Examining scale discriminability in the BDI and CES-D as a function of depressive severity.

Abstract

Scale discriminability is the ability of a measure to discriminate among individuals ordered along some continuum, such as depressive severity. We used a nonparametric item-response model to examine scale discriminability in the Beck Depression Inventory (BDI) and Center for Epidemiclogic Studies Depression Scale (CES-D) in both college and depressed outpatient samples. In the college sample, the CES-D was more discriminating than the BDI, but a standard CES-D cutoff score of 16 overestimated the likely prevalence of depression (45%). The CES-D may be more effective than the BDI in detecting differences in depressive severity in college students but may be less specific. In the depressed outpatient sample, the CES-D was again more discriminating than the BDI. The superior scale discriminability of the CES-D offers one explanation for its poorer specificity in college samples. Endorsing many items that discriminate at low levels of depressive severity can result in scores that exceed a cutoff criterion.The Beck Depression Inventory (BDI; Beck, Shaw, Rush, & Emery, 1979) and the Center for Epidemiologic Studies Depression Scale (CES-D;Radloff, 1977) are among the most frequently used and well-validated self-report measures of depression (for reviews, see Beck, Steer, & Garbin, 1988;Orme, Reis, & Herz, 1986). Although both measures treat depression as a continuous variable, the BDI and the CES-D were initially designed to assess depression in different populations. The BDI was initially intended as a measure of depressive severity for individuals already diagnosed as clinically depressed (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961), whereas the CES-D was originally constructed as a measure of depressive severity for adults in the community (Radloff, 1977). Despite this difference, the BDI and CES-D have been used extensively in both clinical and nonclinical populations, and both have been used to estimate the prevalence of cases of depression in various populations.Traditional approaches to psychometric analyses of inventories, such as the BDI and CES-D, use omnibus statistics, such as item-total correlations or reliability coefficients, that average

Depressive Rumination and Past Depression in Japanese University Students: Comparison of Brooding and Reflection

“…The Japanese version has adequate reliability and validity ( Uehara, et al ., 1995 ;Sakado, Sato, Uehara, Sato, & Kameda, 1996 ; Sato, Ue-CES-D score, indicating that currently depressed participants scored higher on each RRS subscale (although Brooding was more strongly correlated with current depression than Refl ection). 8 The rate of participants in this study that scored more than 16 points on the CES-D was similar to the rate that has been previously reported (45%; Santor, Zuroff , Ramsay, Cervantes, & Palacios, 1995 ). hara, Sakado, Sato, Nishioka, & Kasahara, 1996 ).…”

Adult Attachment, Shame, Depression, and Loneliness: The Mediation Role of Basic Psychological Needs Satisfaction.

“…A coefficient alpha of .92 was obtained in the present study. Convergent validity has been established through positive correlations with the Beck Depression Inventory (r ϭ .86; Santor, Zuroff, Ramsay, Cervantes, & Palacios, 1995).…”

REJOINDER: New Coke, Rosetta Stones, and Functional Data Analysis: Recommendations for Developing and Validating New Measures of Depression

“…Accordingly, there is a need to develop a method by which all measures can be calibrated and compared efficiently and effectively. In earlier work, my colleagues and I illustrated this methodology showing how scores on the BDI and CES-D could be equated using a common metric (Santor et al, 1995), which not only facilitates a comparison of how scales perform at different levels of severity but also allows researchers and clinicians to translate scores from one measure to another.…”

“…There are likely to be both advantages and disadvantages in assessing both illnessand wellness-related constructs within a single scale such as the Center for Epidemiologic Depression Scale (CES-D; Radloff, 1977). Indeed, assessing a broader construct that includes both illness and wellness will likely produce a broader range of scores, which may account for the greater discriminability of the CES-D relative to the Beck Depression Inventory (BDI) in nonclinical populations (Santor, Zuroff, Ramsay, Cervantes & Palacios, 1995), but may also account for the great variability in cut scores (see Santor, 2000, for a brief review).…”