Factor structure of the CES-D and measurement invariance across gender for low-income Puerto Ricans in a probability sample.

OBJECTIVE: The authors of this study aimed to evaluate 2-factor structures for the Center for Epidemiologic Studies Depression Scale (CES-D) reported in the literature to determine which one proves to be a better fit with the data on low-income Puerto Ricans living on the island.
METHOD: The sample consisted of 3,504 civilian noninstitutionalized Puerto Ricans, ranging in age from 18 to 64 who were living in low-income areas of Puerto Rico. A confirmatory factor analysis (CFA) was implemented, and model fit was assessed with the Tucker-Lewis Index (TLI), the comparative fit index (CFI), and the root-mean-square error of approximation (RMSEA).
RESULTS: The CFA results obtained for the Radloff (1977) and Guarnaccia, Angel, and Worobey (1989) models show that neither model obtained adequate fit indexes. Exploratory factor analysis revealed that the 2-factor structure not only obtained better fit indexes (TLI = .948, CFI = .955, RMSEA = .048) but also reflected a better conceptual organization of the factors (basically divided into 2 components: Negative Affect and Positive Affect). When we assessed the model fit by gender, the model consistently fit the data for the female subsample but not for the male one.
CONCLUSIONS: Results confirm the need to address the issues related to measurement equivalence for specific cultural and language gaps in assessment of depressive symptoms. Lack of factor invariance across gender also raises the question of the appropriateness of this measure for the male subsample and the need for further research in the manifestation of depression in this group.