Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Sep 14.
Published in final edited form as: J Fam Psychol. 1997 Sep;11(3):259–270. doi: 10.1037/0893-3200.11.3.259

Adolescent Modeling of Parent Substance Use: The Moderating Effect of the Relationship With the Parent

Judy A Andrews 1, Hyman Hops 1, Susan C Duncan 1
PMCID: PMC9473319  NIHMSID: NIHMS1799475  PMID: 36110397

Abstract

This study examined a hypothesis derived from social learning theory, that adolescents would be more likely to model the substance use of each parent if they had a relatively good relationship with the parent than if their relationship with that parent was relatively poor. Data from 657 adolescents (51% female; 11 to 15 years of age at the 1 st assessment), 357 fathers, and 633 mothers across a 6-year assessment period were used for these analyses. As hypothesized, all adolescents modeled mother’s cigarette use and father’s marijuana use, older adolescents modeled mother’s marijuana use and younger girls and older boys modeled father’s alcohol use if they had a relatively good or moderate relationship with that parent but did not model their parent’s use if the relationship with that parent was relatively poor. Caution is noted in assuming that relatively good relationships with a parent are always protective.

The influence of parental substance use on adolescents’ use of that same substance has been shown in numerous studies (e.g., Andrews, Hops, Ary, Tildesley, & Harris, 1993; Hops, Duncan, & Duncan, 1995). Substance-specific effects of parent use of cigarettes (Fisher, MacKinnon, Angling, & Thompson, 1987; Jacobs, Jerome, Sayers, Spielberger, & Weinberg, 1988; M.Murray, Kiryluk, & Swan, 1985; Rittenhouse & Miller, 1984), alcohol (Ary, Tildesley, Hops, & Andrews, 1993; Gfroerer, 1987; Kandel & Andrews, 1987; Lau, Quadrel, & Hartman, 1990; Needle et al., 1986), and marijuana (Brook, Whiteman, Gordon, & Brook, 1985; Duncan, Duncan, Hops, & Stoolmiller, 1995; Fisher et al., 1987; Forster, 1984;Johnson, Shontz, & Locke, 1984) on adolescent use have been found for 12- to 22-year-olds. Glynn (1981), after a review of the literature, concluded that parent use of a substance was the most powerful influence on an adolescent’s initiation of use of the same substance.

The power of parental substance use has often been attributed to adolescent modeling or imitation of parental behavior, as postulated by social learning theorists (e.g., Bandura & Walters, 1963). However, social learning theory also holds that only the behaviors of valued individuals are modeled. Whether an adolescent values a parent may be determined by the quality of their relationship. Therefore, we hypothesized that adolescents would be more likely to imitate the substance use of the parent in the context of a relatively good relationship with relatively little conflict than in the context of a relatively poor relationship.

An adolescent’s relationship with his or her parents is central to many theories of adolescent substance use. Weak attachment to parents has been identified as a risk factor for adolescent substance use in multiple theories (e.g., social control theory [Elliot, Huizinga, & Ageton, 1985; Elliot, Huizinga, & Manard, 1989], the social development model [Hawkins & Weiss, 1985], family interaction theory [Brook, Brook, Gordon, Whiteman, & Cohen, 1990], and problem behavior theory [Jessor & Jessor, 1977]). In addition, the quality of the bond between parent and adolescent has been shown to have a positive effect in restraining youth from engaging in various deviant or delinquent activities (Elder, 1980; Glueck & Glueck, 1940) and from initiating or increasing their use of substances (Andrews, Hops, Ary, Lichtenstein, & Tildesley, 1991; Duncan, Duncan, & Hops, 1994; Kandel, Kessler, & Marguiles, 1978).

Relatively few studies have included both parent use and the adolescent’s relationship with the parent as risk factors in the same model. Furthermore, most of the studies that have included both factors have assumed an additive model in which relationships with parents and parent use affect adolescent use, either directly (Anderson & Henry, 1994) or indirectly, through peer associations (Kandel & Andrews, 1987; Melby, Conger, Conger, & Lorenz, 1993) or the adolescent’s attitude (Kandel & Andrews, 1987). However, from the tenets of social learning theory, one should conclude that it is the interaction between parent use and the parent–adolescent relationship that is of primary importance, with modeling dependent on the quality of the relationship. Few studies have reported an examination of this interaction (Brook, Whiteman, Gordon, & Brook, 1986; Doherty & Allen, 1994; Foshee & Bauman, 1994). Results reported by Brook and her associates (1986) and Foshee and Bauman (1994) were consistent with social learning theory. Brook et al. (1986) found that the marijuana and illicit drug use of female adolescents who identified with their fathers resembled the substance use of their fathers, whereas father–adolescent substance use was not congruent if the daughter did not identify with her father. Foshee and Bauman (1994) showed that, among nonsmoking adolescents with smoking parents, the adolescent’s subsequent smoking increased as attachment to that parent increased. When parents were nonsmokers, subsequent smoking decreased as attachment to that parent increased. In contrast, Doherty and Allen (1994) showed that the effect of at least one parent smoking on the smoking onset of the adolescent 6 years later was amplified by low family cohesion. However, although these latter findings appear to be in contrast to the predictions of social learning theory, they are based on the parent’s report of family cohesion rather than the adolescent’s car parent’s report on the quality of the dyadic relationship.

These equivocal findings could be due, in part, to methodological and statistical problems extant in the literature. Although, in all of the studies cited, parents reported on their own or their partner’s substance use, the assessment of quality of the relationship was derived from a single respondent, either the adolescent or the parent. In the present study, not only did the adolescent and each parent report on their own substance use, but the assessment of the adolescent–parent relationship included measures from both parent and adolescent. Also, in most longitudinal studies, the amount of time between measurement of parent use and measurement of adolescent use has varied, with lengthy intervals between assessments. This is problematic because of the dynamic nature of substance use during this developmental period. Most adolescents make the transition from nonuse to use and back again over the course of time. Even among adults, quitting and relapsing is a frequent occurrence. Several reviews (Bertrand & Abernathy, 1993; Conrad, Flay, & Hill, 1992) suggest that parental substance use is more predictive of the adolescent’s concurrent rather than subsequent substance use. Thus, studies examining the modeling of parent use require concurrent measures of adolescent and parent use or measures close in time.

