An examination of the internal structure of the Center for Epidemiological Studies-Depression Scale in two medical samples

The original research on the development of the Center for Epidemiological Studies-Depression Scale (CESD) postulated an internal structure of four subscales on the basis of an exploratory factor analysis (principal components). Although later research often reported different numbers of factors and patterns of loadings, this structure has generally been regarded as confirmed in the literature, even though most such studies used subjective methods for evaluating the fit. Recent objective confirmatory studies have replicated the four-factor structure in older adults and college students, although the exact models fitted do not always correspond exactly to the original model. Here two methods, confirmatory factor analysis and a simpler method based on Procrustes rotations, did not replicate the postulated subscale structure in two samples, one of 249 older adults and a second of 238 fibromyalgia patients. Discussion focuses on the methodology for confirming the subscale structure of psychological tests and the items of the CES-D that appear to underlie the difficulties in replication of the subscale structure.