Replicated factor structure of the Beck Depression Inventory

The factor structure of the Beck Depression Inventory (BDI) was examined in an initial sample of 407 patients and replicated in a sample of 370 patients, using principal components analysis and varimax rotation. The entire sample was 98% male, with ages ranging from 22 to 88 years. The incidence of alcoholism was 72%, and the incidence of major mental illnesses was 17%. The stable dominant first factor (cognitive) accounted for 67 to 81% of the common variance, with a correlation of .94 between the factor loadings in the initial and replication analyses. The unstable second factor (vegetative) accounted for 15% to 19% of the common variance, with a correlation of .58 in item loadings. Consequently, the BDI appears to measure the cognitive aspects of depressive severity in a global fashion, as Beck originally intended. Approximately half of the items contribute very little useful predictive information. The fact that BDI total scores are not strongly related to the traditional vegetative symptoms of depression used in psychiatry is an artifact of the original method used to construct the test and is not a statement about the fundamental characteristics of depression.