Psychological Assessment 2005, Vol. 17, No.

1, 110 –114

Copyright 2005 by the American Psychological Association 1040-3590/05/$12.00 DOI: 10.1037/1040-3590.17.1.110


Validation of the Beck Depression Inventory—II in a Low-Income African American Sample of Medical Outpatients
Karen B. Grothe and Gareth R. Dutton
Louisiana State University

Glenn N. Jones
Louisiana State University Health Sciences Center

Jamie Bodenlos and Martin Ancona
Louisiana State University

Phillip J. Brantley
Pennington Biomedical Research Center

The psychometric properties of the Beck Depression Inventory—II (BDI–II) are well established with primarily Caucasian samples. However, little is known about its reliability and validity with minority groups. This study evaluated the psychometric properties of the BDI–II in a sample of low-income African American medical outpatients (N ϭ 220). Reliability was demonstrated with high internal consistency (.90) and good item-total intercorrelations. Criterion-related validity was demonstrated. A confirmatory factor analysis supported a hierarchical factor structure in which the BDI–II reflected 2 first-order factors (Cognitive and Somatic) that in turn reflected a second-order factor (Depression). These results are consistent with previous findings and thus support the use of the BDI–II in assessing depressive symptoms for African American patients in a medical setting.

The Beck Depression Inventory—II (BDI–II; Beck, Steer, & Brown, 1996) is one of the most frequently used measures of the severity of depression in adolescents and adults by both researchers and clinicians (Brantley, Dutton, & Wood, 2004). The Beck Depression Inventory (BDI; Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) was developed in 1961, revised in 1978 (BDI–IA; Beck, Rush, Shaw, & Emery, 1979), and again in 1996 (BDI–II). The BDI–II is more consonant with the American Psychiatric Association’s (1994) Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM–IV) diagnostic criteria for major depressive episode than the earlier forms. Previous research has established that the BDI–II has good psychometric properties. The BDI–II has demonstrated high internal consistency, good test–retest reliability, and good construct and concurrent validity with other common measures of depression in clinical and nonclinical samples (Beck et al., 1996; Whisman, Perez, & Ramel, 2000). Discriminant validity has also been demonstrated through weaker relationships with measures of other psychopathology, such as anxiety (Steer, Ball, Ranieri, & Beck, 1997).

Karen B. Grothe, Gareth R. Dutton, Jamie Bodenlos, and Martin Ancona, Department of Psychology, Louisiana State University; Glenn N. Jones, Department of Family Medicine, Louisiana State University Health Sciences Center; Phillip J. Brantley, Pennington Biomedical Research Center, Louisiana State University System. Gareth R. Dutton is now at the Department of Psychiatry and Human Behavior, Brown University. Correspondence concerning this article (and requests for an extended report of this study) should be addressed to Karen B. Grothe, who is now at the Department of Psychiatry and Human Behavior, University of Mississippi Medical Center, 2500 North State Street, Jackson, MS 392164505. E-mail: 110

Several studies have examined the factor structure of the BDI–II in differing populations with varying consistency. A two-factor solution (Cognitive, Somatic) has been identified for psychiatric outpatients (Beck et al., 1996), primary care medical patients (Arnau, Meagher, Norris, & Bramson, 2001), clinically depressed outpatients (Steer, Ball, Ranieri, & Beck, 1999), depressed geriatric inpatients (Steer, Rissmiller, & Beck, 2000) and college students (Beck et al., 1996; Dozois, Dobson, & Ahnberg, 1998; Whisman et al., 2000). Two of these studies further demonstrated that the BDI–II consists of two first-order factors, representing cognitive and somatic symptoms that comprise one second-order factor of Depression (Arnau et al., 2001; Steer et al., 1999). Alternatively, a few studies have identified a three-factor solution for the BDI–II with adolescent psychiatric outpatients (Cognitive, Somatic, and Guilt/Punishment; Steer, Kumar, Ranieri, & Beck, 1998) and college students (Negative Attitude, Performance Difficulty, and Somatic Elements; Osman et al., 1997). Although these studies indicate that the BDI–II has moderately stable psychometric properties across groups, the majority of subjects were Caucasian and middle-class, with no study including more than 15% of their sample as African American. Therefore, little is known about its psychometric properties in minority and low-income samples. Because racial differences in the expression of depressive symptoms have been found, it cannot be assumed that the reliability and validity for the BDI–II established with primarily Caucasian samples remain accurate for African Americans. Specifically, it has been demonstrated that African Americans evidence more somatic symptoms, in particular, sleep disturbance (Ayalon & Young, 2003; Brown, Schulberg, & Madonia, 1996), and they articulate fewer typical depressive symptoms than depressed Caucasians (Wohl, Lesser, & Smith, 2002). Further, the American Psychological Association’s (2002) Ethical Principles of Psychologists and Code of Conduct explicitly states that it is the

