, marijuana, and cigarettes. Drug Alcohol Depend. 9: 285-303. FRIEDMAN, A. S., UTADA, A. T., GLICKMAN, N. W., and MORRISSEY, M. R. (1987). Psychopathology as an antecedent to, and as a “consequence” of, substance use in adolescence. J. . Drug Educ. 17: 233-244. GOLDBERG, L. R. (1993). The structure of phenotypic personality traits. Am. Psychol. 41: 26-34. HAMMER, T. (1992). Unemployment and use of drug and alcohol among young people: A longitudinal study in the general population. Br. J. Addict. 87: I57 l-l 58 I. HANSELL, S., and MECHANIC, D. (1990). Parent and peer effects on adolescent health behavior. In K. Hurrelmann and F. Lose1 (Eds.), Health Hazards in Adolescence: Prevention and Intervention in Childhood and Adolescence, Vol. 8 (pp. 43-65). New York, NY: Walter de Gruyter. HANSELL, S., and WHITE, H. R. (1991). Adolescent drug use, psychological distress, and physical symptoms. J. Health Sot. Behav. 32: 288-301. HAWKINS, J. D., CATALANO, R. F., and MILLER, J. Y. (1992). Risk and protective factors for alcohol and other drug problems in adolescence and early adulthood: Implications for substance abuse prevention. Psychol. Bull. 112: 64-105.

2598

PETRAITIS ET AL.

HUBA, Cl. J., WINGARD, J. A., and BENTLER, P. M. (1980). Longitudinal analysis of the role of peer support, adult models, and peer subcultures in beginning adolescent substance use: An application of setwise canonical correlation methods. Mulfivur. Behav. Res. 15: 259-279. HUBA, G. J., WINGARD, J. A., and BENTLER, P. M. (I 98 I). Intentions to use drugs among adolescents: A longitudinal analysis. fnt. J. Addict. 16: 33 l-339. JESSOR, R., DONOVAN, J. E., and COSTA, F. M. (1991). Beyond Adolescence: Problem Behavior and Young Adult Development. New York, NY: Cambridge University Press. JESSOR, R., and JESSOR, S. L. (I 977). Problem Behavior and Psychosocial Development. New York, NY: Academic Press. JOHNSON, C. A., PENTZ, M. A., WEBER, M. D., DWYER, J. H., BAER, N., MAcKINNON, D. P., HANSEN, W. B., and FLAY, B. R. (1990). Relative effectiveness of comprehensive community programming for drug abuse prevention with high-risk and low-risk adolescents. J. Consult. Clin. Psychol. 58: 447456. JOHNSON, R. J., and KAPLAN, H. B. (I 990). Stability of psychological symptoms: Drug use consequences and intervening processes. J. He&h Sot. Behav. 3 I : 277-29 I. JOHNSON, V. (1988). Adolescent alcohol and marijuana use: A longitudinal assessment of a social learning perspective. Am. J. Drug Alcohol Abuse 14: 419439. JOHNSON, V., and PANDINA, R. J. (1991). Effects of family environment on adolescent substance use, delinquency, and coping styles. Am. J. Drug Alcohol Abuse 17: 71-88. JOHNSTON, L. D. (1991). Toward a theory of drug epidemics. In L. Donohew, H. E. Sypher, and W. J. Bukoski (Eds.), Persuasive Communication and Drug Abuse Prevention. Hillsdale, NJ: Lawrence Earlbaum Associates. JOHNSTON, L. D., O’MALLEY, P. M., and EVELAND, L. (1978). Drugs and delinquency: A search for causal connections. In D. B. Kandel (Ed.), Longitudinal Research on Drug Use: Empirical Findings and Methodological Issues (pp. 137-156). Washington, DC: Hemisphere. KANDEL, D. B. (1980). Developmental stages in adolescent drug involvement. In D. J. Lettieri, M. Sayers, and H. W. Pearson (Eds.), Theories of Drug Abuse: Selected Contemporary Perspectives (pp. 120-127). Rockville, MD: National Institute on Drug Abuse, Research Monograph 30. KANDEL, D. B. (1989). Issues of sequencing of adolescent drug use and other problem behaviors. Drugs Sot. 3: 55-76. KANDEL, D. B., KESSLER, R. C., and MARGULIES, R. Z. (1978). Antecedents of adolescent initiation into stages of drug use: A developmental analysis. In D. B. Kandel (Ed.), Longitudinal Research on Drug Use: Empirical Findings and hfethodological Issues (pp. 7>99). Washington, DC: Hemisphere. KANDEL, D., SIMCHA-FAGAN, 0.. and DAVIES, M. (1986). Risk factors for delinquency and illicit drug use from adolescence to young adulthood. J. Drug Issues 16: 67-90. KAPLAN, H. B., MARTIN, S. S., JOHNSON, R. J., and ROBBINS, C. (1986). Escalation of marijuana use: Application of a general theory of deviant behavior. J. Health Sot. Behav. 27: 44 61. KAPLAN, H. B., MARTIN, S. S., and ROBBINS, C. (1982). Application of a general theory of deviant behavior: Self-derogation and adolescent drug use. J. Health Sot. Behav. 23: 274-294. KAPLAN, H. B., MARTIN, S. S., and ROBBINS, C. (1984). Pathways to adolescent drug use: Selfderogation, peer influence, weakening of social controls, and early substance use. J. Health Sot. Behav. 25: 270-289. KELLAM, S. G., BROWN, C. H., and FLEMING, J. P. (I 982). The prevention of teenage substance use: Longitudinal research and strategy. In T. J. Coates, A. C. Peterson, and C. Perry (Eds.), Promoting Adolescent Health: A Dialogue on Research and Practice (pp. I7 I-200). New York, NY: Academic Press.

