a 18 year old model of family and peer effects on the drug use

P1: Vendor/FOM
Journal of Youth and Adolescence

pp334-joyo-364605

December 28, 2001

12:11

Style file version July 26, 1999

Journal of Youth and Adolescence, Vol. 31, No. 1, February 2002, pp. 45–56 ( C 2002)

An 18-Year Model of Family and Peer Effects on Adolescent Drug Use and Delinquency
Helen E. Garnier1 and Judith A. Stein2
Received November 27, 2000; accepted June 26, 2001

The interrelationship of family and peer experiences in predicting adolescent problem behaviors was examined in an 18-year longitudinal sample of adolescents (N = 198) from conventional and nonconventional families. Positive associations among early childhood predictors and adolescent problem behaviors were consistent with problem behavior theory. The most powerful predictors of teen drug use and delinquent behaviors were similar behaviors by peers. Peer behaviors, however, were in turn predicted by earlier family-related variables and the quality of peer relationships in childhood.
This study provides supporting evidence that strong peer effects in adolescence reflect even earlier processes in childhood and highlight the importance of linkages from early childhood experiences in family and peer contexts to the development of problem behaviors in adolescence. Implications for prevention and intervention programs are discussed.
KEY WORDS: adolescent problem behaviors; longitudinal structural equation model; peer effects; family effects.

INTRODUCTION

tiple antecedents in childhood and adolescence (Goff and
Goddard, 1999; Jessor et al., 1995). Although youth violence declined slightly in the late 1990s, adolescents are frequently exposed to delinquency, violent crimes, drugrelated crimes, and crimes against persons and property
(U.S. Bureau of the Census, 1999) and drug use (Johnston et al., 2000; National Institute of Drug Abuse (NIDA),
1995). For example, 55% of adolescents have used an illicit drug, 65% have tried cigarettes, and 80% have tried alcohol by the time they leave high school.
Because teens spend more time with peers and less with families during their transition into adulthood, peers have the most important influence on their day-to-day behaviors (Steinberg et al., 1992). Both parents and peers, however, have been found to contribute to adolescent development but in different ways (Dryfoos, 1990; Laible et al., 2000; Stein et al., 1987). Cairns and Cairns (1995) propose that predominant peer effects actually reflect earlier processes in childhood, and stress the importance of processes linking different social organizations across time. Unfortunately, the narrow time frame of many studies comparing parent and peer antecedents of teen behaviors has hampered investigation of the process by which longer-term family effects and proximal peer effects

Current research on high-risk adolescent behaviors more often focuses on groups of problem behaviors rather than on single risk factors in an attempt to discover mulThe Family Lifestyles Project has been supported by Carnegie Corporation Grant # B3970, by USPHS Grant # MH24947 to the late Dr Bernice
Eiduson, Principal Investigator, and by Carnegie Corporation Grant
# B4189 and W. T. Grant Foundation Grant # 92-1488 to Dr Thomas
Weisner, Principal Investigator. The National Institute on Drug Abuse has supported Dr Helen Garnier, Grant # DA05449-02, and Dr Judith
Stein, Grant # DA10170-28. The Department of Psychiatry and Biobehavioral Sciences, University of California, Los Angeles, also has provided important support to the study. Most of all, we thank the more than 200 families participating in the Family Lifestyles study.
1 Research Statistician, Department of Psychiatry and Behavioral
Sciences, University of California, Los Angeles. Received PhD in
Education, University of California, Los Angeles. Interests are adolescent development and related problem behaviors. To whom correspondence should be addressed at Department of Psychiatry and
Behavioral Sciences NPI C9-740, University of California, Los Angeles
90095.
2 Research Psychologist, Department of Psychology, University of
California, Los Angeles. Received PhD in Psychology with a specialization in Measurement and Psychometrics. Currently collaborates with researchers, using structural equation modeling to investigate problem behaviors among at-risk populations.

45
0047-2891/02/0200-0045/0

C

2002 Plenum Publishing Corporation

P1: Vendor/FOM
Journal of Youth and Adolescence

pp334-joyo-364605

December 28, 2001

12:11

Style file version July 26, 1999

46 predict adolescent behaviors. More recent studies have longitudinal rather than cross-sectional data and employ techniques such as structural equation modeling to estimate the contribution of parental and peer pathways to problem behaviors simultaneously (e.g., Blanton et al.,
1997). Yet few studies have followed teens from very early childhood through adolescence.
This study examines the relationships in the early family environment and child and adolescent peer group experiences in the prediction of adolescent drug use and delinquent behaviors. Because nonconventional values have been linked to high-risk behaviors, the role of values in the development of adolescent problems is a central issue explored in the 18-year longitudinal prospective Family Lifestyles (FLS) Project sample of conventional and nonconventional families (e.g., Garnier and Stein, 1998).
Although research has associated nonconventional values with problem behaviors, few studies have explored the protective role of values in the context of families and peer affiliations. The longitudinal nature of this study permits use of structural equation modeling to estimate and control simultaneously the associations between parent and peer antecedents and assess the unique effects of families and peers on adolescent problem behaviors.
Furthermore, children who often are lost in school-based longitudinal studies because they are involved in highrisk behaviors such as heavy drug use and school failure have been retained in this sample (Garnier et al.,
1997).

Problem Behavior Theories
Both theory and research explain the relationships among adolescent high-risk behaviors by the existence of a single “problem behavior syndrome” which is generalizable to different gender, socioeconomic, and ethnic groups
(Elliott et al., 1989; Jessor and Jessor, 1977). Problem behavior theory asserts that nonconventionality in values, the personality, and the social environment is related to multiple problem behaviors (Donovan et al., 1991). Adolescent identification with nonconventional values, for example, predicts greater involvement in drug use behaviors
(Baumrind, 1985; Elliott et al., 1985; Stein et al., 1987) and delinquent behaviors (Hawkins and Weis, 1985). Possible antecedents to these behaviors can be found in the family context where children are exposed to nonconventional values and attitudes, open modeling of parental drug abuse, and extent of primary caretaker’s drug use (Block et al., 1988; Brook et al., 1985; Kumpfer and DeMarsh,
1986), and in the peer context where children acquire the attitudes and behaviors of peers (Harris, 1998).

