Measurement invariance of the alcohol use disorders identification test: Establishing its factor structure in different settings and across gender

Abstract

IntroductionThe Alcohol Use Disorders Identification Test (AUDIT) is an internationally well-established screening tool for the assessment of hazardous and harmful alcohol consumption. To be valid for group comparisons, the AUDIT should measure the same latent construct with the same structure across groups. This is determined by measurement invariance. So far, measurement invariance of the AUDIT has rarely been investigated. We analyzed measurement invariance across gender and samples from different settings (i.e., inpatients from general hospital, patients from general medical practices, general population). MethodsA sample of n = 28,345 participants from general hospitals, general medical practices and the general population was provided from six studies. First, we used Confirmatory Factor Analysis (CFA) to establish the factorial structure of the AUDIT by comparing a single-factor model to a two-factor model for each group. Next, Multiple Group CFA was used to investigate measurement invariance. ResultsThe two-factor structure was shown to be preferable for all groups. Furthermore, strict measurement invariance was established across all groups for the AUDIT. ConclusionA two-factor structure for the AUDIT is preferred. Nevertheless, the one-factor structure also showed a good fit to the data. The findings support the AUDIT as a psychometrically valid and reliable screening instrument.