.

ILLICIT SUBSTANCE USE AMONG ADOLESCENTS

2599

KELLAM, S. Cl., BROWN, C. H., RUBIN, B. R., and ENSMINGER, M. E. (1983). Paths leading to teenage psychiatric symptoms and substance use: Developmental epidemiological studies in Woodlawn. In S. B. Guze, F. J. Earls, and J. E. Barrett (Eds.), Childhood fsychopatho/ogv and Development (pp. 17-S I). New York, NY: Raven Press. LERNER, J. V., and VICARY, J. R. (1984). Difficult temperament and drug use: Analyses from the New York Longitudinal Study. J. Drug Educ. 14: l-8. LETTIERI, D. J., SAYERS, M., and PEARSON, H. W. (Eds.) (1980). Theories on Drug Abuse: Selected Contemporary Perspectives. Rockville, MD: National Institute on Drug Abuse, Research Monograph 30. LEVY, S. J., and PIERCE, J. P. (1990). Predictors of marijuana use and uptake among teenagers in Sydney, Australia. Int. J. Addict. 25: 1179-l 193. MADDAHIAN, E., NEWCOMB, M. D., and BENTLER, P. M. (1988). Adolescent drug use and intentions to use drugs: Concurrent and longitudinal analyses of four ethnic groups. Addict. Behav. 13: 191-195. MCBRIDE, A. A., JOE, G. W., and SIMPSON, D. D. (1991). Prediction of long-term alcohol use, drug use, and criminality among inhalant users. Hisp. J. Behav. Sci. 13: 315-323. MILLER, T. Q. (1994). A test of alternative explanations for the stage-like progression of adolescent substance use in four national samples. Addict. Behav. 19(3): 287-293. MILLER, T. Q., and FLAY, B. R. (I 993). A Framework for Understanding Why Adolescents Experiment with Alcohol and Cigarettes before Marijuana. Unpublished manuscript, Prevention Research Center, University of Illinois at Chicago. MONCHER, M. S., HOLDEN, G. W., and SCHINKE, S. P. (1991). Psychosocial correlates of adolescent substance use: A review of current etiological constructs. fnt. J. Addict. 26: 377-414. NEWCOMB, M. D., and BENTLER, P. M. (1986). Frequency and sequence of drug use: A longitudinal study from early adolescence to young adulthood. J. Drug Educ. 16: IO l-l 20. NEWCOMB, M. D., and BENTLER, P. M. (1987). Changes in drug use from high school to young adulthood: Effects of living arrangement and current life pursuit. J. Appl. Dev. Psychol. 8: 22 l246. PANDINA, R. J., and JOHNSON, V. (1990). Serious alcohol and drug problems among adolescents with a family history of alcoholism. J. Stud. Alcohol 5 I: 278-282. PEDERSEN, W. (1991). Mental health, sensation seeking and drug use patterns: A longitudinal study. Br. J. Addict. 86: 195-204. PETRAITIS, J., FLAY, B. R., and MILLER, T. Q. (I 995). Reviewing theories of adolescent substance use: Organizing pieces in the puzzle. Psycho!. Bull. I 17: 67-86. ROBINS, L. N., and PRZYBECK, T. R. (1985). Age of onset of drug use as a factor in drug and other disorders. In C. L. Jones and R. J. Battjes (Eds.), Etiology of Drug Abuse: Implications fir Prevention (pp. 178-192). Rockville, MD: National institute on Drug Abuse, Research Monograph 56. ROSENTHAL, R. (1984). Meta-analytic Procedures@ Social Research. Beverly Hills, CA: Sage. SCHULENBERG, J., BACHMAN, J. G., O’MALLEY, P. M., and JOHNSTON, L. D. (1994). High school educational success and subsequent substance use: A panel analysis following adolescents into young adulthood. J. Health Sot. Behav. 