Parenting influences on adolescent alcohol use

Research report No. 10, 2004

6. Parent and peer influences compared

It is often proposed that in adolescence, there is a shift away from parental influence, with peers taking on greater importance. While this shift has been demonstrated in research, it is erroneous to assume that parents have no or little influence on their adolescents. Rather, research has shown that both parents and peers have an influence on the developing adolescent.

The influence of parents and peers

We have seen that for typically developing adolescents, the relationship with parents is likely to remain strong. Evidence suggests that well-adjusted adolescents tend to have high quality relationships with their peers (Armsden and Greenberg 1987). Research on decision-making has shown that whether parents or peers are more important depends on the quality of the parent-adolescent relationship (Noller 1994; Wilks 1986).

Generally, parents are considered more vital in decisions concerning vocation or money, and peers are considered more important in decisions about clothes, social activities, and entertainment. According to Noller (1994), when everyday problems arise, adolescents prefer to discuss these with their close friends, rather than parents. While a complete review of the research on peer influence on adolescent alcohol use is beyond the scope of this review, a summary of key studies follows.

The role of peers in adolescent alcohol initiation is generally accepted as crucial. In Australia, the high prevalence of alcohol use among peers of adolescent drinkers has been demonstrated by the NDSHS survey (AIHW 2002); with threequarters of male (73 per cent) and female (78 per cent) recent drinkers reporting that all or most of their friends consume alcohol. This was compared to the abstainers or ex-drinkers where only 21 per cent of males and 30 per cent of females reported that their friends drink alcohol (AIHW 2002).

This review has adopted a bi-directional framework in which to consider parent-adolescent relationships. The research also shows that this bidirectionality is an important element in adolescent-peer relationships. This bidirectional peer selection/peer socialisation process has been demonstrated in a longitudinal study using latent growth modelling (Curran, Stice and Chassin 1997). In this study, an individual's earlier levels of alcohol use were predictive of later changes in their peers' use; conversely levels of peer use at time one were also predictive of later changes in the adolescent's use. This suggests that once adolescents commence drinking they may seek out peers who also drink alcohol. Bamberg and colleagues (2001) propose that parents influence this selection of peers because adolescents who have parents who frequently smoke or drink alcohol are more likely to choose to associate with peers who display these behaviours too.

Recent research has begun to look at the similarities amongst peers, specifically close friends, rather than the wider peer group. In a longitudinal study of 1028 students from Grades 6 to 10, only the alcohol use of a close friend predicted the initiation of alcohol use, and the alcohol use of the wider peer group was not significant (Urberg, Degirmencioglu, and Pilgrim 1997). However, both close friends and peer groups predicted drinking to intoxication (Urberg et al. 1997). Of interest is that once initiation has occurred, there appears to be a change in the peer context, in that those who have tried alcohol are more likely to know fellow drinkers. Thus it appears that peer selection and influence are complementary processes.

The importance of peers and parents was compared in a three-year longitudinal sample of adolescents, who were aged 13-16 years at time one. Reifman et al. (1998) measured progression to heavy drinking and found that friend's drinking was the most significant predictor (odds ratio 1.62), and parental monitoring was the only other significant predictor (OR 0.71). In this study maternal disapproval, rule setting, and parents' drinking were not significant.

Similar findings were reported by Rai et al. (2003) in a comparison of peer and parental influence using data from six studies conducted across ten years. Rai and colleagues found the odds ratios for alcohol use among adolescents whose peers were involved in risk behaviours ranged from 1.37 to 1.92 over the six samples, with an overall average of 1.62. The odds ratios for parental monitoring were 0.35 to 0.69 (indicating that increased monitoring was associated with less alcohol use), with an overall averaged odds ratio of 0.65 (the odds ratios obtained for these parent and peer influences cannot be directly compared, as odds ratios smaller than 1 have a possible range of 0 to 1, and are therefore measured on a different scale to odds ratios greater than 1, which can range from 1 to infinity). It should be noted that since parents often influence adolescents' choice of peers, these studies to a certain extent underplay the extent of parental influence.


These key studies have shown that the effect of peers mediates the influence of parenting factors on adolescents' alcohol use. Peer effects become particularly powerful when parent-adolescent relationships are of poorer quality. The influence of peers is thought to occur through peer modelling, peer pressure, or association with alcohol using peers. However, direct connections between parental monitoring and adolescent alcohol use remained after peer influences were taken into account. The interaction of peers and parents is a complex issue that cannot be easily separated, and thus future research or interventions must consider the interaction of each.

The mediational model

Further evidence of the inter-connected influence of family and peers is shown in the seminal work of Dishion and colleagues, using multiple types of respondents (e.g. adolescents, parents), and multi-method measures (for example, observations and questionnaires). Dishion and colleagues (1999) measured the initiation of alcohol use and found that family, peer, and child characteristics were all connected to initiation of use. Socio-economic status and parental substance use were also significant contextual predictors of the onset of adolescent alcohol use. Statistical modelling showed that the effect of socioeconomic status was mediated through family management (parenting behaviours such as monitoring and limit setting), and in turn family management was mediated through deviant peer associations. Furthermore, after including child characteristics at age 9-10 years, only the boys' antisocial behaviour at this younger age predicted early onset of alcohol use. The authors argued that their research demonstrated the ecological framework of adolescent development, and that more proximal predictors mediated the effects of less proximal predictors.

Other longitudinal work using the Social Development model as a structure for understanding the importance of family, school, social, and community factors supports an ecological view of adolescent development. Guo et al. (2001) measured the alcohol use of 755 adolescents who were followed from age 10-21 years. Using odd ratios adjusted for internalising and externalising behaviours at age ten years, these researchers found that the most significant predictors of alcohol abuse originating from age 10 were associations with deviant peers (OR 1.5), time with antisocial friends (OR 1.3), and alcohol use amongst best friends (OR 1.3). Furthermore, high monitoring as well as clearly defined rules at ten years of age predicted lower alcohol abuse and dependence at 21 years. The size of the odds ratios increased across time and with closeness to the outcome at 21 years, with more powerful associations found for measures taken at age 14 and 16 years.


Mediational models have shown that there are bi-directional influences between adolescents and others. Adolescents are influenced by parents, peers, the community and society; but conversely their behaviour has a reciprocal influence on their families and peers too. The research on parenting factors does not always account for this bi-directionality, or for the existence of multiple spheres of influence.

Research Report 10: Contents | Previous | Next