Introduction

Over the past twenty years, there have been significant achievements in drug abuse prevention. Despite this success, drug use among American adolescents remains a leading preventable health risk. The most recent data reveal that 29% of eighth graders and more than half (53%) of 12th graders have tried an illicit drug (Johnston et al. 2007). One of the most influential innovations in substance abuse prevention theory has been the use of the risk-resiliency paradigm as a context for understanding adolescent problems (Hawkins et al. 1992). Although many factors have been found to contribute to adolescent drug use (Hawkins et al. 1992; Jessor and Jessor 1977; Newcomb and Felix-Ortiz 1992), peer and parental factors have been shown to be especially important both theoretically and empirically in predicting adolescent drug use initiation.

Few studies, however, have addressed how different age groups perceive parent and peer disapproval of drug use and how these perceptions influence intentions to use drugs in the future. For instance, a difference in relationships between parent and peer influence on drug use decision-making might occur when youth begin to strive for independence from their parents and place more emphasis on peer relationships. Attention to the relative importance of these risk and protective factors at different developmental points can have important implications for the design of prevention activities.

Parent Factors

Research indicates that parents and family dynamics are influential and can protect against adolescent substance use when parenting skills, parent-adolescent communication, and levels of warmth and affection are high (e.g., Brook et al. 1986; Dishion and Andrews 1995; Forhand et al. 1997; Kosterman et al. 1997; Patterson and Stouthamer-Loeber 1984; Shedler and Block 1990; Werch et al. 1991). Additional studies have established that adolescents’ alcohol, tobacco, and other drug (ATOD) use is more probable when their parents engage in ATOD use or when parents directly or indirectly convey that these behaviors are normative and common (e.g., Ary et al. 1999; Jackson et al. 1999; Kosterman et al. 1997; Stanton et al. 2000).

Peer Factors

In general, research indicates that adolescents are more likely to engage in behavior when they perceive that their peers will accept the behavior or will also undertake the activities. Therefore, peer norms are influential for problem behavior (Biglan et al. 2004). Individual characteristics such as the perception of the harm of substance use, self-esteem, sensation seeking, and a commitment to conventional social values are predictors of substance use (Hawkins et al. 1992). Peers, however, can play a role in increasing adolescent susceptibility to these risk factors. For instance, youth who perceive little harm in using substances may be easily persuaded to experiment when the behavior is modeled and the substances are available (Henry et al. 2005), and adolescents low in self-esteem might chose to go along with invitations to use when they believe it is the best way to gain peer acceptance (Ackard et al. 2006). Across gender and ethnicities, Simons-Morton (2004) found peer influence and drinking opportunities were positively correlated with drinking initiation.

The degree to which peer influence is responsible for adolescent substance use is controversial. Studies comparing the effects of peer influence and peer selection have claimed selection effects to be as strong or stronger than influence effects in accounting for comparable cigarette use between adolescents and their friends (Ennett and Bauman 1994). Although the study presented here does not directly address peer selection, it cannot be ignored when discussing peer factors with ATOD use. Similarly, the influence of parents (direct and indirect) on peer selection cannot be ignored when investigating peer influences. Thus, to better understand drug use intentions it is imperative to study peer and parental influence together.

Drug Use Intentions

Intervention or prevention programs aimed at adolescents are less likely to reduce use of substances for early initiators of drug use (Ellickson et al. 1993), and most prevention programs are planned for primary prevention. Therefore, they intend to target adolescents who have not yet initiated use (Webb et al. 1995). Although this target group has not yet experimented with substances, they may report developing interests regarding future intentions to use each substance (Andrews et al. 2003). Andrews et al. (2003) found that intention to use alcohol and cigarettes was related to subsequent experimenting with the substances, suggesting that intention to use is an early warning sign of future substance use. In the current study, we measure drug use intentions as a dependent variable because knowledge regarding the age at which young people develop interests to use drugs provides a guide to the timing and characteristics of interventions that can reduce later use.