35: 45-62. SHADISH, W. R. (1996). Meta-analysis and the exploration of causal mediating processes: A primer of examples, methods, and issues. Psychol. Methods 1: 47-65. SHEDLER, J., and BLOCK, J. (1990). Adolescent drug use and psychological health. Am. Psychol. 45: 612-630. SIMCHA-FAGAN, O., GERSTEN, J. C., and LANGER, T. S. (1986). Early precursors and concurrent correlates of patterns of illicit drug use in adolescence. J. Drug Issues 16: 7-28. SIMON& R. L., CONGER, R. D., and WHITBECK, L. B. (1988). A multistage social learning model of the influences of family and peers upon adolescent substance abuse. J. Drag Issues 18: 293315.

2600

PETRAITIS ET AL.

SMITH, G. M., and FOGG, C. P. (1978). Psychological predictors of early use, late use, and nonuse of marijuana among teenage students. In D. 8. Kandel (Ed.), Longitudinal Research on Drug Use: Empirical Findings and Methodological Issues (pp. 101-I 13). Washington, DC: Hemisphere. SMITH, G. M., and FOGG, C. P. (1979). Psychological antecedents of teen-age drug use. Res. Commun. Ment. Health 1: 87-102. STEIN, J. A., NEWCOMB, M. D., and BENTLER, P. M. (1986). The relationship of gender, social conformity, and substance use: A longitudinal study. Bull. Sot. Psychol. Addict. Behav. 5: 125-138. STEIN, J. A., NEWCOMB, M. D., and BENTLER, P. M. (1987a). An 8-year study of multiple influences on drug use and drug use consequences. J. Pers. Sot. Psychol. 53: 109b1105. STEIN, J. A., NEWCOMB, M. D., and BENTLER, P. M. (1987b). Personality and drug use: Reciprocal effects across four years. Pers. Indiv. Dafirences 8: 419-430. TEICHMAN, M., BARNEA, Z., and RAVAV, G. (I 989). Personality and substance use among adolescents: A longitudinal study. Br. .I. Addict. 84: 18 I-190. VANKAMMEN, W. B., and LOEBER, R. (1994). Are fluxuations in delinquent activities related to the onset and offset in juvenile illegal drug use and drug dealing? J. Drug Issues 24: 9-24. VICARY, J. R., and LERNER, J. V. (1983). Longitudinal perspectives on drug use: Analyses from the New York longitudinal study. J. Drug Educ. 13: 275-285. VICARY, J. R., and LERNER, J. V. (1986). Parental attributes and adolescent drug use. J. Adolesc. 9: 115-122. WENG, L., and NEWCOMB, M. D. (1989). Predicting changes in teenage drug use: The role of intention-behavior discrepancy. Genet. Sot. Gen. Psychol. Monogr. 115: 25-48. WHITE, H. R. ( 1992). Early problem behavior and later drug use. J. Res. Crime Delinq. 29: 412429. WHITE, H. R., PANDINA, R. J., and LAGRANGE, R. (1987). Longitudinal predictors of serious substance use and delinquency. Criminology 3: 7 15-740. WILLS, T. A., BAKER, E., and BOTVIN, G. J. (1989). Dimensions of assertiveness: Differential relationships to substance use in early adolescence. J. Consult. Clin. PsychoI. 57: 473-478. WINDEL, M. (1989). Substance use and abuse among adolescent runaways: A four-year follow-up study. J. Youth Adolesc. 18: 33 I-344. WINDEL, M. (1990). A longitudinal study of antisocial behaviors in early adolescence as predictors of late adolescent substance use: Gender and ethnic group differences. J. Abnorm. Psychol. 99: 86-91.

