Abstract: | Nonparametric item response models were used to investigate the psychometric properties of the Beck Depression Inventory in depressed outpatient (N?=?648) and nonpatient college (N?=?1,182) samples. The authors examined item bias between men and women as a function of severity of depression and distinguished group mean differences from item bias. They also assessed the appropriateness of the weights assigned to response option categories. In both depressed outpatient and nonpatient college samples, most items and options discriminated individuals differing in degree of depression, and little gender item bias existed between men and women. Estimated response option weights did, however, correspond more closely to a priori weights in the depressed outpatient sample than in the nonpatient college sample. The greatest amount of bias was observed in Item 14, assessing distortion of body image. (PsycINFO Database Record (c) 2010 APA, all rights reserved) |