Results of this research provide empirical substantiation for social norms being among the best predictors of college student drinking, at least with respect to typical weekly consumption. In this sample, most of the unique variance that was accounted for in drinking by demographics, social norms, motives, and expectancies was attributable to perceived descriptive norms and perceived approval of friends. Conversely, in predicting alcohol problems, coping motives were the strongest predictor, accounting for roughly half of all variance accounted by the entire set of predictors. Negative expectancies and favorable evaluations of negative alcohol effects were also strongly and uniquely associated with alcohol-related problems. Although all three norms variables were associated with problems, this was primarily because of their associations with alcohol consumption.
The positive association between negative expectancies and alcohol-related problems is somewhat counterintuitive, in that one might expect a person who expects negative things to occur as a function of consumption to drink less and have fewer problems. It may be that as long as positive expectancies outweigh negative expectancies, one continues to drink. In addition, given that favorable evaluation of negative effects was positively related to problems, these “negative” effects may not be perceived as negative to all students but may be viewed as desirable states or outcomes for some. Alternatively, and more consistent with these results, is the likelihood that students who drink problematically are most aware of the potential negative effects of drinking.
Additionally, both positive and negative expectancies have typically been associated with alcohol use (e.g., Adams and Nagoshi, 1999
; Bartholow et al., 2000
; Sher et al., 1996
). In the current findings, when examining correlations for expectancies with drinking and problems, positive expectancies were positively correlated to both drinking and problems, whereas negative expectancies were positively correlated only to problems. The finding that negative expectancies showed a robust and positive association with problems in the regression analysis whereas positive expectancies did not may be because of shared variance between positive and negative expectancies and with other variables in the model, as found in previous research (Jones et al., 2001
; Valdivia and Stewart, 2005
). Negative expectancies may explain more unique variance that is not accounted for by other predictors.
This research is consistent with previous suggestions that college student drinkers can be distinguished according to whether they drink primarily for social or enhancement reasons and/or positive reinforcement or as a means of coping and/or negative reinforcement (Mohr et al., 2001
; Stewart and Devine, 2000
; Stewart et al., 2001
; Wilkie and Stewart, 2005
). Most heavy drinking among college students can be attributed to social factors, including social norms and social contexts (Borsari and Carey, 2001
). Although drinking appears to be less motivated by using alcohol as a means of dealing with one’s problems, this kind of drinking is particularly problematic and, we suspect, more strongly linked with the development of longer-term problems and dependence.
A unique contribution of this study is the simultaneous evaluation of multiple predictors of college student drinking and related problems. As such, its results provide tentative implications for the choice of intervention targets and content. Fraternity/sorority membership was uniquely associated with greater consumption and problems, supporting efforts to provide prevention programs for these groups of students (Larimer et al., 2001
). All other things being equal, these results further suggest that strategies that are effective in changing perceived norms (e.g., Lewis and Neighbors, 2007
; Neighbors et al., 2004b
) might have a larger impact on drinking than strategies without a normative component.
Additionally, the results provide support for strategies that are effective in changing alcohol expectancies (Cruz and Dunn, 2003
; Darkes and Goldman, 1993
) as a way to impact alcohol-related problems more directly. Although the present findings support both social norms and alcohol expectancies as possible mechanisms of change for reducing alcohol consumption and problems, social norms-based interventions (as they are typically operationalized) can reach a larger audience with lower cost than expectancy challenge strategies as typically operationalized (i.e., delivered in person and often involving alcohol administration).
The results also support intervention strategies that directly address drinking motivations, such as functional analyses of benefits attributed to alcohol (e.g., enhancement and coping) and exploration of potential substitute behaviors or activities that can have the same effects (Baer et al., 1992
; Kivlahan et al., 1990
The present research also extends previous examinations of the relationships among norms, consumption, and alcohol-related problems (Benton et al., 2006
). The finding that consumption largely mediates the relationship between social norms and problems suggests that students who overestimate the drinking of their peers and who perceive their friends as more approving of alcohol have more problems because they drink more.