Theoretical Model

Catalano and Hawkins’ (1996) Social Development Model (SDM) integrates the influences of parent and peer factors, along with school environmental factors, into a well-designed theoretical model. SDM hypothesizes that youth learn patterns of socializing behavior, whether pro-social or antisocial, from units of family, school, community, and peers. Moreover, individual behavior will follow the predominant behavioral norms and values held by those to whom the individual is bonded, whether they are antisocial or pro-social (Catalano and Hawkins 1996). This model posits that influence from family factors precedes and affects opportunities and influences emanating from peer cultures.

Hypotheses

Because early experimentation with substances is a precursor to regular substance use and concomitant problems, it is important to determine which factors are most likely to delay the onset of substance use. In order to design developmentally appropriate substance abuse prevention programs, it is essential to further our understanding of the nature of the parent and peer influence. In line with the SDM, parents directly influence behavior, attitudes and intentions, and also the peer relationships of their younger youth, whereas older adolescents expect and are often granted more autonomy and spend more time away from their parents and considerably more time with peers. Additionally, consistent with the shifting developmental roles of parents and peers, we expected that the perceptions of key behavioral norms and attitudes (i.e., disapproval toward substance use) held by those to whom the individual has the closest bond would influence drug use intentions.

The current study investigated three hypotheses. First, we hypothesized that perceptions of parent disapproval would be a stronger predictor of drug use intentions than perceptions of peer disapproval for youth in the sixth grade. Second, we expected that perceptions of peer disapproval would be a stronger predictor of drug use intentions than perceptions of parent disapproval for the eighth grade sample. Tests of these two hypotheses indicated the strength of relationships for parent and peer disapproval for use, with adolescent substance use decision making independently assessed at two grade levels. We also wanted to know if there were interaction effects. In particular, we wanted to know if the strength of the relationship between drug use intentions and disapproval (parent and peer) differed, depending on participant grade. Figure 1 visually depicts our third hypothesis: that the relationship between perceived parent and peer disapproval and drug use intentions would be moderated by grade level.

Fig. 1
figure 1

Parent × Grade and peer × Grade interaction models

Full size image

Method

This cross-sectional study made secondary use of data originally collected as part of a larger project during the 2002–2003 academic year. Parents received letters that described the program and informed them that their child would be asked to complete an anonymous survey prior to and after the program. Only those whose parent or guardian had not refused permission, and who themselves assented to participate, completed questionnaires. Using a standardized administration process, prevention providers administered paper and pencil surveys in classroom settings. Students completed the surveys independently and privately at their desks. Implementers carried a blank survey for use when students had questions. This practice assured anonymity. Students also created a secret code for linking pre- to post-tests, so that their identities would never be associated with their responses. The surveys took approximately 30 min to complete.

This evaluation design was approved by a local university’s institutional review board. The sixth grade data used in this study had been collected prior to commencement of the prevention curriculum. The eighth grade sample had been surveyed in the spring of 2003 as one form of comparison group for the original study and had not been exposed to any prevention curricula.

Sample

The sample for the present study consisted of 812 sixth graders and 837 eighth graders from four middle schools. The sixth grade sample, with a median age of 11, was comprised of 385 females and 427 males; the eighth grade sample, with a median age of 14, was comprised of 430 females and 407 males. The majority of respondents were White non-Hispanic (76%), and the remaining 24% were Hispanic (9%), Black (7%), or Other race (8%). Seventy-eight percent of the sample lived with two care providers. School lunch type was used as a socioeconomic status proxy; twenty-three percent reported eligibility for free lunch or partial subsidy.

Measures

All of the measures used in the current study have been used extensively and have been published in prior studies. Peer disapproval was measured using four items adapted from the Monitoring the Future Survey (Johnston et al. 2001) with a 3-point response format that ranged from 1: “approve” to 3: “disapprove.” In this scale, students were asked “how would your close friends feel if you… smoked marijuana regularly; tried cocaine or crack once or twice; had five or more drinks once or twice each weekend; and drove a car after having five or more drinks?” Higher scores indicated higher levels of perceived disapproval. A version of the scale using three of the four items obtained a Cronbach’s Alpha of .92 (Stevenson et al. 1998).

