Cross-cultural comparability of DSM-IV-criteria for alcohol dependence – Findings from classical and modern test theory

Aims: To compare national prevalence rates it is necessary to have an assessment instrument that is “culture-fair“. Recent research showed that especially DSM-IV criteria for alcohol dependence are differently understood across countries. Item-Response-Theory (IRT) provides a framework to evaluate the cross-country comparability of psychological instruments guaranteeing invariant measurement across samples. Method: Data was provided by the project “Standardizing Measurement of Alcohol Related Troubles“ (SMART). DSM-IV-criteria for alcohol dependence were available for N=1889 individuals in nine European countries (Estonia, Czech, Germany, Italy, United Kingdom, Finland, Spain, Poland, Hungary) with about equal sample sizes of 200 individuals each. Exploratory factor analysis and 1-PL and 2-PL-model were performed. Differential Item functioning (DIF) was applied for both item difficulty and item discriminatory power. Results: Reliability was satisfactory for most countries and the total sample. Exploratory factor analysis revealed a one-factor-structure in five out of nine countries and the total sample. The 1-PL-model better fitted the data than the 2-PL-model for the total sample and six countries. DIF-analysis showed that item difficulties are highly linear (proportional) across countries and different split criteria. Item discrimination parameters were not proportional across countries. The sum-scores of criteria do not represent equal distances on the latent dimension. Conclusions: Findings indicate a scale with good reliability and good but not perfect homogeneity. None of the criteria showed consistent DIF indicating good cross-cultural comparability. Although adaptation to the Rasch-model is much higher than expected, scale dignity (additivity) cannot be presumed for the sum of criteria. However, these findings can be used for more economic testing in future surveys, as (adaptively) not all items always have to be used.

DIF-analysis - IRT - alcohol dependence