Garnier and Stein
Family Environment
Unconventional values in families are a key factor in the development of problem behaviors. Parental unconventional values are related not just to their own behaviors and associated risks, but also to the transmission of these values to their children (Garnier and Stein,
1998). Children form their worldview through the values and meanings within the context of the family socializing environment (Berg, 1985). Empirical findings provide supporting evidence of intergenerational transmission of values (Whitbeck and Gecas, 1988). Social learning theories propose that children acquire their values and behaviors from observation and imitation of role models, social reinforcement, and positive expectations of future behaviors (Bandura, 1986). The more important the role models, such as parents, the more likely children will be to imitate their behaviors and internalize values, attitudes, and beliefs. Home environments where unconventional behaviors are socially accepted provide children with models and opportunities to engage in problem behaviors. Family drug use, for instance, has been linked consistently to adolescent drug use (Newcomb and Bentler, 1988; Stein et al.,
1987). Initiation into drug use is related to parental drug use and attitudes about drugs (Baumrind, 1985; Kandel and Yamaguchi, 1987). By providing models for drug use, access to drugs, and motivation for drug use, parents may effectively establish drug use behaviors, reduce children’s internal restraints to drug use, and, through positive attitudes toward drug use, encourage adolescents to seek affiliations with peers who similary use drugs (Duncan and Petosa, 1995).
Parent behaviors and attitudes also have been found to have differential effects on children. It is frequently assumed that fathers have less of an influence on children’s development because they spend less time with their children (Coltrane, 1995). Although fathers have been found to make unique contributions to their children’s outcomes
(Phares and Compas, 1992) including alcohol and illicit drug use (Brook et al., 1985), fathers’ drug use is more weakly associated with children’s behaviors than does mothers’ drug use (Brook et al., 1995). Children may be more directly affected by a mother who is a drug user because she is the main caretaker (Kumpfer and DeMarsh,
1986).
Benefits for parents and children, on the other hand, are related also to certain nonconventional values. For example, the basis of many shared values and beliefs of counterculture movements was a commitment to promoting and protecting a sense of personal worth and meaning in life (Johnston et al., 1994). Parents committed to such values were less likely to promote values that encourage

P1: Vendor/FOM
Journal of Youth and Adolescence

pp334-joyo-364605

December 28, 2001

12:11

Style file version July 26, 1999

Family and Peer Effects deviant behavior in their children and more likely to foster independence and creativity in their children (Garnier and Stein, 1998). Although related to risk-taking behaviors, these values can predict positive health behaviors
(Chassin et al., 1989). Chassin proposes that two kinds of deviance explain the cooccurrence of such different behaviors. “Constructive deviance” is distinguished from
“destructive deviance” as a less negative risk factor associated with independence rather than anticonformity and antisociality. Single parenthood and economic uncertainty are other risk factors in nonconventional families (Eiduson,
1975) that have been linked to both adolescent drug use and delinquency (Elliott and Huizinga, 1983). Although extensive research has associated socioeconomic status
(SES) with children’s development through such factors as parent values, socialization practices, school achievement, and neighborhoods (Elder et al., 1984; Elliott et al.,
1985), other studies have found weak effects of SES on such children’s problem behaviors (Coombs et al., 1986).
Peer Experiences
Peers are an important part of the complex socialization process as teens cope with the developmental tasks of adolescence (Brown, 1990). According to current theories such as Harris’ group socialization theory (Harris, 1998), children acquire the behaviors and attitudes of their peers through their experiences with peers, and then carry these learned behaviors and attitudes into adulthood. Unlike peer relationships in childhood, interactions with friends in adolescence become more frequent and time consuming, and are more autonomous. Similarities between friends may be partly the result of selection of friends who are already similar (Cairns and Cairns, 1995). In other words, the nature of peer group affiliations at adolescence appears to be a complex mix of mutual influences. Teens pick peers somewhat like themselves and also are picked by other teens for attributes they bring from childhood from their family value system, social status, and ethnic background. Peer affiliation also appears to reflect adolescent bonds to family and school (Elliott et al., 1985). Adolescents with close parental relationships tend to associate with friends whose values are similar to those of their parents (Elder, 1980), thus reinforcing parental values. Peers also may reinforce behaviors. For example, greatest drug use occurs when both parents and peers use drugs (Kandel and Wu, 1995).
Similar to parents, peers may have both positive and negative effects on teens. Friends may encourage prosocial behavior as well as academic achievement and aspirations

47
(Stein and Newcomb, 1999). In contrast, negative peer behavior is a “risk marker” for multiple high risk behaviors
(see Dryfoos, 1990). Early delinquency and association with delinquent peers are two of the strongest potential antecedents of adolescent delinquent behavior. Peers who use substances and/or engage in delinquent behaviors may encourage similar behaviors in their friends (Conger and
Reuter, 1996).
Family Lifestyles Study
The Family Lifestyles (FLS) Project is an 18-year longitudinal, prospective study of conventional and nonconventional families and their children. Many nonconventional parents adopted countercultural values stemming from their identification with 1960’s countercultural movements and experimented with their lifestyles. Counterculture lifestyles often exposed children to multiple risks, including the transmission of parents’ nonconventional values that often were associated with positive attitudes toward experimentation and drug use. An interesting feature of the nonconventional families in this study is the role of values in fostering resilience in families and children. Families with a strong commitment to countercultural values made proactive efforts to change their lives and environment (Weisner, 1987) which may have reduced the likelihood of children’s developmental problems throughout elementary and high school (Garnier et al., 1997);
Weisner and Garnier, 1992).
This body of research is extended by including peer predictors simultaneously with family predictors in a longitudinal model predicting adolescent drug use and delinquency. Including peer behaviors in adolescence enables comparison of the unique effects of family and peer antecedents on problem behaviors. Including the quality of peer interactions at the beginning of elementary school enables us to compare family and peer effects early in a child’s development on adolescent behaviors, directly and indirectly through later peer factors. Finally, including longitudinal measures of maternal drug use at 6 and 18 years allows assessment of long-term and concurrent effects of maternal drug use and disentangles the effects of maternal drug use from time of assessment of family variables.
Although child gender has been associated with problem behaviors, such as boys’ greater involvement in substance use (Stein et al., 1993), prior studies failed to find significant gender differences in this sample in drug use and delinquency. Girls were as likely as boys to engage in drug use and delinquency. In addition, relationships between most individual measures used in this study were similar for girls and boys. Gender, therefore, was not included in our longitudinal model.

P1: Vendor/FOM
Journal of Youth and Adolescence

pp334-joyo-364605

December 28, 2001

12:11

Style file version July 26, 1999

48

Garnier and Stein

Hypotheses
The goal of this study is to explore relationships among variables in the family environment and peer behaviors on adolescent drug use and delinquent behaviors.
The following are our hypotheses:
(1) Adolescent problem behaviors would be associated with both the family background and peer behaviors but peer behaviors at adolescence would have stronger associations.
(2) Both early peer interactions and family background, such as lower SES and maternal drug use, would indirectly predict adolescent problem behaviors through similar peer behaviors.
(3) Values socialization in families would differentially predict adolescent behaviors indirectly through teen values and through peer behaviors.
Less conventional parental values would indirectly predict teens’ greater involvement in drug use and delinquent behaviors through similar teen values and through association with friends who engage in similar behaviors.

