Measurement Invariance of English and French Language Versions of the 20-Item Toronto Alexithymia Scale
Abstract. The alexithymia construct is commonly measured with the 20-Item Toronto Alexithymia Scale (TAS-20), with more than 20 different language translations. Despite replication of the factor structure, however, it cannot be assumed that observed differences in mean TAS-20 scores can be interpreted similarly across different languages and cultural groups. It is necessary to also demonstrate measurement invariance (MI) for language. The aim of this study was to evaluate MI of the English and French versions of the TAS-20 using data from 17,866 Canadian military recruits; 71% spoke English and 29% spoke French as their first language. We used confirmatory factor analyses (CFAs) to establish a baseline model of the TAS-20, and four increasingly restrictive multigroup CFA analyses to evaluate configural, metric, scalar, and residual error levels of MI. The best fitting factor structure in both samples was an oblique 3-factor model with an additional method factor comprised of negatively-keyed items. MI was achieved at all four levels of invariance. There were only small differences in mean scores across the two samples. Results support MI of English and French versions of the TAS-20, allowing meaningful comparisons of findings from investigations in Canadian French-speaking and English-speaking groups.