However, to capture the dynamic nature of substance use and to fully understand the development of substance use during adolescence, with its continual change and growth, it is important to examine the phenomenon as a long-term dynamic process. Thus, the present study used generalized estimating equations (GEE) methodology to examine whether serial fluctuations in adolescent substance use correspond to synchronous changes in each parent’s substance use. Using GEE, this study examined the concurrent relation between parent substance use and adolescent use, as a function of the relationship with the parent, across six annual assessments.1

In contrast to most previous studies, the present study examined the independent influence of mothers and fathers, as well as gender and age effects, for each substance separately. Previous research has shown that parental influences and parent–adolescent relationships may be sensitive to the gender of the parent and adolescent as well as to developmental processes (e.g., Andrews et al., 1993; Hops, Duncan, Duncan, & Stoolmiller, 1996). In general, adolescents are more heavily influenced by mothers than by fathers (Baer & Corrado, 1974; Brook et al., 1986; Kandel, 1974; Smart & Fejer, 1972), but this influence is moderated by the age and gender of the adolescent (Andrews et al., 1993). Moreover, the literature suggests that there is increased mother–daughter conflict in adolescence (Steinberg, 1988), indicating that daughters may be less likely to model their mothers. Taken together, these studies suggest that an investigation of the interaction of parent modeling and the relationship with that parent should consider the gender and age of the offspring.

Furthermore, prevalence of use varies by type of substance, age, and gender. Most adolescents initiate their substance use with alcohol and make the transition to cigarette use and then to use of marijuana and hard drugs (Andrews, Hops, Ary, Lichtenstein, & Tildesley, 1991). Similarly, for all ages within adolescence, alcohol use is most prevalent, followed by cigarette use, marijuana use, and, finally, other illicit drug use (National Institute on Drug Abuse [NIDA], 1990). Furthermore, the frequency and prevalence of substance use increase as the adolescent ages (NIDA, 1990), with a period of rapid growth between middle school and high school (Dishion, Capaldi, Spracklen, & Li, 1995). In addition, with the exception of cigarette use (Mermelstein & Borrelli, 1995), use of substances is more prevalent and more frequent among male adolescents than among female adolescents (Kandel & Yamaguchi, 1985; Lewinsohn, Rohde, & Seeley, 1996). This variation in prevalence across substances and between ages and genders, along with identification of diverse risk factors associated with the use of each substance that vary by age and gender (Kaplan & Johnson, 1992), warrants separate analyses for each substance and an investigation of age and gender differences. Furthermore, similar to previous studies (Andrews et al., 1993), we controlled for number of parents by including this variable in analyses.

Thus, the purpose of the present study was to provide a direct test of a facet of social learning theory. We hypothesized that adolescents would model both the use and nonuse of substances by a parent if they had a relatively good relationship with that parent but not if the relationship was relatively poor. To test this hypothesis, we examined, across time, the interaction between parent use and the adolescent’s relationship with the parent as it predicted adolescent substance use.

Method

Sample

Seven hundred sixty-three adolescents 11 through 15 years of age, along with their parents and siblings, were recruited as part of an ongoing 12-year longitudinal study of family influence on adolescent substance use. Families were originally recruited from moderate-sized northwestern urban areas, via newspaper, television, and radio announcements and flyers placed at schools, for a 3-year study investigating family influences on adolescent tobacco use. An effort was made to recruit adolescents who could be considered at risk for tobacco and other substance use. Thus, families were selected disproportionately if one or both parents smoked. Data derived from the first six annual assessments are presented here.

Attrition was high in the initial years, 31% of the sample having left the study between Time 1 (Tl) and Time 4 (T4). By T4, the sample size of 530 had stabilized, leveling off at 496 by Time 6 (T6). A comparison of those who completed all six assessments and those who dropped out before the sixth assessment revealed no differences in the relative proportion of male and female participants. However, those who left the study tended to be older, χ2(4, N = 760) = 9.89, p < .05, and had fathers with less education, χ2(2, N = 760) = 15.32, p < .05. Consistent with other Longitudinal studies (e.g., Kandel, 1984; Newcomb & Bentler, 1988), those who left the study by T6 were more likely to have used cigarettes, χ2(1, N = 760) = 13.12, p < .001; alcohol, χ2(1, N = 760) = 6.38, p < .05; and marijuana, χ2(1, N = 760) = 14.89, p < .001.

Although substance users disproportionately dropped out of the study, the prevalence of substance use among adolescents remaining in the study was similar to the prevalence of use among adolescents in the region. We compared our T4 data from 8th and 11th graders with data obtained from all 8th and 11th graders in Oregon (Oregon Employment Division Research and Statistics, 1989) during the same period. Prevalence of monthly use was comparable across all substances with the exception of cigarettes, in which use was significantly lower for our sample.

Data from 657 target adolescents (335 girls [51.0%] and 322 boys) and from 357 fathers and 633 mothers with complete data for each analysis were used in the present study. At the first assessment, target adolescents were primarily Caucasian (92.4%) and ranged in age from 11 to 15 years (M = 13.15, SD = 1.67). Most (54.3%) adolescents lived in two-parent households; 42.8% lived with a single mother, and 2.9% lived with a single father. Nearly all parents had completed high school (94.7% of mothers and 93.1% of fathers), and more than two thirds had some college education (67.8% of mothers and 70.0% of fathers). Comparisons of the T6 sample with 1990 census data from the county from which participants were recruited revealed no differences in terms of race–ethnicity, gender, or household sire. However, the present sample had a higher proportion of single mothers (37.2% vs. 21% in die county), and adults in the present sample were better educated than adults in the county (38.3% vs. 22.2% with a college degree).

Procedure

Simultaneously, in separate rooms, adolescents and their parents completed parallel self-report questionnaires assessing the extent of their own substance use and their perception of the parent–adolescent relationship, as well as other psychosocial variables. Before completing the questionnaire, adolescents were asked to hold their breath for 20 seconds and then breathe into an air bag to provide an assessment of expired-air carbon monoxide. This procedure has been shown to enhance the validity of adolescents’ self-reports regarding tobacco use (D. M. Murray, O’Connell, Schmid, & Perry, 1987). Each family was paid $35 for each of the first three annual assessments, and each participating family member was paid $25 for the fourth through sixth assessments.

Measures

Adolescent substance use.