Frequent contacts and strict confidentiality contributed to
98% sample retention over 18 years. Extra resources were allocated to maintain contact with about 15% of the sample who were more difficult to locate. Comparisons of families lost to follow-up with those who were retained indicated that the more troubled, drug-using families were not lost.
A small amount of missing data at the 18-year followup resulted in a final sample size of 198 for this current study. Changes in the Original Selection Categories
The nonconventional family arrangements were used for sample selection to identify parents who were experimenting with their lifestyles in the 1970s. Generally, families became more stable over time with nonconventional families changing their family organization more often than did the conventional families. At 18 years, 39% of the original single mothers, 36% of the social contract couples, 14% of those living in communes, and 73% of the conventionally married families were still in their original family arrangement.

METHOD
Measures
Participants
The FLS has followed a sample of 205 families since
1974–75: 47 single mothers, 53 social contract couples
(nonlegally married), 54 in communes, and a conventional comparison sample of 51 two-parent married couples
(Eiduson, 1975). These families were selected through a random sample of obstetricians in major urban areas of
California. The sample represents a range of working to upper–middle class mostly Euro-American families. All parents were between the 20th and 90th percentile on a combined socioeconomic and educational scale when selected. The average age of entry into the study for the mothers was 25 years; fathers were slightly older with an average age of 28. Mothers had completed an average of
14 years of education, and fathers 16 years by the time the child was 6. Seventy-five percent were about to have their first child. Forty-seven percent of the sample children are girls, and 53% are boys.
Sample Retention
Extensive interviews, questionnaires, home observations, staff ratings, and clinical assessments were used to collect 15 waves of data on families and their children.

Face-to-face interviews and questionnaires beginning at the last trimester of pregnancy through 18 years were used to obtain data on family background, values, and drug use from mothers. Questionnaires were used at 18 years to obtain adolescent reports of their values, behaviors, and perceived peer behaviors. Teachers rated children’s competencies in Grades 1 and 2.
Maternal Drug Use
Mothers identified the extent of their illicit drug use in interviews at 6 years and in questionnaires and interviews at 18 years. Latent variables indicated maternal drug use at 6 years and at 18 years. Drug use indicators at
6 and 18 years include marijuana, cocaine, and downers
(coded 0 – no use to 5 – at least once a week). The latent constructs are labeled Maternal Drug Use. Comparison of substance use among FLS mothers with adults in similar age groups from the National Household Survey on
Drug Abuse (Garnier and Weisner, 1992; NIDA, 1995) indicate maternal substance use in this sample reflects a similar pattern found in national surveys. Marijuana was the most frequent substance used, and drug use declined with age.

P1: Vendor/FOM
Journal of Youth and Adolescence

pp334-joyo-364605

December 28, 2001

12:11

Style file version July 26, 1999

Family and Peer Effects

49

Maternal and Adolescent Values

Adolescent Drug Use

Mothers were assessed at the trimester of their pregnancy on 8 values, and the teens were assessed at 18 years on the same 8 values3 (Weisner and Rochford, 1980):
(a) alternative achievement goals such as self-fulfillment
(10 items, α = 0.89), (b) pronaturalism (7 items, α =
0.91), (c) humanism (9 items, α = 0.86), (d) immediate gratification (6 items, α = 0.90), (e) antiauthority (7 items, α = 0.92), (f) acceptance of alternative sources of knowledge such as intuitive sources (6 items, α = 0.89), (g) sex egalitarianism (8 items, α = 0.84), and (h) nonmaterialism (6 items, α = 0.91). Two sums were created
(Garnier and Stein, 1998): “traditional/achievement values” and “humanitarian/egalitarian values.” Included in the traditional/achievement values (scored in the conventional direction) were conventional achievement, future orientation, conventional authority, and acceptance of rational, scientific knowledge. Included in humanitarian/ egalitarian values were pronaturalism, sex egalitarianism, and lack of materialism.

Teens also described their own drug use from items in a questionnaire designed for adolescents to obtain measures of drug use, delinquent behaviors, and perceived peer behaviors (Newcomb and Bentler, 1988). Teen reports were used to create summary scores measuring alcohol, marijuana, and hard drug use. Type and frequency of substance use in the last 6 months (coded 1 – never to 7 – more than once a day) were summed separately for alcohol (3 items), cannabis (2 items), and hard drugs other than marijuana (18 items).

Family SES

Adolescent Delinquency

Family SES was measured at the beginning of the study using Hollingshead’s scale combining mother’s and father’s education and occupation (Hollingshead, 1975).
SES was coded on a scale from 1 (low SES) to 5 (high
SES).

Adolescents also reported how many times they had been involved in deviant activities in the last 6 months
(coded 0 – none to 6 – six or more). Sums were created for aggressive behaviors such as “gotten into a serious fight at school or work” (5 items, α = 0.85), criminal behaviors such as “carried a deadly weapon” (5 items, α = 0.60), and crimes involving theft or burglary such as “taken something from a store without paying for it”
(4 items, α = 0.70). The latent construct is labeled Teen
Delinquency.

Child Social Competence
Teachers rated children in Grades 1 and 2 with the use of two rating scales: The Pupil Behavior Rating Scale
(Lambert et al., 1979) and The Classroom Behavior Inventory (Schaefer et al., 1978). Ratings included peer interactions, classroom adaptation, interpersonal behavior, consideration, independence, dependence, introversion, and extroversion, α (Grade 1) = 0.75; α (Grade 2) = 0.78.
Principal component analyses were used to create the two measures of social competence. The latent construct is labeled Child Social Competency.
3 The following are examples of items used to measure value orientations:

(a) To succeed in what I think is important is one of my strongest desires
(Alternative achievement goals). (b) I see myself as a “natural person,” in tune with the environment rather than exploiting it (Pronaturalism).
(c) If accomplishing a task or goal requires a long-term commitment,
I generally won’t consider it (Immediate gratification). (d) Despite its faults I still believe it important to respect the authority of the government (Antiauthority). (e) Women should become actively involved in politics and community problems as well as in their families (Gender egalitarianism). (f) I make a conscious effort to limit my accumulation of possessions (Nonmaterialism).

Peer Drug Use
Adolescents reported how many of their friends were involved in substance use behaviors (coded 1 – none to 5 – all). Three measures were created (alcohol use, marijuana use, and hard drug use) and used to measure the latent construct labeled Peer Drug Use.