1%) of the sample met criteria for current major depressive disorder.34). 1998). The average number of chronic illnesses for this sample was 1. To assess validity. Thirty-six percent were employed either full. & Mather. Specifically. A demographic questionnaire was administered to all participants in order to collect the following information: age. BDI–II. SD ϭ 10. minority population. Specifically. education level. self-criticalness. The PRIME-MD Mood Module is a structured interview consisting of yes–no questions originally designed for use by primary care physicians (Spitzer et al. They were recruited from the waiting rooms of two medical outpatient clinics at a teaching hospital in the southeastern United States. average education level was lower (M ϭ 11.’s (1996) original factor loadings of the BDI–II for the psychiatric outpatient sample: sadness.63 (SD ϭ 10. we used confirmatory factor analysis (CFA) in an attempt to replicate the hierarchical factor structure identified in previous studies.26. low-income. Ratings are summed to calculate a total BDI–II score. employment status. had to demonstrate a fifth-grade comprehension level to ensure adequate understanding of subsequent questions. Results Sample Characteristics Of the 220 participants (115 women. Thus. The purpose of the present study was to examine the reliability and validity of the BDI–II in a sample of African American. Participants were approached while waiting for their regularly scheduled medical appointments. This facility provides medical care to a predominantly low-income.6%. ␹2(188. The remaining 12 items were restricted to load on the Somatic factor.5. Figure 1 presents the model. The first-order factors were treated as indicators of the second-order factor. Each item contains four statements reflecting varying degrees of symptom severity. 38% was single. To identify the second-order Depression factor. McGrew.BRIEF REPORT 111 responsibility of psychologists to ensure that assessment instruments used in our field demonstrate sound psychometric properties in a variety of ethnic minority groups (p. Participants were administered the oral comprehension portion of the Woodcock–Johnson III Tests of Achievement (Woodcock. 2001). ranging from 3 to 16 years. and we examined criterion-related validity with the use of a brief structured interview. Study personnel then administered the Mood Module of the PRIME-MD orally prior to scoring the BDI–II. Although 60% of the patients had graduated high school.75). the nine Cognitive factor symptoms were constrained to have zero loadings on the Somatic factor and vice versa. Method Sample Participants included self-identified African American patients who were at least 18 years of age. Eligible persons were administered a demographic questionnaire and completed the BDI–II individually when possible.5% (95% CI ϭ 23. Spitzer et al. self-dislike. 105 men) in the current study. we attempted to replicate the Arnau et al. guilty feelings. pessimism. Respondents are instructed to circle the number (ranging from zero to three.51 (SD ϭ 1. Primary Care Evaluation of Mental Disorders (PRIME-MD) Mood Module. suggesting that the model does not explain all the covariance among the items and that residual variance requires explanation. The normal theory maximum-likelihood chi-square test for the CFA model was significant. 1994). We performed the CFA with EQS 6. However. The mean BDI–II total score was 12. medical patients. and PRIME-MD results indicated that 29. patients were administered Items 7–13 on the comprehension subtest. The Institutional Review Board of Louisiana State University Health Sciences Center approved this study. punishment feelings. provided a brief description of the study. but the two factors were allowed to correlate by virtue of both being indicators of the underlying depression factor. Written informed consent was obtained. we conducted a CFA to verify the same factor structure in the current minority sample.61) between depression diagnoses made with the PRIME-MD and diagnoses reached by mental health professionals using lengthier clinical interviews and who were blind to PRIME-MD results. 36. 1072.91). uninsured. see also Okazaki & Sue. other indices of goodness of fit indicated that the two-factor solution resulted in an adequate fit for the data Procedure Six doctoral students in clinical psychology were trained in the administration of the PRIME-MD Mood Module and suicide assessment.49. N ϭ 220) ϭ 281. we constrained the two first-order factors to have equal covariances (unstandardized loadings) with the Depression factor. Those who did not pass the comprehension test were excluded from the study. with a range of 0 to 6. we established reliability by examining item-total correlations and the internal consistency of item scores. SD ϭ 1. Thus. All of the standardized path coefficients of the symptoms for the two first-order factors were salient (Ͼ.35).1 (Bentler. The interviewers approached patients in clinic waiting rooms. (2001) and Steer et al. The BDI–II consists of 21 items assessing symptoms of depression experienced during the past 2 weeks. Twenty-eight percent of the sample was married. past failure. gender. indicating increasing severity) that corresponds with the statement that best describes them. suicidal thoughts or wishes.39). the reading level of the BDI–II. the two factors correlated with each other (r ϭ . As expected. as correct performance on these items is indicative of at least a fifth-grade comprehension level. and chronic medical conditions. ages ranged from 20 to 81 (M ϭ 49. Woodcock–Johnson III Oral Comprehension. and 34% was divorced or widowed. (1999) findings of two first-order factors (Cognitive and Somatic) that are indicators of a second-order factor (Depression).98). Patients . and 12 (4%) did not pass the oral comprehension test. 2003). and worthlessness.or part-time. The PRIME-MD covers the diagnostic criteria for major depressive disorder as outlined by the DSM–IV. p Ͻ . Participants were informed that they would receive a snack after completion of the study as an incentive to participate. Referrals were made to mental health services on the basis of responses to the BDI–II and PRIME-MD. which can range from 0 to 63. (1994) reported satisfactory agreement (␬ ϭ . which was followed by the measure of oral comprehension. The following BDI–II items were restricted to load on the Cognitive factor on the basis of Beck et al. Forty-six of 281 patients approached (16%) declined participation. marital status. and went with interested patients to a private examination room or a separate area within the waiting room..01. Factorial Validity Given that the BDI–II factor structure has been replicated in several samples. Measures Demographic questionnaire.