Adolescents’ current alcohol, cigarette, and marijuana use was measured via their self-report of current (rate of use in the last 24 hr, month, and 6 months) and lifetime (an ordinal variable ranging from never tried and used to … but don’t anymore to at least once a day) use. For this study, extent of use was ignored. Adolescents were considered current users (coded as 1) if they reported currently using a substance “at least once in awhile” and had a rate of use that was greater than zero times per month over the previous 6 months. Adolescents were considered current nonusers (coded as 0) if they were never users (report of never tried, along with a rate of zero times per month in the last 6 months) or previous but not current users (report of quitting, along with a rate of zero times per month in the last 6 months).

Parent substance use.

Each parent reported on his or her own substance use. Parent alcohol use was measured with a 7-point ordinal scale, and parents cigarette and marijuana use was measured with 5-point ordinal scales. For all substances, the scales ranged from never and quit to extensive use (alcohol: daily; cigarettes: greater than a pack a day; and marijuana: once a day). As with the adolescents, extent of use was ignored, and current users were considered those who currently used at least occasionally (coded as 1 [nonuse was coded as 0]).

Parent–adolescent relationships.

The quality of the relationship between each adolescent and his or her respective parent was measured by the adolescent’s report on the Appraisal of Mother and Appraisal of Father subscales of the Conflict Behavior Questionnaire (Prinz, Foster, Kent, & O’Leary, 1979) and each parent’s report on this questionnaire’s Appraisal of the Adolescent subscale. The Conflict Behavior Questionnaire was designed to obtain an evaluation of the parent–adolescent dyadic relationship directly from the parent and the adolescent. The mother–adolescent relationship construct was created by summing the subscales measuring the mother’s appraisal of the adolescent and the adolescent’s appraisal of his or her mother. The correlations of these two scales were .36, .41, .42, .42, .46, and .31 for Assessments 1–6, respectively. Similarly, the father–adolescent relationship construct was created by summing the subscales measuring the father’s appraisal of the adolescent and the adolescent’s appraisal of his or her father. The correlations between these two scales were .34, .45, .46, .45, .49, and .41 for Assessments 1–6, respectively. These subscales were scored so that a high score reflected a positive parent–adolescent relationship with less dyadic conflict.

Data Analysis

Statistical methods for the analysis of longitudinal data often perform poorly in settings that incorporate missing observations, attrition, time-varying covariates, and other factors that may make standard multivariate procedures inappropriate. GEE methodology (Liang & Zeger, 1986; Zeger & Liang, 1986) provides an approach that extends the generalized linear models framework of McCullagh and Nelder (1983) for correlated observations, such as those arising from repeated measures over time on a sample of independent individuals. GEE can handle a variety of assumed correlation structures (e.g., independence, exchangeable, autoregressive, and fully specified) and a number of mean–variance relations (e.g., normal, Poisson, and binomial). Zeger and Liang (1986) have demonstrated that, using GEE, it is possible to obtain consistent estimates of coefficients for mean structure models and good statistical tests in large samples (assuming the model is correct), even if the assumed correlational matrix is misspecified. Test statistics will be most powerful, however, when the assumed correlation matrix most closely approximates the true correlation matrix.

The general analytic strategy for using GEE with longitudinal data is to view the analysis as a regression model with correlated residuals. The correlation between observations over time is therefore viewed as a nuisance parameter. The model is focused on the outcome mean unconditional on its prior history. GEE is an attractive analytic format because it allows flexibility in deciding how to handle repeated measures when the outcome is not necessarily normally distributed and the correlation structure of the repeated observations is not known. Applications of GEE can be found in Duncan et al. (1995); Duncan, McAuley, Stoolmiller, and Duncan (1993); Hill et al. (1994); and Hops et al. (1996). Additional details on the methodology of GEE are available in Duncan et al. (1995) and Zeger and Liang (1986).2

In the present study, GEE was used to examine the effect of changes in the independent variable (e.g., parent substance use–nonuse of alcohol, cigarettes, and marijuana) on changes in the dependent variables (e.g., adolescent use–nonuse of alcohol, cigarettes, and marijuana) over a 6-year period. Synchronous models were examined (independent variables predicting the dependent variable at the same time point) because the purpose of the study was to investigate concurrent relations between parent and adolescent substance use over the 6-year period. If adolescent modeling varies as a function of the relationship with the parent, then a significant interaction would be expected between parental use and the parent–adolescent relationship. Furthermore, if the tenets of social learning theory were supported, it would be expected that more modeling would occur if the adolescent had a relatively good as opposed to a relatively poor relationship with the parent.

For these particular analyses, an autoregressive correlational structure with a binomial probability density function was assumed, because the dependent variable was dichotomous. In the autoregressive structure, correlations depend only on the time separating two repeated observations. Mth-order autoregressive assumes that all correlations are a function of m parameters, which are raised to the ∣t − t'∣ power. The size of the exponent increases for higher order lags and implies that the correlation decreases (see Duncan et al., 1995, for more details regarding correlation matrices and mean–variance relations within GEE).

The models encompassed a relatively large number of interactions; thus, for reasons of parsimony, models were tested via backward elimination (Cohen & Cohen, 1983). In the initial multivariate model, all main effects (e.g., parent use, relationship with parent, age, gender, and single–two-parent status) and interactions of interest (age and gender with parent use and relationship with parent) were included. Through backward elimination, starting with the highest order interaction, the nonsignificant interaction (p > .05) with the smallest effect size was dropped from the model, and the model was reestimated. Thus, use of this procedure results in lower order interactions associated with significant higher order interactions remaining in the model. To explore the significant interaction effects across time, we computed simple slopes and simple effects from the robust estimated parameters and the associated covariance matrix using standard techniques (Aiken & West, 1991).

To aid in the interpretation of interactions, we trichotomized the parent–adolescent relationship variable into high, medium, and low levels, using, in most cases, plus one standard deviation and above to represent a good parent–adolescent relationship, minus one standard deviation and below to represent a poor parent–adolescent relationship, and between minus and plus one standard deviation to represent a moderate parent–adolescent relationship. In addition, to understand the interactions with age, we divided adolescents into two age groups: younger adolescents who were less than 14 years of age (n = 386) and older adolescents who were 14 years of age and older (n = 271) at the first assessment. This age breakdown typifies the middle versus high school student.

Results

Description of Use

Adolescent use.