Peer Delinquency
Teens also described how many of their friends were engaged in delinquent behaviors (coded 1 – none to 5 – all). Summary scores measured aggressive behaviors such as “get into fights” (3 items, α = 0.75), criminal behaviors such as “carried a deadly weapon” (5 items, α = 0.78); and crimes involving thefts such as “been busted for burglary or robbery” (4 items, α = 0.65).
Analysis
The proposed longitudinal predictive path model was tested with latent variable structural equation models; initially, confirmatory factor analysis was performed to obtain correlations among the latent and measured constructs without any imputation of precedence (Bentler,

P1: Vendor/FOM
Journal of Youth and Adolescence

pp334-joyo-364605

December 28, 2001

12:11

Style file version July 26, 1999

50

Garnier and Stein

1995). Fit was assessed with chi-square values, the Comparative Fit Index (CFI) (Bentler, 1990), the Robust CFI
(RCFI) (Bentler and Dudgeon, 1996), and the root mean square error of approximation (Hu and Bentler, 1999).
Satorra–Bentler chi-square and RCFI values are reported because the Satorra–Bentler chi-square and RCFI are preferable when data are multivariately kurtose (Bentler and
Dudgeon, 1996). Indirect effects were assessed before nonsignificant direct paths were dropped from the model.
Correlated error residuals were allowed between similar variables measured at two time points. A few additional theoretically plausible correlated error residuals were allowed based on the Lagrange Multiplier Test (LM test)
(Chou and Bentler, 1990). The hypothetical model reflects a sequential, longitudinal timeline in which early family background determinants predict Maternal Drug Use and
Child Social Competency at 6 years. These variables predict 18-year variables (Maternal Drug Use, Peer Drug Use,
Peer Delinquency, and Teen Traditional/Achievement Values and Humanitarian/Egalitarian Values) which then predict Teen Delinquency and Teen Drug Use.

the adjusted standard errors. The final path model fit quite well: (S–B) χ 2 = 310.05, 252 df, RCFI = 0.94,
RMSEA = 0.034. Indirect effects are presented in
Table III. To obtain accurate estimates of indirect effects, the direct paths were included temporarily even if not significant in the final model. Because the sample size was small for the number of parameters estimated, a bootstrap procedure tested the stability of the final parameter estimates (Chan et al., 1998). One hundred separate samples were randomly derived from the sample with replacement, and the final model was run 100 times using each derived sample. Ninety-five percent confidence intervals were computed for all parameters in the final path model after the parameters obtained from the 100 analyses were averaged across the randomly selected samples. All parameter estimates in Fig. 1 fell within the 95% confidence intervals for the averaged parameter estimates, and the nearly identical or identical to the averaged estimates from the bootstrap analysis.

RESULTS

Higher SES was positively associated with traditional/achievement values and negatively associated with humanitarian/egalitarian values. Maternal traditional/ achievement and humanitarian/egalitarian values were negatively associated.

Confirmatory Factor Analysis
The maximum likelihood (ML) fit statistics of the initial confirmatory factor analysis (CFA) model were close to acceptable values, ML χ 2 (213, N = 198) = 398.57,
CFI = 0.88, RMSEA = 0.067. Because the data were multivariately kurtose (normalized estimate = 55.13), the more preferable and appropriate Satorra–Bentler (S–B) χ 2 statistic was examined and was considerably better than the maximum likelihood estimator, S–B χ 2 = (213,
N = 198) = 306.17, RCFI = 0.90, RMSEA = 0.047.
After adding two reasonable correlated error residuals to the CFA, the fit indexes improved markedly (S–B χ 2 (211,
N = 198) = 273.96, RCFI = 0.93, RMSEA = 0.039).
Additional correlated error residuals were between the following variables: Teen alcohol use and peer alcohol use, teen aggression and peer aggression. Table I reports the summary statistics and factor loadings of the measured variables on their latent variables. Table II reports the correlations among all of the latent and measured variables.
Path Model
The final path model, after all nonsignificant predictive paths and correlations were deleted, is depicted in
Fig. 1. The standard errors of the regression paths were adjusted for the extent of the nonnormality (Bentler and
Dijkstra, 1985), and significance levels are based on

Correlations Among Background Predictors

Prediction of Maternal Drug Use
Maternal Drug Use at 6 years was negatively predicted by maternal traditional/achievement values and positively predicted by maternal humanitarian/egalitarian values. Maternal Drug Use at 18 years was predicted by
Maternal Drug Use at 6 years and lower Family SES. Family SES did not predict Maternal Drug Use at 6 years.
Prediction of Child Social Competency
Child Social Competency was not significantly predicted by any of the early family background measures.
Prediction and Correlates of Teen Values and Peer Behaviors
Maternal values assessed at the third trimester of pregnancy were significant predictors of analogous teen values assessed at 18 years. Lower Family SES predicted higher Peer Delinquency. Maternal Drug Use at 6 years predicted greater Peer Drug Use. Child Social Competence positively predicted Teen Traditional/Achievement Values and negatively predicted Peer Delinquency. There were

P1: Vendor/FOM
Journal of Youth and Adolescence

pp334-joyo-364605

December 28, 2001

12:11

Style file version July 26, 1999

Family and Peer Effects

51

Table I. Summary Statistics of Measured Variables, and Factor Loadings from the Confirmatory Factor Analysis
Variable
Background
Maternal Traditional/Achievement Values
Maternal Humanitarian/Egalitarian Values
Family SES
6-Year Predictors
Maternal Drug Use
Marijuana
Cocaine
Downers
Child Social Competencyc
Year 6
Year 7
18-Year Predictors
Maternal Drug Use
Downers/uppers
Hard drugs
Marijuana/LSD
Peer Drug Use
Friends drink
Friends use marijuana
Friends use other drugs
Peer Delinquency
Thefts
Aggression
Criminal behavior
Teen Traditional/Achievement Values
Teen Humanitarian/Egalitarian Values
18-Year Outcomes
Teen Drug Use
Drinking
Marijuana
Other drugs
Teen Delinquency
Thefts
Aggression
Criminal behavior

Factor loadinga

Mean (SD)

Skewness

Kurtosis

10.87 (0.63)
8.68 (0.75)
3.38 (1.27)

−0.01
−0.08
−0.68

−0.41
−0.74
0.08

NAb
NA
NA

0.84 (1.32)
0.30 (0.77)
0.18 (0.62)

1.42
3.14
4.24

0.63
10.36
19.35

0.52
0.57
0.38

0.03 (0.96)
0.00 (0.98)

−0.50
−0.83

−0.13
−0.18

0.53
0.68

0.08 (0.80)
0.08 (0.43)
0.45 (1.02)

5.25
6.34
2.44

28.94
44.17
5.22

0.62
0.73
0.67

2.92 (1.22)
2.20 (1.25)
1.55 (0.86)

0.22
0.79
1.87

−1.07
−0.54
3.56

0.59
0.88
0.77

3.12 (1.14)
3.12 (1.37)
4.45 (1.55)
14.70 (1.09)
13.92 (1.43)