The parental disapproval scale, The Parental Attitude Toward Drug Use Scale, was adapted from the Communities That Care survey (Arthur et al. 2002). The original scale was reliable, with Cronbach’s Alphas = .77 and .69 for sixth grade males and females respectively and .83 and .79 for eighth grade boys and girls respectively (Arthur et al. 2002). This measure consisted of three items with a 4-point scale that ranged from 1: “very wrong” to 4: “not wrong at all.” The scale was recoded so higher scores indicated higher levels of disapproval. Questions asked, “how wrong do your parents feel it would be for you to…” drink various types of alcohol; smoke cigarettes; and smoke marijuana?

Intention to use drugs in the future was a five item scale with a 5-point response format that ranged from 1: “definitely not” to 5: “definitely will.” In this scale, students were asked “Do you think you will use any of these within the next year… cigarettes; various alcoholic beverages with a note ‘excluding use during religious ceremonies;’ marijuana or hashish, cocaine or other hard drugs, and inhalants.” Higher scores indicated increased intentions to use drugs within the next year. The original five-item scale, used in several studies, has demonstrated good internal consistency (Cronbach’s Alpha = .77; Macaulay et al. 2002).

Results

Psychometric Properties

Scale scores were examined for reliability, distribution (skewness and kurtosis), and homogeneity. The peer disapproval (Cronbach’s Alpha = .85) and parent disapproval (Cronbach’s Alpha = .85) scales were exponentially transformed due to unacceptable negative skewness; whereas the criterion, drug use intentions (Cronbach’s Alpha = .82) scale was log transformed due to unacceptable positive skewness. Alpha scores obtained for these measures resemble those in prior studies. Scatter plots indicated there was no violation of homoscedasticity.

Coding the Demographic Variables

The demographic variables were effects-coded for these analyses (see Table 1 for the coding protocol). Due to the small overall number of minority participants (24%), we combined participants who identified as Hispanic/Latino, Black and Other race into one group of Hispanic, Black and Other (HBO) and those who identified as White non-Hispanic into the second group. Lunch eligibility was a proxy measure for socioeconomic status (SES); therefore the terms “full pay” and “free or partial subsidy” are used to indicate higher and lower SES, respectively.

Table 1 Effects coding of variables
Full size table

Demographic Variability

Because the literature indicates differences in risk and protective factors for substance use across race and ethnicity, we performed crosstabs analyses to identify demographic differences among the schools. Of the four schools, one was comprised of more Black (χ 2 = 25.59, p < .001) and Latino (χ 2 = 44.57, p < .001) students. Another school had a higher proportion of students eligible to receive free or partial subsidy lunch (χ 2 = 114.89, p < .001). As a control for demographic differences, the demographic variables race, SES and gender were included in Block 1 of all the following regression analyses.

Testing Hypotheses

For these analyses, alpha was set at p < .01. Stepwise multiple regression was used to test the first two predictions. Block 1 of the analyses included the demographic variables of race, SES and gender. Perceived parent and peer disapproval toward substance use were entered into Block 2. The first analysis examined the sixth grade sample and the second analysis examined the eighth grade sample.

Table 2 shows that the independent variables together accounted for about 17% of the variance in intentions to use ATOD in the future (F (5, 772) = 30.58, p < .001). Using Cohen’s (1992) guidelines, an R 2 of .17 is between a medium and large effect. Change statistics indicated perceived parent disapproval (R 2 change = .123) contributed the most unique variance to the regression equation (12%), and perceived peer disapproval (R 2 change = .031) accounted for an additional 3% of the variance; the demographic variables accounted for a little over 1% of the variance (R 2 change = .012). With respect to our first hypothesis, parent disapproval (β = −.310) had the strongest relationship to intentions to use ATOD in the future, and peer disapproval (β = −.188) also had a significant negative influence over and above parent disapproval.

