Purpose: The study examined the expectation that the joint distribution of perceived drinking norms (others' drinking) and actual levels of drinking (personal drinking) varied by race/ethnicity. Because it is well known that women and men differ in their drinking levels, the joint distribution may also vary by gender and may also vary by gender across ethnic groups. The discrepancy between drinking norm's and personal drinking may vary by quantity of drinking and it will be of interest to examine whether the effects of ethnicity and gender are robust when heavy drinking status is also considered. Clinical trail data indicate that challenging the misperception of drinking norms among college students can reduce heavy drinking, but success in culturally diverse college campuses may depend on whose norms are presented.

Methods: The sample was recruited from the student body on the campus of a large urban university in the southeastern United States. The sample protocol followed guidelines for institutions participating in the National College Health Assessment Spring 2000 survey conducted under the auspices of the National College Health Association. There were 1,236 students who participated in the survey and of those 1,201 (97.2%) provided completed questionnaires. Women were 63.6 percent of the sample. The average age of the participants was 24 years old (sd = 7.4). Of those who participated 21 percent identified as Whites, 18.4 percent as Blacks, 48.3 percent as Hispanic, and 12.3 percent as Asians. Survey participants were 86.6 percent undergraduates, 10.5 percent graduates, and 2.9 percent non-matriculated students.

Results: The results of the MANOVA showed that the main effect for gender (Wilk's lambda = 0.98, df = 2/868, p < 0.0001; eta2 = 0.02), the main effect for ethnic group (Wilk's lambda= 0.98, df = 6/1736, p < 0.008; eta2 = 0.01), and the main effect for heavy drinking status (Wilk's lambda= 0.75, df = 6/868, p < 0.0001; eta2 = 0.25) were statistically significant. None of the interaction terms were statistically significant. As exhibited by the magnitude of eta2, the differences introduced by gender (2%) and race/ethnic status (1%) while statistically significant are not as strong as are differences for heavy drinking status. Post-hoc analysis suggests that the mean of the joint distribution is different for Whites, Hispanics (higher) versus Blacks, Asians (lower)

Implications: While the degree of moderation in the joint distribution introduced by differences between race/ethnic groups is very modest, it is of theoretical and practice interest. According to self-regulation theory the success in reducing drinking by challenging the misperception of drinking norms rests on creating discrepancy. Maximizing discrepancy would be a function of presenting corrected drinking norms that are relevant and familiar rather than generic. It remains for additional empirical tests to corroborate the prediction derived from this study. That is, the efficacy of interventions for reducing college student drinking using a norms challenge approach will be increased by matching normative information on the basis of the individual's self-identified race/ethnicity.