1.07
1.74
1.31
−0.20
−0.11

1.23
4.32
1.37
0.80
0.14

0.74
0.64
0.81
NA
NA

6.87 (3.47)
3.50 (2.26)
19.34 (3.37)

0.55
1.74
4.35

−0.69
2.79
21.83

0.45
0.91
0.64

1.87 (3.58)
4.78 (2.69)
1.18 (2.27)

2.58
0.50
2.95

6.64
1.35
10.94

0.60
0.33
0.67

factor loadings significant, p ≤ 0.001.
– Not applicable. c Competency scores standardized.

a All

b NA

significant positive correlations between Peer Drug Use and Peer Delinquency residuals, negative correlations between Teen Traditional/Achievement Values and Peer
Drug Use and Delinquency residuals, and a modest but significant correlation between Peer Drug Use and Teen
Humanitarian/Egalitarian Values residuals.
Prediction of Teen Delinquency
The large path coefficient between Peer Delinquency and Teen Delinquency demonstrates the powerful and significant relationship of peer behaviors to teen problem behaviors. Higher Teen Traditional/Achievement Values and
Humanitarian/Egalitarian Values significantly predicted

less Teen Delinquency but the contribution was relatively smaller. This model explained 71% of the variance in the
Teen Delinquency latent variable. There was a significant negative indirect effect of Family SES ( p < 0.01) on Teen
Delinquency mediated through Peer Delinquency and a significant negative indirect effect of Child Social Competence ( p < 0.05) on Teen Delinquency mediated through both Peer Delinquency and Teen Traditional/Achievement
Values.
Prediction of Teen Drug Use
Teen Drug Use was strongly associated with Peer
Drug Use. This model explained 76% of the variance

P1: Vendor/FOM
Journal of Youth and Adolescence

pp334-joyo-364605

December 28, 2001

12:11

Style file version July 26, 1999

52

Garnier and Stein
Table II. Correlations Among Measured and Latent Variables
Variables

I. Maternal Traditional/
Achievement Values
II. Maternal Humanitarian/
Egalitarian Values
III. Family SES
IV. Maternal Drug Use—
6 years
V. Child Social
Competency
VI. Maternal Drug Use—
18 years
VII. Peer Drug Use
VIII. Teen Traditional/
Achievement Values
IX. Teen Humanitarian/
Egalitarian Values
X. Peer Delinquency
XI. Teen Drug Use
XII. Teen Delinquency
∗p

I

II

III

IV

V

VI

VII

VIII

IX

X

XI

XII

—
−0.43∗∗∗

—

0.23∗∗∗ −0.15∗
—
−0.29∗∗∗ 0.33∗∗∗ −0.20∗
−0.15∗

0.16∗

0.11

−0.19∗

0.17∗

−0.20∗∗

−0.11
0.16∗

0.14∗
0.00

−0.10
−0.07

−0.11

0.22∗∗∗ −0.07

−0.12
−0.12
−0.09

0.06
0.13
0.06

—
−0.16

—

0.52∗∗∗ −0.21∗

—

0.24∗∗ −0.27∗∗ 0.24∗∗
—
−0.13
0.18∗ −0.17∗ −0.35∗∗∗
0.14

−0.21∗∗ −0.17
−0.06
0.22∗
−0.03
0.20∗

0.16∗
−0.25∗∗
−0.21∗
−0.23∗

—

0.08

0.15∗

0.28∗∗
0.25∗∗
0.26∗∗

0.86∗∗∗ −0.45∗∗∗ 0.04
—
0.88∗∗∗ −0.30∗∗∗ 0.19∗∗ 0.66∗∗∗
—
0.69∗∗∗ −0.46∗∗∗ −0.12 0.83∗∗∗ 0.64∗∗∗ —

−0.06

—

≤ 0.05; ∗∗ p ≤ 0.01; ∗∗∗ p ≤ 0.001.

Fig. 1. Significant paths among latent and measured variables in the structural equation model predicting teen drug use and delinquency.
Latent variables are circles; measured variables are rectangles. Path coefficients are standardized. Small circles represent error variances.
Robust standard errors were used to determine the significance of the standardized paths (a = p < 0.05, b = p < 0.01, c = p < 0.001).

P1: Vendor/FOM
Journal of Youth and Adolescence

pp334-joyo-364605

December 28, 2001

12:11

Style file version July 26, 1999

Family and Peer Effects

53

Table III. Indirect Effects of Latent and Measured Variables on Teen
Outcomes With Direct Effects Controlled
Indirect effects on
Variable
Maternal Conventional Values
Nonconventional Maternal Values
Family SES
Maternal Drug Use—6 years
Child Social Competence—6 years
∗p

Teen
Teen
Drug Use Delinquency
−0.04
0.05
—
0.14∗
−0.22∗

−0.01
−0.03
−0.15∗∗
—
−0.20∗

≤ 0.05; ∗∗ p ≤ 0.01.

in the Teen Drug Use latent variable. Also, there was a significant positive indirect effect of Maternal Drug
Use at 6 years on Teen Drug Use ( p < 0.05) and a significant negative indirect effect of Child Social Competence on Teen Drug Use ( p < 0.05). The direct effects of both of these predictors were mediated through Peer
Drug Use.
DISCUSSION
As teens cope with the developmental tasks of adolescence, their families and peer groups present them with both opportunities and risks. This study demonstrates the simultaneous effects of both family and peer domains on problem behaviors and the relationships between the early family environment and peer experiences that predict adolescent problem behaviors. Childhood and adolescent predictors were significantly associated with adolescent problem behaviors in the directions expected by problem behavior theory. However, childhood predictors impacted peer relationships and value systems that were concurrent with and, yet, predictive of teen outcomes.
Peer Group Experiences
Consistent with the body of research on adolescent problem behaviors, peer behaviors were by far the strongest predictors of adolescent problem behaviors. Drug use and delinquency are typically social behaviors, and the adolescents in this study tended to become involved in the same behaviors as their friends. Furthermore, their involvement in one type of behavior significantly increased the likelihood of involvement in other problem behaviors. Teens who engage in problem behaviors generally are inclined to associate with other teens who are involved in similar behaviors (Cairns et al., 1995; Cairns and
Cairns, 1995). Peer pressure may have encouraged teens to engage in these behaviors, although social selection

