Gender differences in the factor structure of the Alcohol Use Disorders Identification Test in multinational general population surveys

Abstract Background Most gender-specific studies of the Alcohol Use Disorders Identification Test (AUDIT) have focused on gender differences in thresholds for hazardous drinking. This study examines gender differences in the factor structure of the AUDIT in general-population surveys. Methods General-population surveys from 15 countries provided 27,478 current drinkers’ responses to the AUDIT and related measures. We used single-group confirmatory factor analysis (CFA) to evaluate goodness-of-fit of three hypothesized models for responses to the AUDIT by men and women in each country. Bayesian Information Criteria (BIC) using a maximum likelihood robust (MLR) estimator was evaluated to identify the best fitted model. We then assessed factorial invariance within country surveys where fit indices were acceptable for both genders. Gender-specific internal consistency and concurrent validity were also evaluated in all 15 countries. Results CFA revealed that the fit indices of 2-factor or 3-factor models were consistently better than fit indices for a 1-factor model in 14 of 15 countries. Comparisons of BIC values indicated that the 2-factor solution was the best fitted model. Factorial invariance tests in data from 3 countries indicated that the factor loadings and thresholds of the AUDIT were invariant across gender. The internal reliability and concurrent validity of AUDIT and its subscales were acceptable in both genders. Conclusions A two-factor model best describes AUDIT responses across general-population surveys in 12 of 15 countries, with acceptable internal reliability and concurrent validity, and supports a gender-invariant structure in at least three of those countries.