At T1, 50%, 30%, and 25% of girls and 46%, 22%, and 20% of boys were considered users of alcohol, cigarettes, and marijuana, respectively. The proportions increased with time to 80%, 43%, and 34% of girls and 69%, 43%, and 36% of boys at T4 and 83%, 45%, and 41% of girls and 79%, 47%, and 41% of boys at T6. The majority of alcohol (T1: 75%; T4: 67%; and T6: 64%) and marijuana users (T1: 71%; T4: 71%; and T6: 77%) used less than four times per month across assessments. Adolescent use was moderately correlated across substances but decreased with time. The correlations were .59 (alcohol and cigarettes), .62 (alcohol and marijuana) and .70 (marijuana and cigarettes) at Tl; .47 (alcohol and cigarettes), .50 (alcohol and marijuana), and .63 (marijuana and cigarettes) at T4; and .44 (alcohol and cigarettes), .43 (alcohol and marijuana), and .53 (marijuana and cigarettes) at T6.

Parent use.

At T1, 78% of mothers and 76% of fathers drank alcohol, 30% of mothers and 31% of fathers smoked cigarettes, and 21% of mothers and 19% of fathers used marijuana. At T4, 67% of mothers and 59% of fathers drank alcohol, 26% of mothers and 23% of fathers smoked cigarettes, and 15% of mothers and 15% of fathers smoked marijuana. At T6, 64% of mothers and 60% of fathers drank alcohol, 21% of mothers and 22% of fathers smoked cigarettes, and 13% of mothers and 10% of fathers smoked marijuana. In two-parent homes, correlations across parents were moderate, averaging .50 for alcohol, .45 for cigarettes, and .70 for marijuana.

Synchronous Generalized Estimating Equations Predicting Adolescent Use

Mother ‘s use.

Table 1 presents the estimated regression coefficients and associated robust t statistics for the synchronous GEE analyses predicting adolescent use from mother’s use and the adolescent’s relationship with his or her mother. Thus, for these models, mother’s use, the mother–adolescent relationship, and adolescent use were measured concurrently.

Table 1.

Significant Predictors of Adolescent Use From Mother’s Use and Relationship With Mother

Substance use
Alcohol
(n = 633; 2,996
observations)
 Cigarettes
(n = 629; 2,978
observations)
 Marijuana
(n = 622; 2,956
observations)
Predictor variable or variables β t(2985) β t(2971) β t(2946)
Age 0.38 1.18 0.67 5.13*** 0.99 6.70***
Gender 0.02 0.10 0.25 1.89 −0.01 −0.10
Marital status −0.47 −3.79*** −0.49 −3.84 −0.75 −5.52***
Mother’s use 0.33 1.89 0.33 2.83 0.37 2.02*
Relationship with mother −0.04 −0.63 −0.30 −6.00*** −0.20 −2.84**
Age × Gender 1.32 2.79**
Age × Mother’s Use 0.77 2.27* 0.33 1.24
Gender × Mother’s Use 0.16 0.59
Mother’s Use × Relationship 0.29 3.67*** −0.01 −0.08
Age × Relationship −0.06 −0.65
Gender × Relationship −0.27 −2.86**
Age × Gender × Mother’s Use −1.13 −2.24*
Age × Gender × Relationship
Age × Mother’s Use × Relationship 0.50 2.40*
Open in a new tab
*

p < .05.

**

p < .01.

***

p < .001.

The hypothesized interaction of mothers’ alcohol use with the mother–adolescent relationship in the prediction of adolescent alcohol use was not significant; mother’s use and relationship with mother were independent predictors of alcohol use, but both were dependent on the age and gender of the adolescent.3 However, as hypothesized, the interaction of mother's use with the mother–adolescent relationship was significant in the prediction of both the adolescent’s cigarette use and his or her marijuana use, the latter moderated by the age of the adolescent. Examination of the simple effects provided support for a social learning perspective. The relationship between mother’s and adolescent’s cigarette use was significant for those adolescents with relatively moderate, β = 0.33, t(2971) = 2,83, p < .01, and relatively good, β = 0.63, t(2971) = 4.37, p < .001, relationships with their mother but not for those with relatively poor relationships, β = 0.04, t(2971) = 0.27, ns. A similar effect was noted for marijuana use among older adolescents. Among older adolescents, adolescent–mother marijuana use was concordant for those with relatively moderate, β = 0.69, t(2946) = 3.65, p < .001, and good, β = 1.18, t(2946) = 4.63, p < .001, relationships with their mother but not for those with relatively poor relationships, β = 0.21, t(2946) = 0.96, ns. For younger adolescents, although the relation between mother’s and adolescent’s marijuana use was significant only for those with a relatively moderate relationship with their mother, β = 0.37, t(2946) = 2.02, p < .05, concordance varied little as a function of the relationship with mother: relatively good relationship, β = 0.36, t(2946) = 1.36, ns, and relatively poor relationship, β = 0.38, t(2946) = 1.86, ns.

Father’s use.

The results of the GEE analyses for prediction of the adolescent’s substance use from father’s use and relationship with the father are shown in Table 2. The hypothesized interaction between father’s alcohol use and relationship with the father in the prediction of alcohol use was moderated by the age and gender of the adolescent. An examination of the four-way interaction suggested that the effect was in the expected direction for younger girls: relatively poor relationship, β = 0.48, t(1579) = 1.33, ns; relatively moderate relationship, β = 0.68, t(1579) = 3.00, p < .01; and relatively good relationship, β = 0.89, t(1579) = 2.98, p < .01. The effect was also in the predicted direction for older boys: relatively poor relationship, β = 0.31, t(1579) = .58, ns; relatively moderate relationship, β = 1.03, t(1579) = 2.90, p < .01; and relatively good relationship, β = 1.76, t(1579) = 3.53, p < .001. The effect of the relationship with father was opposite to that predicted for younger boys. For these boys, the relation between father’s use and adolescent’s use was significant for those with relatively poor, β = 0.68, t(1579) = 2.11, p < .05, relationships with their father but not for those with relatively moderate, β = 0.42, t(1579) = 1.85, ns, or relatively good, β = 0.16, t(1579) = 0.51, ns, relationships. There was no relation between father’s use and adolescent’s use for older girls: relatively poor, β = −0.20, t(1579) = −0.34, ns; relatively moderate, β = −0.23, t(1579) = −0.57, ns; and relatively good, β = −0.26, t(1579) = −0.69, ns.

Table 2.