Table 2 Stepwise regression analysis predicting drug use intentions with demographic and perceived disapproval variables for sixth graders
Full size table

Table 3 shows that the independent variables together accounted for about 37% of the variance in intentions to use ATOD in the future (F (5, 785) = 90.46, p < .001). An R 2 of .37 is a large effect. Change statistics indicated perceived peer disapproval (R 2 change = .291) contributed the most unique variance (29%) to the regression equation, and perceived parent disapproval (R 2 change .069) accounted for an additional 7% of the variance; the demographic variables accounted for less than 2% of the variance (R 2 change = .016). With regard to our second hypothesis, peer disapproval (β = −.455) had the strongest relationship to intentions to use ATOD in the future, whereas parent disapproval (β = −.287) was also statistically significant over and above peer disapproval. Girls in the eighth grade had higher intentions to use drugs within the next year than their male peers did (β = −.136).

Table 3 Stepwise regression analysis predicting drug use intentions with demographic and perceived disapproval variables for eighth graders
Full size table

Both sixth and eighth grade samples were included in the final analysis. To determine if the independent variables were significant predictors of the dependent variable, we tested the models shown in Fig. 1 by conducting hierarchical multiple regression analyses. The moderator hypothesis posited an interaction between a focal independent variable (peer or parent disapproval) and the moderator (grade level). To eliminate problematic multicollinearity effects between the first order term and the higher order term, ‘peer disapproval’ and ‘parent disapproval’ were centered (Aiken and West 1991). To center the variables, scores were put into deviation score form by subtracting the sample mean from the individual’s scores (Cohen et al. 2003).

Three distinct steps were used in the analysis. Block 1 included the demographic variables race, SES and gender. Block 2 included the main effects of grade, peer disapproval and parent disapproval, and Block 3 included the interaction terms parent disapproval × grade and peer disapproval × grade.

Table 4 shows that the independent variables and the interaction terms together accounted for about 43% of the variance in intentions to use ATOD in the future (F (8, 1560) = 145.07, p < .001). An R 2 of .43 is a large effect. Gender, grade, peer disapproval, and parent disapproval were significant predictors of drug use intentions in the full model. The peer disapproval by grade interaction was significant, but the parent disapproval by grade interaction was not. The non-significant interaction indicates that sixth and eighth grade respondents did not differ in terms of the influence of perceived parent disapproval on their drug use intentions within the next year.

Table 4 Hierarchical regression analysis predicting drug use intentions with grade level as a moderator
Full size table

Because a significant interaction was found, we investigated the conditional effect to better understand the structure of the relationship (Aiken and West 1991). The simple slopes for the relationship between drug use intentions and grade level were calculated for the mean level of the centered variable peer disapproval and for one standard deviation above and below the mean (Aiken and West 1991). Participants with high perceptions of peer disapproval had lower intentions to use drugs in the future than those who had low or medium peer disapproval in both the sixth grade (t = −.490, p < .001) and the eighth grade (t = −18.924, p < .001). Beta is larger for the eighth grade participants (−.551) than the sixth grade participants (−.256) indicating greater influence on the dependent variable for the older youth.

Discussion

The goal of the present study was to investigate the relative strengths of the relationships of perceived parental disapproval and perceived peer disapproval with drug use intentions at two different points in youth development. We expected perceptions of parent disapproval toward substance use to be a stronger predictor of drug use intentions for sixth graders and perceptions of peer disapproval to be stronger for eighth graders. The results supported these hypotheses and were consistent with the social development model and reasonably consistent with prior studies (e.g., Catalano and Hawkins 1996; Gerstein and Green 1993; Webb et al. 1995).

We also were interested in whether the regression coefficient when regressing drug use intentions onto disapproval (parent and peer) for sixth graders was different from the corresponding regression coefficient for eighth graders. Results of the interaction analyses indicated no difference between the grades in influence of parent disapproval, but for eighth graders peer disapproval showed more influence on substance-using intentions (i.e., there was a moderating effect for grade on the role of peer influence, but not for the role of parent influence).

These findings are important in extending the theoretical understanding of the role of two major factors influencing adolescent substance use at two different levels of maturity. One inference is that a turning point in parent and peer influence on drug use decision-making may occur at a time when youth begin to strive for independence and become more closely attached to peers. Of special importance is the finding that, although the strength of peer influence is stronger when youth are in the eighth grade, parent influence still is a significant factor.

