The Beck Depression Inventory (BDI) is widely used to document the pre
valence of depression in sufferers of chronic pain and in research des
igns about this population. Williams and Richardson (1993) initially p
osed the question, ''What does the BDI measure in chronic pain?''. Res
ults from their study found 3 independent constructs, which differed s
omewhat from those obtained in analyses with other non-pain subsamples
. In our reappraisal of the question, we used confirmatory factor anal
ytic procedures to assess the dimensionality of the BDI. Specifically,
we questioned whether a hierarchical model in which a second-level de
pression construct underlies 3 constituent first-level constructs is r
easonable for the data. Our results, based on a sample of 247 chronic
pain patients, corroborated the adequacy of this model. The first-leve
l constituent constructs were labeled Negative Attitudes/Suicide, Perf
ormance Difficulty, and Physiological Manifestations and were conceptu
ally similar to first-level constructs obtained with other subsamples.
Furthermore, external psychological measures and selected questionnai
re items were used to assess convergent and discriminant validity of s
cales operationalizing the factor-analytically derived constructs. Wit
h these analyses, we clarify the constituents of depression as measure
d by the BDI. The findings from this study have implications for more
refined epidemiologic and clinical research with chronic pain patients
.