Objective: To evaluate the dimensionality of the CES-D in four samples of women at various life stages using exploratory and confirmatory factor analyses with one-, two-, three-, four-factor solutions commonly identified in prior research.

Methods: The samples consisted of 246 female university students, 430 low-income single mothers with young children, 503 adult female current and former smokers, and 651 older women living on farms. Each participant completed all 20 items of the CES-D. In addition to exploratory and confirmatory factor analysis, internal consistency was assessed with Cronbach’s alpha.

Findings: Scale reliabilities for the entire 20 items ranged from .86 to .92 across the four samples. However, the single-factor solution was not supported by confirmatory factor analysis [CFA] and nonloading items varied across samples. The three-factor solution (depressed/somatic, positive and interpersonal) emerged as optimal across all four samples. CFA factor loadings showed few items are reliable indicators (loadings > .65) of the underlying factor(s). Most reliable items were consistent across all samples, but the remaining 10 items have been found not to load on any factors in at least one analysis, either in the literature or in the present study.

Conclusions/Implications: The inability to find a single factor structure across samples may be due as much to the heterogeneity of items as to demographic differences. The strong internal consistency contrasted with the finding that the scale has multiple dimensions may be a function of differences across items in the ability to reliably measure depressive symptoms. Further work is needed to determine if these findings are replicable.

Back to Depression and Suicide
Back to 15th International Nursing Research Congress
Sigma Theta Tau International
July 22-24, 2004