Unidimensionality and Bandwidth in the Center for Epidemiologic Studies Depression (CES–D) Scale

• DOI: 10.1207/s15327752jpa8601_03
• Publication Date: 2006-01-25T17:34:30Z
• Authors: James P. Stansbury; L. Douglas Ried; Craig A. Velozo
• Keywords: Raw score; Classical test theory; Differential item functioning

In this study, we compared classical test theory (CTT) and item response (IRT) approaches in analyzing the Center for Epidemiological Studies Depression (CES-D) Scale (Radloff, 1977). Standard analyses, as well Rasch (1960) both revealed departures from unidimensionality a sample of 2,455 older persons responding to CES-D. Positive affect items scale performed poorly overall, their removal reducing scale's bandwidth only slightly. Modeling depression scores derived measures raw totals showed subtle but important differences statistical inference. The assessment depressive risk was slightly enhanced by using 16-item obtained results analysis dependent variable. Confirmatory factor parallel verified advantages removing positively worded items. IRT CTT techniques proved be complementary study can usefully combined improve measuring depression.