Abstract

Measurement invariance is typically assumed when assessing drinking-related constructs across distinct groups of respondents. However, measurement properties of motivations related to mood maintenance and stress relief may differ in those experiencing symptoms of depression or anxiety. Invariance of social and coping drinking motives were explored with a sample of 4133 junior enlisted Air Force and Navy personnel. Measurement did not differ in those with depression symptoms. In contrast, those with anxiety symptoms differed in measurement of both motives. The impact of non-equivalence was demonstrated with a mediation model in which anxiety and depression predicted drinking motives, which in turn predicted heavy drinking. Incorporation of the partial invariance of the social motives factor attenuated the estimate of the mediated effect of social drinking motives by almost half compared to the estimate with invariance assumed. These results suggest that lack of measurement invariance could seriously bias or alter conclusions from tests of theoretical models and highlight the need for researchers to carefully consider the measurement properties of their constructs prior to model estimation.

Copyright (c) 2009 Elsevier Ltd. All rights reserved.