Do We Adequately Measure the Alcohol-Specific Behavior of College Students? Measurement Implications From a Latent Class Analysis

Although college students are recognized as being at-risk for heavy alcohol use, the patterns and intensity of alcohol-specific behaviors (e.g., drinking game participation, BAC) are not well understood. Given heterogeneity in alcohol use, it is important to examine patterns of alcohol-specific behaviors to determine types of drinkers that may be at increased risk of experiencing consequences. This presentation aims to identify discrete alcohol subgroups that are meaningfully distinguished by their patterns of engagement in specific behaviors.

Daily self-report data were collected from a multi-ethnic sample of college students during the fall of students’ senior year (n = 557, 56% female). Participants completed up to 14 consecutive days of diary reports (75% completed all 14 days). Latent class analysis was used to identify subgroups of college students based on eight indicators: any weekend drinking; any Thursday drinking; any drinking on Sunday through Wednesday; engaging in drinking games; pregaming; getting drunk; consuming at least 5 (men)/4 (women) drinks within two hours; and maximum estimated blood alcohol content (eBAC) in the 14 days.

Four latent classes were identified: Nondrinkers (27% of the sample), Frequent and Heavy Drinkers (38%), Social Weekend Drinkers (18%), and Intoxicated, Assorted Drinkers (17%). The Heavy and Frequent Drinkers had very high probabilities of endorsing the seven alcohol use behaviors as well as a maximum eBAC above .16%. The Intoxicated, Assorted Drinkers evidenced considerable heterogeneity in their drinking behaviors, with 99.5% reporting drinking on Friday/Saturday night but 65% reporting pregaming, 53% reporting drinking game participation, and 54% reporting consuming 5+/4+ drinks within two hours. Half of the individuals in this class reported a maximum eBAC of .08% to .16% with an additional quarter reporting a maximum eBAC above .16%, suggesting an increased risk for alcohol-related consequences. AUDIT scores, indicating alcohol use problems, significantly related to class membership, thus validating the latent classes. Gender was not related to class membership.

These findings demonstrate the benefit of measuring college drinking in a holistic way to identify critical behavior patterns that indicate heightened risk for related problems. These findings suggest that the use of single, commonly used measures of drinking behavior, such as consuming 5+/4+ drinks within two hours, may not be optimal for identifying individuals who are at greatest risk. The latent classes identified here suggest an opportunity to tailor interventions to subgroups of college students based on their entire profile of use. Future work will extend this study to examine transitions in behavior throughout eight semesters of college.