These results provide encouragement and hope for parents. As adolescents age, they believe a number of behaviors to be under their own personal control and not open to adult regulations (Goldstein and Tisak 2006). If adolescents do not consider others to possess authority over antisocial behaviors, they might be resistant to others’ attempts to regulate them through intervention and prevention efforts (Goldstein and Tisak 2006). With respect to the present findings, it appears that despite these beliefs, youth retain consideration of parental preferences at least through eighth grade.

The present findings can be usefully integrated with developmental patterns reported in other studies. For example, Webb et al. (1995) found stronger relationships to intentions to use alcohol for family factors in fifth graders, but peer factors were more strongly related to intentions for sixth graders. Sheppard et al. (1987) found parents are influential prior to experimentation with ATOD, but the peer group becomes important when a young person has already made the decision to use substances. Wood et al. (2004) found that parental influence moderated peer influences into late adolescence. Thus, parents can have an encouraging, lasting impact on their youth despite the increasing role of peer influence. Findings from this study indicate the importance of peer relationships for adolescents. Programs with peer involvement in the planning and promotion of alcohol free social activities (e.g., Komro et al. 1996) and peer leaders (e.g., Perry et al. 1993; 1996; St. Pierre et al. 1992) are encouraging.

Prior studies (e.g., Xueqin Ma and Shive 2000; Griffin et al. 2000) have shown that the relationships of substance use and risk and protective factors predicting use have varied in important ways between boys and girls. Griffin et al. (2000) found that boys smoked more cigarettes than girls in social situations, and girls smoked more at home when they were unsupervised. Girls also may use cigarettes as a weight loss device or to assist in maintaining their weight (Grigg et al. 1996). Ackard et al. (2006) found that girls reported higher prevalence rates than boys on low self-esteem (47.15% girls, 24.56% boys); and depression (63.52% girls, 33.35% boys), two predictors of substance use.

The most recent results from the Monitoring the Future survey reported that both eighth and tenth graders demonstrated fewer and smaller gender differences in the use of substances than in past years. The authors speculated that girls might tend to date and then try to be like older boys, who are in age groups considerably more likely to use drugs (Johnston et al. 2007). In the eighth grade sample and the combined sixth and eighth grade sample of the present study, there was a small but significant effect indicating that girls reported higher intentions to use ATOD than did boys. These findings indicate a need for prevention programs that take into account the variations between boys and girls in risk and protective factors and substance use behaviors. Self-esteem, for example, might be a significant factor for girls who date older boys - especially if they submit to substance use to be better received - but of less importance to boys.

Limitations, Future Directions, and Implications for Application

This study was limited in its use of cross-sectional data, so the possibility of historical differences between cohorts cannot be eliminated. Sampling a wider range of ages would help to clarify the developmental process. Disapproval data were based upon youth’s perceptions of their parents and peers and not directly elicited from these sources. Therefore, we cannot eliminate the possibility that youth who intend to use alcohol tend to perceive others as more accepting of substance use than they actually are (see Bauman and Ennett 1996). Parenting factors were not investigated independently to determine a connection between youth perceptions of parental disapproval and actual parenting behavior as influences on future intentions to use substances. Future studies would benefit from obtaining information about peer and parent attitudes directly from these sources and tracking participants over time to examine a longitudinal developmental process.

Parent/child connections, parental management, and parenting style are important factors to investigate in order to inform the development of prevention programs focusing on the family (Ackard et al. 2006; Mounts 2000, 2002, 2004). Direct investigation of external authority (parents and peers; e.g., Goldstein and Tisak 2006) could provide additional information on personal decision-making with respect to substance use or intentions to use in the future.

This investigation is of interest to prevention practitioners because it provides evidence that parents do have impact on their youth’s involvement in drug use decision-making, even as the effects of peers gains importance. Program architects can plan developmentally appropriate programs to target specific age groups and genders. By eighth grade, interventions designed to build affiliation with pro-social peers become increasingly valuable.