and socialization rather than a unidirectional effect of peer pressure could help explain complex linkages among peers, values, and behaviors (Reed and Rountree, 1997).
Support was found for reciprocal effects of peer relationships in this study. Less social competency in childhood was associated with later involvement with more deviant peers. Early Family Environment
Identifying the link between maternal values and adolescent problem behaviors 18 years later is an important contribution of this study. The effects of values begin in childhood; they are associated with choices adolescents make in acquiring their own values and developing social skills needed to assume adult roles. Adolescents in this sample reported values similar to those of their mothers along 2 dimensions: traditional achievement and authority values and a concern for individual status in society, and an orientation toward more humanistic/egalitarian values associated with interpersonal or social needs. A stronger commitment to values of either type predicted fewer delinquent behaviors. Encouraging a respect for self and others may have reduced tendencies toward aggressive behaviors or associating with those who might engage in aggressive behaviors. Teen’s humanistic/egalitarian value system, though more strongly associated with teen drug use bivariately
(r = 0.19), did not predict more drug use when the model included peer drug use. The different associations between values and drug use and delinquency, however, deserve comment. Chassin et al.’s conclusions that two types of deviance, “constructive” versus “destructive,” are independent predictors of adolescent health behaviors and are both possible pathways to negative and positive health behaviors are consistent with the findings of this study
(Chassin et al., 1989). Constructive deviance, although a risk factor for independence and sensation seeking, is not a high risk factor for multiple problem behaviors and is, in fact, unrelated to tolerance for deviant behavior. Moreover, teens high in constructive deviance tend to engage in more health-protective behaviors. Most likely, the values underlying constructive and destructive deviance are similar to the domains of humanistic/egalitarian and lower traditional achievement/authority values used in this study.
The efforts of many nonconventional families to implement their values into their daily lives may have facilitated the transmission of parental values (Garnier and
Stein, 1998). The significant but relatively small magnitude of associations between maternal and teen values, though, indicates that additional factors mediated the

P1: Vendor/FOM
Journal of Youth and Adolescence

pp334-joyo-364605

December 28, 2001

12:11

Style file version July 26, 1999

54 transmission and formation of values. Possible factors include influences of societal institutions and the media, change in parent values, the child’s perception of parental values, quality of parent–child interactions, and peer influences (Grusec and Goodnow, 1994; Whitbeck and Gecas,
1988). Teens without a close parental relationship are more likely to associate with friends who have different values from their parents (Elder, 1980). If these friends are involved in problem behaviors, teens may be more likely to develop and model values consistent with those problem behaviors and engage in those behaviors as well.
Early maternal drug use was another risk factor indirectly linked with adolescent drug use through peer drug use. Early maternal drug use most likely indicates chronic, heavier drug use signaling a more deviant lifestyle to which children were exposed and which could increase their exposure and attraction to deviant peers. Early family SES also was indirectly linked to adolescent problem behaviors through teens’ associations with deviant peers.
To a large degree, family SES determines the demographic makeup of the school and community environments within which adolescents live, and helps to define the peer groups to which adolescents are exposed. This does not mean that families remained in one specific geographic location for
18 years, but rather that family SES generally remained stable for these families.

Limitations
One of the limitations of this study is the use of adolescent reports of their peers’ behaviors rather than independent peer reports. Future studies would benefit by including independent peer reports of behaviors. Another limitation is the exclusion of fathers’ values. Only maternal values were included in this study because children generally spend more time with and have more exposure to mothers who are the main caretakers (Kumpfer and
DeMarsh, 1986). In addition, only 72 fathers provided longitudinal data from the trimester to 18 years, which was inadequate for structural models. However, maternal and paternal values are strongly correlated among parents who both participated in this study (trimester canonical correlation = 0.85, p < 0.001; 18-year canonical correlation = 0.74, p < 0.001) (Garnier and Stein, 1998), which may not hold for cases in which fathers cannot be included. A fairly large number of parameters are estimated despite a rather small sample size. However, the bootstrap analysis provided further evidence that the model is plausible and not capitalizing on chance relationships in the data or nonrepresentative individuals who skew the results. Garnier and Stein
CONCLUSION
This study highlights the importance of processes that link family and peer contexts over time in the development of adolescent problem behaviors. The findings support the proposal by Cairns and Cairns (1995) that predominant peer effects reflect earlier processes in childhood, and emphasize the need to include longitudinal data from both peer and family contexts in studies of trajectories leading to adolescent problem behaviors. In different ways, both experiences at home with parents and outside the home with peers are associated with children’s development. What children learn at home from their parents, they bring to the peer group and “when there is an array of peer groups, kids sort themselves out on the basis of their individual characteristics and gravitate to the group that provides the best fit” (Harris, 1998, 285). Furthermore, the emergence of different predictive structures for teen problem behaviors suggests the importance of allowing for effects of both a general problem behavior syndrome as well as unique and possibly positive effects of individual value orientations.
This study supports current suggestions for interventions with high-risk youth. The child, the family, and peers are all potential areas for multiple interventions. Programs should include early intervention efforts and focus on common antecedents, such as early social competence with peers and values or lack of a positive value system underlying the tendency to engage in related problem behaviors rather than separate behaviors. Programs aimed at helping adolescents make choices and decisions about their lives should include an understanding of the social influences in the adolescents’ environment and in their personal history including the values of adolescents and their families.
REFERENCES
Bandura, A. (1986). Social Foundations of Thought and Action: A Social
Cognitive Theory. Prentice-Hall, Englewood Cliffs, NJ.
Baumrind, D. (1985). Familial antecedents of adolescent drug use: A developmental perspective. In Jones, C. L., and Battjes, R. J. (eds.),
Etiology of Drug Abuse: Implications for Prevention (National Institute on Drug Abuse Research Monograph No. 56, U.S. Department of Health and Human Services Publication No. 1415). U.S.
Government Printing Office, Washington, DC, pp. 13–44.
Bentler, P. M. (1990). Comparative fit indexes in structural models.
Psychol. Bull. 107: 238–246.
Bentler, P. M. (1995). EQS Structural Equations Program. Multivariate
Software, Encino, CA.
Bentler, P. M., and Dijkstra, T. (1985). Efficient estimation via linearization in structural models. In Krishnaiah, P. R. (ed.), Multivariate
Analysis VI. North-Holland, Amsterdam, pp. 9–42.
Bentler, P. M., and Dudgeon, P. (1996). Covariance structure analysis:
Statistical practice, theory, and directions. An. Rev. Psychol. 47:
563–592.