Significant Predictors of Adolescent Use From Father’s Use and Relationship With Father

Substance use
Alcohol
(n = 356; 1,596
observations)
 Cigarettes
(n = 357; 1,598
observations)
 Marijuana
(n = 350; 1,570
observations)
 Predictor variable or variables β t(1579) β t(1592) β t(1559)
Age 0.64 1.65 0.89 4.91*** 1.59 5.68***
Gender 0.07 0.24 0.02 0.11 0.13 0.43
Marital status 0.22 0.75 −0.29 −0.73 −0.25 −0.54
Father’s use 0.42 1.85 0.16 0.87 0.98 4.60***
Relationship with father −0.03 −1.68 −0.20 −3.41*** −0.49 −3.34***
Age × Gender 1.33 2.31* 0.26 −0.63
Age × Father’s Use 0.61 1.45
Gender × Father’s Use 0.26 0.81
Father’s Use × Relationship −0.26 −1.16 0.27 2.23*
Age × Relationship −0.64 −1.59 0.46 2.43*
Gender × Relationship −0.29 −1.04 0.12 0.67
Age × Gender × Father’s Use −1.53 −2.4*
Age × Gender × Relationship 0.66 1.27 −0.51 −2.04*
Age × Father’s Use × Relationship 0.99 2.26*
Gender × Father’s Use × Relationship 0.46 1.41
Age × Gender × Father’s Use × Relationship −1.22 −2.09*
Open in a new tab
*

p < .05.

***

p < .001.

Contrary to expectations, the interaction of relationship with father and father’s cigarette use in the prediction of adolescents’ cigarette use was not significant. Only relationship with father predicted adolescents’ cigarette use. However, in the prediction of adolescents’ marijuana use, the interaction between relationship with father and father’s use was in the hypothesized direction. For adolescents with relatively poor relationships with their father, father’s use was not related to the adolescent’s use, β = −0.44, t(1559) = 1.70, ns; however, the association was significant for adolescents with more moderate relationships with their father, β = 0.98, t(1559) = 4.60, p < .001, and for adolescents with relatively good relationships with their father, β = 1.52, t(1559) = 4.03, p < .001.4 Thus, the hypothesis, derived from social learning theory, that father use would interact with the father–adolescent relationship to predict adolescent use was supported for alcohol for younger boys and older girls and for marijuana for all adolescents.

Discussion

Although this study has several strengths, including the examination of concurrent substance use by parents and adolescents over 6 years and multiple respondents, two methodological weaknesses may limit the generalizability of our findings. First, the sample consisted of volunteers from the community who answered advertisements. A comparison of the demographics of our sample with community demographics suggested that our sample was better educated and consisted of more single mothers than the community at large. Indeed, most of the parents in our sample had graduated from high school. Caution is needed in generalizing our findings to a less educated sample of parents. Second, those adolescents who had left the study by T6 were more likely to be substance users, which also affected the representativeness of the sample. Despite this limitation, the substance use of adolescents in our sample was comparable to the substance use of adolescents in the region.

The goal of this research was to examine a specific hypothesis derived from social learning theory (Bandura & Walters, 1963), that adolescents will be more likely to model the substance use of their parent if they have a relatively good, less conflictual relationship with that parent than if the relationship with that parent is relatively poor. Our results provide considerable support for this hypothesis. All adolescents modeled their mother’s use of cigarettes and their father’s use of marijuana if they had a relatively moderate or good relationship with that parent and did not model the substance use of the parent if the relationship was relatively poor. Age and gender effects were found for some substances. Older adolescents modeled their mother’s use of marijuana, and younger girls and older boys modeled their father’s use of alcohol, if they had a relatively moderate or good relationship with the parent; they did not model the parent’s use if they had a relatively poor relationship with the parent.

Most etiological theories of adolescent substance use have assumed that a good parent–adolescent relationship is protective, delaying or decreasing substance use (Brook et al., 1990; Hawkins & Weiss, 1985) and buffering the effect of other risk factors (Elder, 1980; Farrell, Barnes, & Banerjee, 1995). However, the results of the present research suggest that a relatively good parent–adolescent relationship is not always protective. In many cases, parent use, a variable often conceptualized as a risk factor, appears to influence substance use, but only within the context of a relatively good or moderate relationship with the patent. Within the context of a poor relationship, parent use usually is not related to adolescent use.

Within a social learning framework (e.g., Bandura & Walters, 1963), two explanations for adolescent–parent concordance in substance use are plausible. First, adolescents may imitate their parent’s use if they value that parent. Second, parents may encourage and reinforce the adolescent’s use if they themselves use and if they have a good relationship with the adolescent. One or both mechanisms may help to explain the relation between parent and adolescent use found in this study. Perhaps, among this educated sample, parents’ attitude toward the substance use of their offspring is more liberal. Thus, if parents use and the dyadic relationship with their son or daughter is relatively moderate or good, they may be more willing to use with or offer the substance to the adolescent, and the adolescent, in turn, may be more willing to accept their offer.

Dishion, Patterson, and colleagues (Dishion, 1990; Dishion et al., 1995; Patterson, Reid, & Dishion, 1992) have shown that adolescents with conflictual parent–adolescent relationships are more likely to interact with deviant peer groups that condone the use of substances. Our results suggest that the substance use of the adolescent in this context is not dependent on the substance use of his or her parents. Thus, the substance use of adolescents with conflictual parent–adolescent relationships is more likely to be influenced by the substance use of their friends, who are likely to use substances.

A word of caution in interpretation is needed here. In the present study, experimental users and less frequent users were not distinguished from heavy users. Different etiological factors may well differentiate these two groups of users. The theorizing of Dishion, Patterson, and colleagues (Dishion, 1990; Dishion et al., 1995; Patterson et al., 1992) is based primarily on the behavior of antisocial boys, who are most likely heavier substance users. In contrast, our sample consisted predominantly of experimental or less frequent users, and thus our findings may be less generalizable to this antisocial group.

Mother’s Versus Father’s Influence

It is noteworthy that the results of this study indicate, in contrast to the findings of others (Brook et al., 1985; Kandel, 1974; Rittenhouse & Miller, 1984; Smart & Fejer, 1972), that both parents influence their child’s substance use. However, our results suggest that the relative influence of mothers and fathers varies across substances. For example, adolescents appear to model their mother’s smoking as a function of the relationship with the mother; however, they do not appear to model their father’s smoking. Given mothers’ heavier caretaking responsibilities and more active relationship with their adolescent (Steinberg, 1990), adolescents most likely spend more time with their mother than with their father. Because cigarettes are used throughout the day, adolescents may be more likely to see and imitate their mother’s use or nonuse. In contrast, concordance between parent and adolescent alcohol use as a function of the parent–adolescent relationship was found for fathers but not mothers. Fathers’ drinking, which is more likely to occur in the evenings, may be more frequent and thus more apparent than that of mothers (NIDA, 1990). Fathers may also be more willing than mothers to share a drink with their adolescent if the dyadic relationship is good and involves little conflict.