WINEFIELD, A. H., TIGGEMANN, M., WINEFIELD, H. R., and GOLDNEY, R. D. (1993). Growing Up with Unemployment: A Longitudinal Study of Its Psychological Impact. London: Routledge.

RESUMEN Este trabajo repas 10s resultados de 58 estudios longitudinales sobre el uso de sustancias ilicitas (USI) entre 10s adolescentes. ArregM 384 resultados conforme a tres tipos de influencia (a saber, social, de actitud, e intrapersonal) y cuatro niveles de influencia (a saber, lejana, distante, cercana, e inmediata). La mayorla de la evidencia reconfirm6 la importancia de varios factores predictores de US1 (p. ej., intenciones y previo comportamiento relacionado con sustancias, patrones de amistad y comportamientos de @ales, ausencia de padres que presten apoyo, temperament0 sicol6gico), revel6 que algunas variables que se consideraban

ILLICIT SUBSTANCE USE AMONG ADOLESCENTS

2601

factores predictores bien establecidos pueden no serlo (p. ej., el comportamiento de 10s padres, la indulgencia de 10s padres, la depresion, la falta de confianza en si mismo), e identifico varias variables en que 10s resultados eran o escasas o no consistentes (p. ej., el papel de las politicas publicas tocantes al USI, la representation en 10s medios de comunicacion masiva de1 USI, varios estilos de desempefiar las responsabilidades de ser padres, estados afectivos, percepciones de desaprobacion de 10s padres hacia el USI, y las capacidades de rechazo de uso de sustancias especificas). Tambibn se consideran posibles pautas para futuras investigaciones.

Ce memoire examine les resultats de 58 etudes longitudinales de l’usage de substances illicites parmi les adolescents. Les 384 conclusions sont &parties selon trois types d’influence (a savoir sociale, attitudinale et interpersonnelle) et quatre degres d’influence (a savoir ultime, eloignee, proche et immediate). L’ensemble de l’tvidence reaffirme l’importance de plusieurs predispositions a l’usage de substances illicites (c’estMire les intentions et le comportement associes a l’usage prealable de substances, le choix d’amitits et le comportement de semblables, l’absence de soutien parental, le temperament psychologique) et demontre que quelques variables considerees comme predispositions, certaines ne le sont pas (c’est-a-dire l’attitude parentale, l’indulgence parentale, la depression, le manque de confiance en soi) et revtle plusieurs variables ou les result&s ttaient soit.rares soit inconcluants (c’est-A-dire le role des lois qui regissent l’usage de substances illicites, sa divulgation par les medias, la facon d’tlever les enfants, les ttats affectifs, la perception de la d&approbation des parents et l’incapacitt a refuser la consomption de ladite substance). Les domaines dans lesquels la recherche se poursuit sont abordts ici.