P1: Vendor/FOM
Journal of Youth and Adolescence

pp334-joyo-364605

December 28, 2001

Family and Peer Effects
Berg, D. H. (1985). Reality construction at the family/society interface:
The internalization of family themes and values. Adolescence 20:
605–618.
Blanton, H., Gibbons, F. X., Gerrard, M., Conger, K., and Smith, G. E.
(1997). Role of family and peers in the development of prototypes associated with substance use. J. Fam. Psychol. 11: 271–288.
Block, J., Block, J. H., and Keyes, S. (1988). Longitudinally foretelling drug usage in adolescence: Early Childhood personality and environmental precursors. Child Dev. 59: 336–355.
Brook, J. S., Whiteman, M., Balka, E. B., and Cohen, P. (1995). Parent drug use, parent personality, and parenting, J. Gen. Psychol. 156:
137–151.
Brook, J. S., Whiteman, M., and Gordon, A. S. (1985). Father absence, perceived family characteristics, and stage of drug use in adolescence. Br. J. Dev. Psychol. 2: 87–94.
Brown, B. B. (1990). Peer groups and peer cultures. In Feldman, S., and Elliott, G. R. (eds.), At the Threshold. Harvard University,
Cambridge, pp. 171–196.
Cairns, R. B., and Cairns, B. D. (1995). Lifelines and Risks: Pathways of Youth in Our Time. Cambridge University, New York.
Cairns, R. B., Leung, M. C., and Cairns, B. D. (1995). Social networks over time and space in adolescence. In Crockett, L. J., and Crouter,
A. C. (eds.), Pathways Through Adolescence. Erlbaum, Mahwah,
NJ, pp. 1–12.
Chan, W., Yung, Y. F., Bentler, P. M., and Tang, M. L. (1998). Tests of independence for ordinal data using bootstrap. Educ. Psychol.
Meas. 58: 221–240.
Chassin, L., Presson, C. C., and Sherman, S. J. (1989). “Constructive” vs. “Destructive” Deviance in adolescent health-related behaviors.
J. Youth Adolesc. 18: 245–262.
Chou, C. P., and Bentler, P. M. (1990). Model modification in covariance structure modeling: A comparison among likelihood ratio,
Lagrange Multiplier, and Wald tests. Multivar. Behav. Res. 25: 115–
136.
Coltrane, S. (1995). The future of fatherhood: Social, demographic, and economic influences on men’s family involvements. In Marsiglio,
W. (ed.), Fatherhood: Contemporary Theory, Research, and Social
Policy. Research on Men and Masculinities Series 7. Sage, Thousand Oaks, CA.
Conger, R. D., and Reuter, M. A. (1996). Siblings, parents, and peers:
A longitudinal study of social influences in adolescent risk for alcohol use and abuse. In Brody, G. (ed.), Sibling Relationships: Their Causes and Consequences. Ablex, Norwood, NJ, pp. 1–30.
Coombs, R. H., Fawzi, F. I., and Gerber, B. E. (1986). Patterns of cigarette, alcohol and other drug use among children and adolescents: A longitudinal study. Int. J. Addict. 1: 897–
913.
Donovan, J., Jessor, R., and Costa, F. M. (1991). Adolescent health behavior and conventionality–unconventionality: An extension of problem-behavior theory. Health Psychol. 10: 52–61.
Dryfoos, J. G. (1990). Adolescents at Risk: Prevalence and Prevention.
Oxford University, New York.
Duncan, D. F., and Petosa, R. (1995). Social and community factors associated with drug use and abuse among adolescents.
In Gullotta, T. P., Adams, G. R., and Montemayor, R. (eds.),
Substance Misuse in Adolescence. Sage, Thousand Oaks, CA, pp. 56–91.
Eiduson, B. T. (1975). Drug using parents and their children: Part I anticipatory socialization behaviors, Report to the National Institute of Drug Use, UCLA, Los Angeles, CA.
Elder, G. H., Jr. (1980). Adolescence in historical perspective. In
Adelson, J. (ed.) Handbook of Adolescent Psychology. Wiley, New
York.
Elder, G. H., Liker, J. K., and Cross, C. E. (1984). Parent–child behavior in the Great Depression: Life course and intergenerational influences. Life Span Dev. Behav. 6: 109–158.
Elliott, D., and Huizinga, D. (1983). Social class and delinquent behavior in a national youth panel. Criminology 21: 149–177.

12:11

Style file version July 26, 1999

55
Elliott, D. S., Huizinga, D., and Ageton, S. S. (1985). Explaining Delinquency and Drug Use. Sage, Beverly Hills, CA.
Elliott, D. S., Huizinga, D., and Menard, S. (1989). Multiple Problem
Youth: Delinquency, Substance Use, and Mental Health Problems.
Springer, New York.
Garnier, H. E., and Stein, J. A. (1998). Values and the family: Risk and protective factors for adolescent problem behaviors. Youth Soc. 30:
89–120.
Garnier, H. E., Stein, J. A., and Jacobs, J. (1997). The process of dropping or stopping out of high school: A 19-year perspective. Am. Ed. Res.
J. 34: 395–419.
Garnier, H. E., and Weisner, T. S. (1992). Drug use in conventional and nonconventional families: Longitudinal studies of families and children. Paper presented at American Educational Research Association, San Francisco.
Goff, B. G., and Goddard, H. W. (1999). Terminal core values associated with adolescent problem behaviors. Adolescence 34: 47–59.
Grusec, J. E., and Goodnow, J. J. (1994). Impact of parental discipline methods on the child’s internalization of values: A reconceptualization of current points of view. Dev. Psychol. 30: 4–19.
Harris, J. R. (1998). The Nurture Assumption: Why Children Turn Out the Way They Do. Free Press, New York.
Hawkins, J. D., and Weis, J. G. (1985). The social developmental model:
An integrated approach to delinquency prevention. J. Prim. Prev.
6: 73–97.
Hollingshead, A. B. (1975). Four Factor Index of Social Position. Yale
University, Department of Sociology, New Haven, CT.
Hu, L. T., and Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Struct. Eq. Model. 6: 1–55.
Jessor, R., and Jessor, S. L. (1977). Problem Behavior and Psychosocial
Development. Academic Press, New York.
Jessor, R., Van Den Bos, J., Vanderryn, J., Costa, F. M., and Turbin,
M. S. (1995). Protective factors in adolescent problem behavior:
Moderator effects and developmental change. Dev. Psychol. 31:
923–933.
Johnston, H., Larana, E., and Gusfield, J. R. (1994). Identities, grievances, and new social movements. In Blankenhorn, D., Bayme,
S., and Elshtain, J. B. (eds.), Rebuilding the Nest. Family Service
America, Milwaukee.
Johnston, L., O’Malley, P., and Bachman, J. (2000). Monitoring the
Future: National Results on Adolescent Drug Use, 1999. National
Institute on Drug Abuse, Bethesda, MD.
Kandel, D. B., and Wu, P. (1995). The contributions of mothers and fathers to the intergenerational transmission of cigarette smoking in adolescence. J. Res. Adolesc. 5: 225–252.
Kandel, D. B., and Yamaguchi, K. (1987). Job mobility and drug use:
An event history analysis. Am. J. Soc. 92: 836–878.
Kumpfer, K. L., and DeMarsh, J. (1986). Family environmental and genetic influences on children’s future chemical dependency. J. Child.
Contemp. Soc. 18: 49–91.
Laible, D. J., Carlo, G., and Raffaelli, M. (2000). The differential relations of parent and peer attachment to adolescent adjustment.
J. Youth Adolesc. 29: 45–59.
Lambert, N. M., Hartsough, C. S., and Bower, E. M. (1979). Process for the Assessment of Effective Student Functioning. Publishers Test
Service, Monterey, CA.
National Institute of Drug Abuse (1995). National Household Survey on Drug Abuse: Main Findings 1992. National Institute of Drug
Abuse, U.S. Department of Health and Human Services, Publication No. (ADM) 94-3012, U.S. Government Printing Office,
Rockville, MD.
Newcomb, M. D., and Bentler, P. M. (1988). Consequences of Adolescent
Drug Use. Sage, Newbury Park, CA.
Phares, V., and Compas, B. E. (1992). The role of fathers in child and adolescent psychopathology: Make room for daddy. Psychol. Bull.
111: 387–412.
Reed, M. D., and Rountree, P. W. (1997). Peer pressure and adolescent substance use. J. Quant. Crim. 13: 143–180.