Gender and Developmental Effects

Although the results of this study suggest that, for some substances, the interaction of parent relationship with parent use is moderated by age and gender, no previous studies have examined the moderating effect of these variables. Hence, the interpretation of these effects is more difficult and must be derived from the substance use and general developmental literature. We found mother–adolescent concordance in marijuana use as a function of the relationship with the mother for older but not younger adolescents. Marijuana use increases as the adolescent ages (NIDA, 1990) and is considered more normative for older than younger adolescents. Mothers may be sensitive to normative expectations and may be more willing to encourage the use of marijuana by their older than by their younger adolescents if they have a relatively moderate or good relationship with the adolescent.

Father–adolescent concordance in alcohol use as a function of the relationship with the father was moderated not only by the age of the adolescent but also by the adolescent’s gender. Thus, the hypothesis derived from social learning theory was supported only for younger girls and older boys. As adolescents age, they spend less time in the family setting and more time with peers (Montemayor, 1982). This may be particularly true for girls, as a result of increased mother–daughter conflict (Steinberg, 1988). For older girls, this decrease in the amount of time spent with the family may explain the finding of no relation between the alcohol use of fathers and that of their older daughters. Although there was an interaction between parent use and relationship with the parent for younger boys, the effect was opposite to the predicted direction. For younger boys, a relatively poor relationship amplified rather than moderated the effect of parent use. Boys, in general, are less conforming and more rebellious than girls (Baumrind, 1985), and these behaviors may be exacerbated by their youth. Thus, if these boys have relatively poor relationships with their father, they may be more willing to engage in an antisocial activity that they have seen in the home: their father’s alcohol use.

Conclusions

This study was a rather simple test of a very specific hypothesis. It did not consider the covariation between the two independent variables (e.g., that the substance use of the parent can affect the parent–adolescent relationship) or that other factors are influential in determining the substance use of the adolescent (e.g., modeling of peer use). In addition, it did not examine other aspects of parenting skills, such as parental monitoring. However, results regarding the influence of parent use on adolescent use as a function of the parent–adolescent relationship are quite clear. Supporting the tenets of social learning theory (Bandura & Walters, 1963), we found parent–adolescent congruence in use for those with relatively moderate or good relationships with that parent, but not for those with relatively poor relationships with that parent, among all adolescents for mother’s cigarette use and father’s marijuana use, among older adolescents for mother’s marijuana use, and among older boys and younger girls for father’s alcohol use. Thus, any treatment efforts that ignore the parent–adolescent relationship or the parents’ use of specific substances and their potential interaction are bound to have less of an impact than a combined venture.

Acknowledgments

This research was supported by Grant DA 03706 from the National Institute on Drug Abuse. An earlier version of this study was presented at the biennial meeting of the Society for Research in Adolescence, March 1993, San Diego, CA.

We wish to thank Peggi Rodgers for her contributions in the preparation of this article.

Footnotes

1.

Consistent with most previous research (e.g., Coombs, Paulson, & Richardson, 1991; M. Murray et al., 1985; Stanton & Silva, 1992), we chose to investigate the effect of parent use versus nonuse on adolescent use versus nonuse of a given substance rather than the effect of the quantity of parent use on the quantity of adolescent use.

2.

GEE was originally developed as an SAS (1985) macro by M. Rezaul Karim and Scot L. Zeger of the Department of Biostatistics, Johns Hopkins University; it was adapted for use with the SPSS (1990) environment for the present study. As a means of facilitating the use of GEE, SPSS macro language was used for passing parameters. Documentation for the statistical program used in the present study can be obtained from Oregon Research Institute, 1715 Franklin Boulevard, Eugene, Oregon 97403. For more detailed statistical background, see Zeger and Liang (1986) and references therein.

3.

Investigation of the significant Age × Gender × Mother’s Use interaction indicated that although mother’s alcohol use was significantly related to use among younger girls, β = 0.49, t(2985) = 2.31, p < .05, and older boys, β = 1.09, t(2985) = 3.79, p < .001, it was not significantly related to use among younger boys, β = 0.33, t(2985) = 1.89, ns, and older girls, β = 0.12, t(2985) = 0.39, ns. Exploration of the interaction of gender with the mother–adolescent relationship indicated that the mother–adolescent relationship was inversely related to alcohol use among girls, β = −0.31, t(2985) = −4.46, p < .001, but not among boys, β = −0.04, t(2985) = −0.63, ns.

4.

An interpretation of the Age × Gender × Relationship interaction suggested an inverse association between relationship with father and use among younger boys, β = −0.49, t(1559) = −3.33, p < .001; younger girls, β = −0.37, t(1559) = −2.98, p < .01; and older girls, β = − 0.41, t(1559) = −3.51, p < .001, but not among older boys, β = −0.03, t(1559) = −0.24, ns.