.10).25. The fit criterion (␹2/df) was 1. ␣ ϭ .90). As L. Somatic factor. Chou & Bentler. and the nonnormed fit index was . Cognitive factor. p Ͻ . This statistic is robust to violations of normality and is probably a more appropriate test for item-level data (see.81. Coefficients are standardized loadings. The Satara–Bentler scaled chi-square also indicated that the model was an excellent fit. according to Hatcher’s (1994) criteria (shown in parentheses) for acceptability of fit. the root-mean-square residual was . was appropriate. in which two first-order factors (Cognitive and Somatic) represented one second-order factor (Depression). Clark and Watson (1995) pointed out. e.90).g.34 (and Ͻ2.93 (and Ͼ.87).02. we concluded that acceptance of the hierarchical model. BDI ϭ Beck Depression Inventory.0).112 BRIEF REPORT Figure 1. ␣ ϭ .90. Therefore. ␹2(118.04 (and Ͻ. 1995). A hierarchical model of the Beck Depression Inventory—II factor structure. . E. N ϭ 220) ϭ 201.94 (and Ͼ. the comparative fit index was . ␣ ϭ . Internal Consistency Internal consistency of the BDI–II and its factors was quite high (BDI–II total score.

33 .24 .21 .37 . Past failure 4. and by Arnau et al.27 .40 .19 .24 . Loss of energy 16.98 .30 .25 . Loss of pleasure 5. Agitation 12.45 . the BDI–II demonstrated excellent internal consistency (.15 .58 — .39 .09 .28 .36 .57 — .28 .18 . rii ϭ .19 .90.64 — .25 .33 .54 — .48 .34 .33 .33 .40 . With regard to reliability.37).35).46 .28 . (1997).19 .51 .19 .37 .35 .44 .22 . and Corrected Item-Total Correlations (rtot) Among Beck Depression Inventory—II Items for African American Medical Outpatients Symptom 1. Irritability 18. Osman et al.90 .74 .29 .23. (2001) for primary care medical patients (. SD ϭ 7. the fit indices for the present sample of African American medical patients actually appear better than the fit indices Osman et al.21 .30 .55 — .57 .41 .17 .18 .47 .32 .35 .56 — . (1999) for clinically depressed outpatients (.05 .29 .50).12.24 .56 .25 .38 . Loss of interest in sex M SD rtot 1 — .99 .07 .36 .67 — .52 .39 . and item intercorrelations of the 21 BDI–II items for the current sample are presented in Table 1. African American medical patients as it was for the primarily Caucasian samples of previous work.57 .33 .23 .01.37 .36 .54 .73 . ␩2 ϭ .39 .34 . yet the fit indices suggest the model is an acceptable fit.41 .19 .33 .47 .20 .15 . (1997) obtained in their sample of college students for the Beck et al. as the use of somatic items in measures of depression for medical patients has been criticized.23 . rii ϭ .22 .26 .29 . (1996) correlated two-factor model.38 .32 . Guilty feelings 6. These results were also similar in magnitude to those identified by Arnau et al.25 .43 — .20 .31 .94 . which is similar to those reported by Beck et al.30 .36 . (2001) for medical outpatients..41 . Means.41 .81 .49 .86 1.35 .22 .30 .25 .22 . by Steer et al.27 . Suicidal thoughts or wishes 10.28 .40 .27 .83.34 . which was comparable with coefficient alphas for primarily Caucasian samples reported by Beck and colleagues (1996) for psychiatric outpatients (.50 .31 . Changes in appetite 19.36 .30 . Cohen’s d ϭ 1.31 .31 .30 .32 . Validation of the BDI–II in a medical setting is particularly salient.95 .39 . Punishment feelings 7.50 — .33 .32 .34 . Pessimism 3. Tiredness or fatigue 21.10 .26 .38 .28 .31 .36 .25 . item-total correlations ranged from .16 .43 .76 .25 .39 .30 .27 . Sadness 2.44 . Evidence for criterion validity revealed that depressed patients had higher BDI–II scores than nondepressed patients.90) in this sample.24 .42 .86 . low-income medical