In a random sample of 891 Brazilian adolescents 15 to 19 years of age, half of the participants reported alcohol consumption during the past year. One-third of the participants reported binge drinking. Having their most important groups of friends from school was associated with a higher likelihood of binge drinking. Individual SES was not associated with binge drinking, but residential SES was associated with binge drinking.
The prevalence of alcohol consumption in our study in Belo Horizonte, Brazil, was similar to previous studies conducted among adolescents in Brazil [5, 56, 57]. However, some studies have reported lower prevalence compared with the present survey [32, 58]. The prevalence of binge drinking in the present study (36%) was similar to that reported by studies from other countries (27%–47%), and our results are comparable to figures reported previously in Brazil (35%) . Although the comparison of binge drinking prevalence between countries is somewhat complicated by differences in standard drink sizes and other definitions, the high rates found for alcohol consumption among adolescents represent a significant public health concern. Binge drinking was associated with being an adolescent male, and this result corroborates findings in previous studies [5, 56].
Some studies of alcohol consumption and binge drinking in adolescents have shown a statistically significant association with socioeconomic status (family income, personal income and unemployment), but conflicting results have also been reported in the literature. Some studies confirmed that alcohol consumption is negatively associated with socioeconomic status [59, 60], while others found positive associations between higher socioeconomic status and high-risk drinking [56, 61] as well as low socioeconomic conditions and high-risk drinking [30, 32]. Consistent with other studies [55, 59–61], our results indicate a higher prevalence of binge drinking among adolescents who live in less vulnerable areas. Also consistent with our results, Humensky (2010)  indicated that a higher socioeconomic status in adolescence, as measured by parental education and household income, is associated with higher rates of substance use, particularly binge drinking. Because the SVI also includes family income and educational backgrounds, these results consider the disposable income of those who live in less vulnerable areas where adolescents are more likely to buy their own alcohol and consume it at rates and in quantities without guidance. According to a longitudinal study developed in a representative sample of adolescents in Sweden with multilevel analysis of a period with large economic changes, unemployment is statistically associated with binge drinking . A decrease in incomes leads to lower consumption of all goods, including alcohol because adolescents have to allocate the resources they can spend on different goods .
In this paper, we estimated the predictors of adolescent drinking behavior to identify the role of social friendship networks on binge drinking. Adolescents who reported that the majority of their groups of friends/close friends were from school rather than from church or hobbies/other activities had higher odds of binge drinking. This result corroborates finding in the National Longitudinal Study of Adolescent Health in the US, suggesting that an increase of 10% in the proportion of classmates who drink will increase the likelihood of drinking among peers by 4% . With similar results, a classroom-randomized controlled trial that compared control classes with those receiving an evidence-based substance use (including alcohol) prevention program with 541 adolescents in the US indicated that students with classroom friends who used substances were more likely to increase their own use . However, our study revealed that differences emerged when analyses were conducted separately by gender. In our study, the adolescent females who reported that the majority of their groups of close friends were from school had a higher likelihood of binge drinking compared with those who reported that their best friends were from church or from hobbies and other activities. However, no statistically significant association was found for male adolescents. These results are consistent with previous evidence of the importance of friends on alcohol drinking in adolescence. At least two studies have demonstrated that peer drinking has a greater impact on the average amount of alcohol consumed by girls consumed versus the amount consumed by boys [53, 62]. Furthermore, girls are more likely to socialize with and be persuaded by friends to drink alcohol compared with boys .
When we compared the adolescents who reported that the majority of their groups of close friends were from church with the adolescents who reported that the majority of their groups of close friends were from hobbies and other activities, we found that having friends from church was associated with less binge drinking. This result is similar to that found using other surveys, which suggest the association between attending church/religiosity and a lower likelihood of binge drinking [5, 20, 63, 64]. However, because we did not collect information on the adolescents’ religious beliefs or church attendance frequency, we were unable to tease out the relative contributions of having friends in church-based networks versus religious participation or religious beliefs.