References

  1. Aiken LS, & West SG (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage. [Google Scholar]
  2. Anderson AR, & Henry CS (1994). Family system characteristics and parental behaviors as predictors of adolescent substance use. Adolescence, 29, 405–420. [PubMed] [Google Scholar]
  3. Andrews JA, Hops H, Ary D, Lichtenstein B, & Tildesley E (1991). The construction, validation, and use of a Guttman scale of adolescent substance use: An investigation of family relationships. Journal of Drug Issues, 21, 556–572. [Google Scholar]
  4. Andrews JA, Hops H, Ary D, Tildesley E, & Harris J (1993). Parental influence on early adolescent substance use: Specific and nonspecific effects. Journal of Early Adolescence, 13, 285–310. [Google Scholar]
  5. Ary DV, Tildesley E, Hops H, & Andrews J (1993). The influence of parent, sibling and peer modeling and attitudes on adolescent use of alcohol. International Journal of the Addictions, 28, 853–880. [DOI] [PubMed] [Google Scholar]
  6. Baer DJ, & Corrado JJ (1974). Heroin addicts’ relationships with parents during childhood and early adolescent years. Journal of Genetic Psychology, 124, 99–103. [DOI] [PubMed] [Google Scholar]
  7. Bandura A, & Walters RH (1963). Social learning and personality development. New York: Holt, Rinehart & Winston. [Google Scholar]
  8. Baumrind D (1985). Familial antecedents of adolescent drug use: A developmental perspective. In Jones CL & Battjes RJ (Eds.), Etiology of drug abuse: Implications for prevention (NIDA Research Monograph 56, DHHS Publication No. ADM 85–1335, pp. 13–44). Washington, DC: National Institute on Drug Abuse. [PubMed] [Google Scholar]
  9. Bertrand LD, & Abernathy TJ (1993). Predicting cigarette smoking among adolescents using cross-sectional and longitudinal approaches. Journal of School Health, 63, 98–103. [DOI] [PubMed] [Google Scholar]
  10. Brook JS, Brook DW, Gordon AS, Whiteman M, & Cohen P (1990). The psychosocial etiology of adolescent drug use: A family interactional approach. Genetic, Social, and General Psychology Monographs, 116, 111–267. [PubMed] [Google Scholar]
  11. Brook JS, Whiteman M, Gordon AS, & Brook DW (1985). Father’s influence on his daughter’s marijuana use viewed in a mother and peer context. Advances in Alcohol and Substance Abuse, 4, 165–190. [DOI] [PubMed] [Google Scholar]
  12. Brook JS, Whiteman M, Gordon AS, & Brook DW (1986). Father-daughter identification and its impact on her personality and drug use. Developmental Psychology, 22, 743–748. [Google Scholar]
  13. Cohen J, & Cohen P (1983). Applied multiple regression/correlation analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. [Google Scholar]
  14. Conrad KM, Flay BR, & Hill D (1992). Why children start smoking: Predictors of onset. British Journal of Addictions, 87, 1711–1724. [DOI] [PubMed] [Google Scholar]
  15. Coombs RH, Paulson MJ, & Richardson MA (1991). Peer vs. parental influence in substance use among Hispanic and Anglo children and adolescents. Journal of Youth and Adolescence, 20, 73–88. [DOI] [PubMed] [Google Scholar]
  16. Dishion TJ (1990). The family ecology of peer relations in middle childhood. Child Development, 61, 874–892. [DOI] [PubMed] [Google Scholar]
  17. Dishion TJ, Capaldi D, Spracklen KM, & Li F (1995). Peer ecology of male adolescent drug use. Development and Psychopathology, 7, 803–824. [Google Scholar]
  18. Doherty WJ, & Allen W (1994). Family functioning and parental smoking as predictors of adolescent cigarette use: A six year prospective study. Journal of Family Psychology, 8, 347–353. [Google Scholar]
  19. Duncan TE, Duncan SC, & Hops H (1994). The effects of family cohesiveness and peer encouragement on the development of adolescent alcohol use: A cohort-sequential approach to the analysis of longitudinal data. Journal of Studies on Alcohol, 55, 588–599. [DOI] [PubMed] [Google Scholar]
  20. Duncan TE, Duncan SC, Hops H, & Stoolmiller M (1995). An analysis of the relationship between parent and adolescent marijuana use via generalized estimating equations methodology. Multivariate Behavioral Research, 30, 317–339. [DOI] [PubMed] [Google Scholar]
  21. Duncan TE, McAuley E, Stoolmiller M, & Duncan SC (1993). Serial fluctuations in exercise behavior as a function of social support and efficacy cognitions. Journal of Applied Social Psychology, 23, 1498–1522. [Google Scholar]
  22. Elder GH (1980). Adolescence in historical perspective. In Adelson J (Ed.), Handbook of adolescent psychology (pp. 3–46). New York: Wiley. [Google Scholar]
  23. Elliot DS, Huizinga D, & Ageton SS (1985). Explaining delinquency and drug use. Beverly Hills, CA: Sage. [Google Scholar]
  24. Elliot DS, Huizinga D, & Manard S (1989). Multiple problem youth: Delinquency, substance use, and mental health problems. New York: Springer-Verlag. [Google Scholar]
  25. Farrell MP, Barnes GM, & Banerjee S (1995). Family cohesion as a buffer against the effects of problem-drinking fathers on psychological distress, deviant behavior, and heavy drinking in adolescents. Journal of Health and Social Behavior, 36, 377–385. [PubMed] [Google Scholar]
  26. Fisher DG, MacKinnon DP, Angling MD, & Thompson JP (1987). Parental influences on substance use: Gender differences and stage theory. Journal of Drug Education, 17, 69–85. [DOI] [PubMed] [Google Scholar]
  27. Forster B (1984). Upper middle class adolescent drug use: Patterns and factors. In Stimmel B (Ed.), Alcohol and drug abuse in the affluent (pp. 27–36). Binghamton, NY: Haworth Press. [DOI] [PubMed] [Google Scholar]
  28. Foshee V, & Bauman KE (1994). Parental attachment and adolescent cigarette smoking initiation. Journal of Adolescent Research, 9, 88–104. [Google Scholar]
  29. Gfroerer J (1987). Correlation between drug use by teenagers and drug use by older family members. American Journal of Drug and Alcohol Abuse, 13, 95–108. [DOI] [PubMed] [Google Scholar]
  30. Glueck S, & Glueck E (1940). Juvenile delinquents grown up. New York: Commonwealth Fund. [Google Scholar]
  31. Glynn TJ (1981). From family to peer: A review of transitions of influence among drug-using youth. Journal of Youth and Adolescence, 10, 363–383. [DOI] [PubMed] [Google Scholar]
  32. Hawkins JD, & Weiss JG (1985). The social-development model: An integrated approach to delinquency prevention. Journal of Primary Prevention, 6, 73–97. [DOI] [PubMed] [Google Scholar]
  33. Hill HA, Schoenbach VJ, Kleinbaum DG, Strecher VJ, Orleans CT, Gebski VJ, & Kaplan BH (1994). A longitudinal analysis of predictors of quitting smoking among participants in a self-help intervention trial. Addictive Behaviors, 19, 159–173. [DOI] [PubMed] [Google Scholar]
