Analysis and revision of the Depressive Experiences Questionnaire: examining scale performance as a function of scale length.

Revisions of the Depressive Experiences Questionnaire (DEQ; Blatt, D'Afflitti, & Quinlan, 1976) have failed to replicate the degree of orthogonality routinely observed with the original Dependency and Self-Criticism scales. Item performance on the DEQ was examined by computing correlation coefficients between factor-derived scores and unit-weighted composite scores for Dependency and Self-Criticism as a function of (a) the importance of individual items in predicting factor-derived Dependency and Self-Criticism scores and (b) scale length. Analyses identified sets of unit-weighted items that optimally preserve the psychometric properties of the original DEQ scales, including between-scale orthogonality, while reducing the number of items used to measure Dependency and Self-Criticism. Findings were replicated in college (N = 172) and clinical (N = 83) samples. Limitations of exploratory principal components and confirmatory factor analysis as tools for revising scales are discussed.