Using item response theory to improve alcohol dependence screening for african american and white male and female college students

Background: Alcohol-dependent college students rarely seek treatment, despite the availability of efficacious interventions. Overly simplistic screening procedures, failing to account for differential patterns across demographic groupings, exacerbate this when students in need of intervention are not identified. Objective: The criterion validity of the Alcohol Use Disorders Identification Test was evaluated, emphasizing sexrelated and race-related invariance. Methods: Item Response Theory was used to evaluate data (collected: 2008 to 2009) from 1500 lifetime-drinking, southeastern US college students, across sex and self-identified race (ie, White, non-Hispanic and Black/African American). Alcohol dependence criteria were used in the sensitivity/specificity analyses. Results: Item Response Theory results led to item 9's removal (for race-related noninvariance) and a cut score of 4 (problems/dependence subscale) offered more balanced intergroup scoring. Discussion: Optimal college student screening may differ by sex and race.