SAMENVATTING Onder adolescenten: Een matrix van prospectieve voorspellers. Dit artikel geefi een overzicht van de resultaten van 59 prospectieve studies naar illegaal middelengebruik onder adolescenten. 384 bevindingen, voortkomend uit de resultaten, zijn gerangschikt naar drie typeringen van befnvloeding (namelijk: sociaal, attitude en intrapersoonlijk) en vier niveaus van befnvloeding (namelijk: uiterst, ver, nabij en direct). Een overvloed aan bewijzen bevestigt nogmaals het belang van verscheidene voorspellers van illegaal middelengebruik (zoals: intenties tot middelengebruik en voormalig aan-middelengebruikgerelateerd gedrag, vriendschapspatronen en peer-gedrag, afwezigheid van ondersteuning vanuit de ouders en psychisch temperament). Daamaast bleken sommige variabelen, waarvan dit we1 gedacht werd, geen goede, vastgestelde

2602

PETRAITISETAL.

voorspellers te zijn. (bij voorbeeld: het gedrag van de ouders, de toegestane vrijheid door de ouders, depressie, lage zeltiaardering). Bij verscheidene andere variabelene bleken de bevindingen te schaars of inconsistent te zijn (bij voorbeeld: de rol van het overheidsbeleid met betrekking tot illegaal middelengebruik, beelden over illegaal middelengebruik vanuit de massa-media, bepaalde opvoedingsstijlen, stemmingen, waarnemingen van ouderlijke atkeuring van illegaal middelengebruik en middel-specifieke weigeringstechnieken). In de discussie worden nieuwe terreinen voor toekomstig onderzoek besproken.

ILLICIT SUBSTANCE USE AMONG ADOLESCENTS

2603

THE AUTHORS John Petraitis received his Ph.D. in Applied Social Psychology from Loyola University of Chicago. After working as a Senior Research Associate at the Prevention Research Center at the University of Illinois at Chicago, he accepted his current position as an Associate Professor of Psychology at the University of Alaska Anchorage. He has published papers and chapters on the use of tobacco, alcohol, and other substances among adolescents, and he is currently working on grants to study the etiology of substance use among adolescents. Brian R. Flay received his D.Phil. in Social Psychology from Waikato University in New Zealand in 1976. After receiving postdoctoral training at Northwestern University (Evanston, Illinois) under a Fulbright-Hays Fellowship, he started research on smoking prevention at the University of Waterloo (Ontario, Canada). He continued this work and developed work in the areas of drug use prevention and the use of mass media for smoking cessation at the University of Southern Cali-

2604

PETRAITIS ET AL.

forma. Since 1987, Dr. Flay has been the Director of the Prevention Research Center in the School of Public Health, University of Illinois at Chicago, where he continues research in the above areas as well as AIDS and violence prevention, theories of adolescent behavior development and change, evaluation research methods, and community prevention of high risk social and health behaviors including violence. Todd Q. Miller is an Associate Professor of Medicine in the Department of Preventive Medicine and Community Health at the University of Texas Medical Branch. He has received a grant from the American Lung Association and currently has a grant from the National Institute on Drug Abuse that examines how state-of-the-art statistics can be applied to longitudinal studies that seek to understand the social psychology etiology of adolescent substance use. Edward J. Torpy currently works at SPSS, Inc. as a technical marketing specialist. Previously he worked as a Research Specialist at the Prevention Research Center at the University of Illinois at Chicago where he participated in writing literature reviews and analyzing several large data sets. He is currently pursuing a Ph.D. in Applied Social Psychology at Loyola University of Chicago. His research interests include the study of attitude structures and values hierarchies. Brenda Greiner received her M.A. in Clinical Psychology from DePaul University (Chicago, Illinois) in 1990 where she also served as a Research Assistant in the Department of Psychology. From 1990 to 1995 she worked as a Research Specialist at the Prevention Research Center at the University of Illinois at Chicago.