Studies developed by Valente et al. (2007) ; Jamison and Myers (2008) ; Kiuru et al. (2010) , previously examined the influence of the peer group on alcohol consumption in adolescents, but they were limited to investigating just the school based peer network. Other studies investigated the influence of the peer group on alcohol consumption in adolescents based on their religiosity or church attendance [5, 63]. We are not aware of a previous study that compared the different type of groups of friends from different contexts in relation to alcohol consumption by adolescents. It is important to consider friendship networks in multiple contexts when examining peer influences on adolescent behavior. Our study was not restricted to nominated peer groups, so it was possible to compare groups of friends from different contexts such as church, school, family and hobby groups. Comparing the different sources of peer connections, we found that adolescent binge drinking was strongly associated with school-based friendship networks relative to church-based networks.
Throughout the course of a lifetime, friendships can direct development through support, modeling, and assistance, but their significance is heightened during adolescence . Peers have the potential to both positively and negatively influence behavior, depending on the behavior endorsed by the peer group .
Public health interventions targeted to the peer network might be more cost-effective than previous estimates have suggested because the health-promoting behavior of one person may spread to others via social networks [10, 12]. The extent to which the school is a functional community with supportive social relationships, social participations in school activities, and shared norms, goals, and values may also moderate the individual risk of initiating adverse health behaviors such as high alcohol consumption . Our results showed that students having their most important group of friends coming from church seems to be a protective factor with respect to binge drinking. The protective effect of social capital might reflect the effect of norms and social controls on curtailing deviant and dangerous alcohol consumption in communities in which individuals are more bonded to each other . Thus, our findings underscore the importance of deeply exploring how the context can determine alcohol consumption.
`Our results should be interpreted while considering some limitations. Because this study of the association between type of friendship network and binge drinking used cross-sectional data, two principal threats to causal inference should be considered: homophily and unobserved confounding. Homophily refers to the idea that individuals select different types of social network connections based partly on their own behavior preferences. For example, some students may enjoy associating with others based on their shared preference for observing religious proscriptions against drinking alcohol. In other words, the associations that we observed may have been driven by the tendency for “birds of a feather to flock together”, rather than by the network characteristics per se (e.g., social norms that influence and regulate behaviors within certain types of social networks). Secondly, our finding may reflect the influence of unmeasured confounding factors that affect both patterns of friendship networks and drinking behavior. For example, the strictness of school policies against drinking may influence both the formation of different friendship connections as well as the prevalence of binge drinking. Because we did not collect information on variables such as school policies, we cannot rule out the possibility that unmeasured or unobserved factors could have led to the patterns we observed.
We did not measure and control for variables such as “parental supervision” in our survey. Thus, if students who associated with peers from school were also those who were less likely to be supervised by their parents, the association between having school friends and binge drinking could have been confounded by such omitted variables.
We assured the participants in this study that their responses would remain anonymous. However, there are some limitations associated with our measures of behaviors and negative outcomes because they rely on self-reports of alcohol consumption and binge drinking. It is possible that the actual amount of alcohol consumption and binge drinking could have been either under-reported or exaggerated. Furthermore, our sampling approach did not include individuals who were excluded from or had otherwise left school-based education, and the Vulnerability Index (SVI) was calculated on an ecologic level and not on an individual basis.
The level of social capital to which an individual may gain access through their social networks depends on the structural characteristics of the networks as well as on the amount of social capital that other individuals in the network possess. Although they may be positive, friendship networks can also be viewed as negative, especially with regard to binge drinking during adolescence. Corroborating our results, at least three other studies have also found an association between peer influence and alcohol consumption and binge drinking [10, 11, 13].
We were unable to obtain information on the size, density, quality, or proximity of the friendship network or the normative drinking behaviors within those networks. Further studies are needed to investigate the importance of the detailed characteristics of friendship networks (size, density, quality, and proximity), parental control, religious beliefs and participation. Another important issue for future research is to measure the perceived norms among different types of friendships. Ideally, studies should measure the subjective norms surrounding drinking behavior within school-based versus church-based networks (e.g., “How frequently do my friends from school/church engage in binge drinking?” and “Do my friends (in school or in my church group) approve/disapprove of drinking?”).