Abstract
OBJECTIVE: The Center for Epidemiologic Studies Depression Scale (CES-D) is a widely used instrument for measuring depressive symptoms. Though conventional factor analytic evaluations supported the use of four sub-scales for the CES-D, existing studies have yet to adopt the bi-factor analytic approach in psychometric assessment of the 20-item inventory. The present study aimed to apply both confirmatory factor analysis and exploratory bi-factor analysis to evaluate the dimensionality of the CES-D. METHODS: Current scoring practice of the CES-D (single-factor, four-factor, and second-order models) was tested using confirmatory factor analyses in a sample of 706 Chinese persons with insomnia and depressive symptoms. As an alternative, exploratory bi-factor analysis was conducted to examine the utility of the general depression factor and specific factors. RESULTS: Existing measurement models on the CES-D did not provide an adequate model fit to the data in terms of model fit indices and discriminant validity. The bi-factor model revealed a general depression factor that accounted for the majority of the item variance. The three specific factors (somatic symptoms, positive affect, and interpersonal problems) provided little unique information over and above the general factor and plausibly represent a methodological artifact rather than a substantive factor. CONCLUSION: The present study demonstrated empirical support for the bi-factor model as a realistic representation of the underlying structure of the CES-D. Researchers and clinicians are better served by simply using a single measure of depression.