  34. Hops H, Duncan TE, & Duncan SC (1995, March). The relationship between parent and adolescent substance use: An analysis of longitudinal data via generalized estimating equation methodology. Paper presented at the 14th annual scientific session of the Society of Behavioral Medicine, San Francisco, CA. [Google Scholar]
  35. Hops H, Duncan TE, Duncan SC, & Stoolmiller M (1996). Parent substance use as a predictor of adolescent use: A six-year lagged analysis. Annals of Behavioral Medicine, 18, 157–164. [DOI] [PubMed] [Google Scholar]
  36. Jacobs GA, Jerome A, Sayers S, Spielberger CD, & Weinberg H (1988). Family smoking patterns and smoking among eighth and tenth grade students. Applied Psychology: An International Review, 37, 289–299. [Google Scholar]
  37. Jessor R, & Jessor SL (1977). Problem behavior and psychosocial development. New York: Academic Press. [Google Scholar]
  38. Johnson GM, Shontz FC, & Locke TP (1984). Relationships between adolescent drug use and parental drug behaviors. Adolescence, 19, 295–299. [PubMed] [Google Scholar]
  39. Kandel DB (1974). Inter- and intragenerational influences on adolescent marijuana use. Journal of Social Issues, 30, 107–135. [Google Scholar]
  40. Kandel DB (1984). Marijuana users in young adulthood. Archives of General Psychiatry, 41, 200–209. [DOI] [PubMed] [Google Scholar]
  41. Kandel DB, & Andrews K (1987). Processes of adolescent socialization by parents and peers. International Journal of the Addictions, 22, 319–342. [DOI] [PubMed] [Google Scholar]
  42. Kandel DB, Kessler RC, & Marguiles RZ (1978). Antecedents of adolescent initiation into stages of drug use: A developmental analysis. Journal of Youth and Adolescence, 7, 13–40. [DOI] [PubMed] [Google Scholar]
  43. Kandel DB, & Yamaguchi K (1985). Developmental patterns of the use of legal, illegal, and medically prescribed psychotropic drugs from adolescence to young adulthood. In Jones CL & Battjes RJ (Eds,), Etiology of drug abuse: Implications for prevention (pp. 193–235). Washington, DC: National Institute on Drug Abuse. [PubMed] [Google Scholar]
  44. Kaplan HB, & Johnson RJ (1992). Relationships between circumstances surrounding initial illicit drug use and escalation of drug use: Moderating effects of gender and early adolescent experiences. In Glantz M & Pickens R, (Eds.), Vulnerability to abuse (pp. 239–358). Washington, DC: American Psychological Association. [Google Scholar]
  45. Lau RR, Quadrel MJ, & Hartman KA (1990). Development and change of young adults’ preventive health beliefs and behavior: Influence from parents and peers. Journal of Health and Social Behavior, 31, 240–259. [PubMed] [Google Scholar]
  46. Lewinsohn PM, Rohde P, & Seeley JR (1996). Alcohol consumption in high school adolescents: Frequency of use and dimensional structure of associated problems. Addiction, 91, 375–390. [DOI] [PubMed] [Google Scholar]
  47. Liang KY, & Zeger SL (1986). Longitudinal data analysis using generalized linear models. Biometrika, 73, 13–22. [Google Scholar]
  48. McCullagh P, & Nelder JA (1983). Quasi-likelihood functions. Annals of Statistics, 11, 59–67. [Google Scholar]
  49. Melby JN, Conger RD, Conger KJ, & Lorenz FO (1993). Effects of parental behavior on tobacco use by young male adolescents. Journal of Marriage and the Family, 55, 439–454. [Google Scholar]
  50. Mermelstein RJ, & Borrelli B (1995). Women and smoking. In Stanton AL & Gallant SJ (Eds.), The psychology of women’s health: Progress and challenges in research and application (pp. 309–348), Washington, DC: American Psychological Association. [Google Scholar]
  51. Montemayor R (1982). The relationship between parent-adolescent conflict and the amount of time adolescents spend alone and with parents and peers. Child Development, 53, 1512–1519. [Google Scholar]
  52. Murray DM, O’Connell CM, Schmid LA, & Perry CL (1987). The validity of smoking self-reports by adolescents: A reexamination of the bogus pipeline procedure. Addictive Behaviors, 12, 7–15. [DOI] [PubMed] [Google Scholar]
  53. Murray M, Kiryluk S, & Swan V (1985). Relation between parents’ and children’s smoking behaviour and attitudes. Journal of Epidemiology and Community Health, 39, 169–174. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. National Institute on Drug Abuse. (1990). National Household Survey on Drug Abuse population estimates 1988. Washington, DC: U.S. Department of Health and Human Services. [Google Scholar]
  55. Needle R, McCubbin H, Wilson M, Reineck R, Lazar A, & Mederer H (1986). Interpersonal influences in adolescent drug use: The role of older siblings, parents and peers. International Journal of the Addictions, 21, 739–766. [DOI] [PubMed] [Google Scholar]
  56. Newcomb MD, & Bentler PM (1988). Consequences of adolescent drug use: Impact on the lives of young adults. Newbury Park, CA: Sage. [Google Scholar]
  57. Oregon Employment Division Research and Statistics. (1989). Is Oregon’s future at risk? A profile of Oregon’s youth. Salem: Author. [Google Scholar]
  58. Patterson GR, Reid JB, & Dishion TJ (1992). A social teaming approach: IV. Antisocial boys Eugene, OR: Castalia. [Google Scholar]
  59. Prinz RJ, Foster S, Kent RN, & O’Leary KD (1979). Multivariate assessment of conflict in distressed and nondistressed mother-adolescent dyads. Journal of Applied Behavior Analysis, 12, 691–700. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Rittenhouse JD, & Miller JD (1984). Social learning and teenage drug use: An analysis of family dyads. Health Psychology, 3, 329–345. [DOI] [PubMed] [Google Scholar]
  61. SAS Institute. (1985). SAS/IML user’s guide: Version 5 edition. Cary, NC: Author. [Google Scholar]
  62. Smart RG, & Fejer D (1972). Drug use among adolescents and their parents: Closing the generation gap in mood modification. Journal of Abnormal Psychology, 79, 153–160. [DOI] [PubMed] [Google Scholar]
  63. SPSS. (1990). SPSS reference guide. Chicago: Author. [Google Scholar]
  64. Stanton WR, & Silva PA (1992). A longitudinal study of the influence of parents and friends on children’s initiation of smoking. Journal of Applied Developmental Psychology, 13, 423–434. [Google Scholar]
  65. Steinberg L (1988). Reciprocal relation between parent-child distance and pubertal maturation. Developmental Psychology, 24, 122–128. [Google Scholar]
  66. Steinberg L (1990). Autonomy, conflict, and harmony in the family relationship. In Feldman SS & Elliot GR (Eds.), At the threshold: The developing adolescent (pp. 259–276). Cambridge, MA: Harvard University Press. [Google Scholar]
  67. Zeger SL, & Liang KY (1986). The analysis of discrete and continuous longitudinal data. Biometrics, 42, 121–130. [PubMed] [Google Scholar]

RESOURCES