This research reiterates the influence that friends and parents have on drinking. Many existing intervention approaches for college student drinking focus on descriptive norms for typical students. Although this approach has shown considerable promise, these results suggest that interventions that also incorporate friends and parents might improve effectiveness. In the context of brief interventions, additional discussion of strategies for addressing friends’ expectations regarding alcohol use might be worthwhile. In these data, perceived parent approval of drinking had both direct and indirect unique effects on alcohol problems, suggesting the utility of more widely incorporating effective intervention approaches that already incorporate parents (Turrisi et al., 2001
Additional work is needed to identify and address alcohol use as a means of regulating negative affect and cognitions. Whether the addition of components such as coping-skills training to existing interventions would be sufficient for coping drinkers is unclear, and more work needs to be done specifically targeting this population.
Several limitations should be considered in interpreting the present results. The cross-sectional nature of the data limits our ability to draw causal inferences. Although the preponderance of previous literature supports the assumption that the predictors included here do have some causal influence on alcohol consumption and problems, it also seems clear that at least in some cases the temporal relationship is probably bidirectional. For example, although changes in norms have been associated with changes in drinking, the reverse is also true, suggesting, at least to some extent, that estimates of perceived norms may reflect behavioral justification (i.e., perceived norms, to some degree, probably represent personal drinking behavior; Kahler et al., 2003
; Neighbors et al., 2006a
). As noted above, this may also in part explain the strong relationship between negative expectancies and alcohol-related problems.
In addition, we included a relatively large number of predictors in our regression analyses. The resulting alpha inflation combined with the absence of specific hypotheses regarding the relative contribution of each predictor necessitates caution in interpreting the results. Our sample was restricted to male/female students who reported drinking five/four drinks on at least one occasion in the previous month, and results may not generalize to samples that include a large proportion of abstainers and light drinkers. In addition, the sample consisted of first-year students and may not generalize to older students who have had more extensive exposure to the college environment.
Another limitation is the disjunctive in reference groups between descriptive and injunctive norms. Ongoing research has begun to evaluate the importance of considering different reference groups in operationalizing social norms. In this research, we elected to use reference groups according to how they have been most often operationalized in the literature (typical students for descriptive norms and friends and parents for injunctive norms). Additional work is needed to tease apart the influences of different reference groups and their importance in evaluating the impact of both descriptive and injunctive norms.
Finally, the results of this research are limited to the constructs that were included and the manner in which they were assessed. The indicator of typical drinks per week has been empirically shown to be a relatively good indicator of alcohol consumption in comparison with other indicators of consumption (Borsari et al., 2001
; Smart et al., 2000
). Less research has focused on different measures of alcohol problems among college students (Read et al., 2006
), and additional work is needed to evaluate the extent to which different factors may be associated with different kinds of problems.
With respect to predictors of drinking outcomes, those we chose to include were based on our initial review of the literature and the primary aims of the larger study from which the data were drawn. In retrospect, we might have included additional variables that have also been associated with drinking among college students and young adults, such as attitudes (Benton et al., 2006
) and/or personality variables (e.g., impulsivity/disinhibition, extraversion/sociability, and neuroticism/emotionality; Baer, 2002
Despite these limitations, the present research offers a rare examination of the relative influence of several commonly included factors related to alcohol consumption and problems among college students. The majority of studies examining the etiology of problem drinking focus on a single factor or smaller set of factors designed to test narrow theoretical propositions. Although this strategy is appropriate for theory testing, it ignores the considerable overlap in variance accounted for by other predictors that are not included in the models and does not give any indication of which are the best predictors of drinking outcomes. Additional studies evaluating multiple factors related to consumption and problems using longitudinal designs are a logical next step. Ultimately, systematic examination of the relative contribution of factors may help us to better understand mechanisms of drinking and develop more effective interventions that precisely target factors most strongly associated with problematic drinking.