P1: Vendor/FOM
Journal of Youth and Adolescence

pp334-joyo-364605

December 28, 2001

12:11

56
Schaefer, E. S., Edgerton, M., and Aaronson, M. (1978). Classroom
Behavior Inventory. Unpublished 42-item research instrument.
Stein, J. S., and Newcomb, M. D. (1999). Adult outcomes of adolescent conventional and agentic orientations: A twenty-year longitudinal study. J. Early Adolesc. 19: 39–65.
Stein, J. S., Newcomb, M. D., and Bentler, P. M. (1987). An 8-year study of multiple influences on drug use and drug use consequences.
J. Pers. Soc. Psychol. 53: 1094–1105.
Stein, J. S., Newcomb, M. D., and Bentler, P. M. (1993). Differential effects of parent and grandparent drug use on behavior problems of male and female children. Dev. Psychol. 29: 31–43.
Steinberg, L., Dornbusch, S. M., and Brown, B. B. (1992). Ethnic differences in adolescent achievement: An ecological perspective. Am.
Psychol. 47: 723–729.

Style file version July 26, 1999

Garnier and Stein
U.S. Bureau of the Census (1999). Statistical Abstract of the United
States: 1999 (119th edn.). U.S. Census Bureau, Washington, DC.
Weisner, T. S. (1987). Socialization for parenthood in sibling caretaking societies. In Lancaster, J., Rossi, A., Altmann, J., and Sherrod, L.
(eds.), Parenting Across the Life Span. Aldin, New York, pp. 237–
270.
Weisner, T. S., and Garnier, H. E. (1992). Nonconventional family lifestyles and school achievement: A 12-year longitudinal study. Am.
Ed. Res. J. 29: 605–632.
Weisner, T. S., and Rochford, E. B. (1980). Family Lifestyles Project
Values Questionnaire. Unpublished manuscript, Sociobehavioral
Group, University of California, Los Angeles.
Whitbeck, L. B., and Gecas, V. (1988). Value attributions and value transmission between parents and children. J. Marr. Fam. 50: 829–840.

References: Bandura, A. (1986). Social Foundations of Thought and Action: A Social Cognitive Theory Baumrind, D. (1985). Familial antecedents of adolescent drug use: A developmental perspective Etiology of Drug Abuse: Implications for Prevention (National Institute on Drug Abuse Research Monograph No. 56, U.S. Department of Health and Human Services Publication No. 1415). U.S. Bentler, P. M. (1990). Comparative fit indexes in structural models. Bentler, P. M. (1995). EQS Structural Equations Program. Multivariate Software, Encino, CA. Bentler, P. M., and Dijkstra, T. (1985). Efficient estimation via linearization in structural models. In Krishnaiah, P. R. (ed.), Multivariate Analysis VI Bentler, P. M., and Dudgeon, P. (1996). Covariance structure analysis: Statistical practice, theory, and directions pp334-joyo-364605 December 28, 2001 (1997). Role of family and peers in the development of prototypes associated with substance use Block, J., Block, J. H., and Keyes, S. (1988). Longitudinally foretelling drug usage in adolescence: Early Childhood personality and environmental precursors Brook, J. S., Whiteman, M., Balka, E. B., and Cohen, P. (1995). Parent drug use, parent personality, and parenting, J Brook, J. S., Whiteman, M., and Gordon, A. S. (1985). Father absence, perceived family characteristics, and stage of drug use in adolescence Brown, B. B. (1990). Peer groups and peer cultures. In Feldman, S., and Elliott, G Cairns, R. B., and Cairns, B. D. (1995). Lifelines and Risks: Pathways of Youth in Our Time Cairns, R. B., Leung, M. C., and Cairns, B. D. (1995). Social networks over time and space in adolescence Chan, W., Yung, Y. F., Bentler, P. M., and Tang, M. L. (1998). Tests of independence for ordinal data using bootstrap Chassin, L., Presson, C. C., and Sherman, S. J. (1989). “Constructive” vs Chou, C. P., and Bentler, P. M. (1990). Model modification in covariance structure modeling: A comparison among likelihood ratio, Lagrange Multiplier, and Wald tests Coltrane, S. (1995). The future of fatherhood: Social, demographic, and economic influences on men’s family involvements Conger, R. D., and Reuter, M. A. (1996). Siblings, parents, and peers: A longitudinal study of social influences in adolescent risk for Coombs, R. H., Fawzi, F. I., and Gerber, B. E. (1986). Patterns of cigarette, alcohol and other drug use among children and adolescents: A longitudinal study Donovan, J., Jessor, R., and Costa, F. M. (1991). Adolescent health behavior and conventionality–unconventionality: An extension of Dryfoos, J. G. (1990). Adolescents at Risk: Prevalence and Prevention. Duncan, D. F., and Petosa, R. (1995). Social and community factors associated with drug use and abuse among adolescents. Eiduson, B. T. (1975). Drug using parents and their children: Part I anticipatory socialization behaviors, Report to the National Institute Elder, G. H., Jr. (1980). Adolescence in historical perspective. In Adelson, J Elder, G. H., Liker, J. K., and Cross, C. E. (1984). Parent–child behavior in the Great Depression: Life course and intergenerational influences Elliott, D., and Huizinga, D. (1983). Social class and delinquent behavior in a national youth panel 12:11 Style file version July 26, 1999 Elliott, D. S., Huizinga, D., and Menard, S. (1989). Multiple Problem Youth: Delinquency, Substance Use, and Mental Health Problems. Garnier, H. E., and Stein, J. A. (1998). Values and the family: Risk and protective factors for adolescent problem behaviors