patients.45 .67.39 .82 .20 . In addition.30 .56 1.28 .28 . These estimates of internal consistency were of similar magnitudes (total BDI–II.35 . Self-criticalness 9. Standard Deviations.23 . A receiver operating characteristic analysis has been presented elsewhere (see Dutton et al.38 . The lowest item-total correlation for our sample was for the suicidal thoughts or wishes item (. Self-dislike 8. It has been suggested that somatic symptoms often endorsed by medical patients cause these patients to score highly on measures of depression.40 .07 .93).22 . (1998).42 .40 . Analyses indicated that patients with a diagnosis of current major depression had significantly greater BDI–II total scores (M ϭ 23. standard deviations.66) compared with patients without this diagnosis (M ϭ 8.38 .43 .18 .56 .47 .94).24 .41 .41 .25 .27 .41 .31 .46 .08 .34. Somatic factor. and a more appropriate index of internal consistency is the average interitem correlation (rii).23 .45 . Our results demonstrate strong evidence for the reliability and validity of the BDI–II in this sample and are quite consistent with previous findings for primarily Caucasian samples.37 .32 .22 1.35 .78 .38 .32 .24 . The presence or absence of major depression in the present sample was established with the PRIME-MD. Factorial validity was demonstrated through acceptance of the hierarchical model composed of two first-order factors (Cognitive and Somatic) that both reflect a second-order factor (Depression). Changes in sleeping pattern 17.36 .36 .41 .57 — .77 . and Dozois et al.39 .43. Indecisiveness 14.57 .32.35 to .55 — .33 .32 .34 .35 .25 .70 .35 — .33 .24 .15 .43 .49 — .41 .27 .28 .37 .30 . 2003).28 .33 .30 .23 .34 .72 .99 .17 . p Ͻ .BRIEF REPORT 113 alpha is influenced by the number of items.28 . Criterion Validity Although the BDI–II was not specifically designed for diagnostic purposes.39 .47 — .25 . t(218) ϭ 12. and patients with major depression should have higher BDI–II scores on average than patients without this diagnosis.26 .37 .30 . rii ϭ .40 .41 . 2004).39 .34 . Cognitive factor.27 .44 .32 . SD ϭ 8.56 — .49 .61 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 — .30 .44 .40 .44 . Discussion The present study examined the factor structure of the BDI–II in a sample of African American. This factor structure appears to be as appropriate for low-income.25 .61 .32 . Indeed. which may be related to the finding that African Americans are less likely to endorse suicidal thoughts than Caucasians (Ayalon & Young. Loss of interest 13.25 .21 .29 .92) and college students (.35 . Concentration difficulty 20.53 2 — .29 .35 .90).28 . regardless of Table 1 Intercorrelations.49 .37 . Worthlessness 15.93 .24 .53 .33 1.56 .63 — .07 .31 .37 . Crying 11.43 .29 .45 — .13 .78 .55 — .34 .38 . corrected item-total correlations. Means.45 .35 .93 .92 . it assesses the presence and severity of depressive symptoms.45 .21 . Our results are particularly parallel to the results of Steer and colleagues (1999) and Dozois and colleagues (1998) in that the chi-square test for the current CFA model was significant.33 .50 . (1996).

. 705–713. although used in the present study to enhance comparison with previous research. (1995). Rissmiller. R. S. (2003). (2003). Beck Depression Inventory: Manual (2nd ed. (2004). L. (2004). & Sue. Journal of CrossCultural Psychology.. Dobson. I. A. the generalizability of these results is limited to low-income. (1983). Clark. our sample was restricted to African Americans.. Brantley. J.. & Gibbons. Thousand Oaks. T. W. J. Hahn. Diagnostic and statistical manual of mental disorders (4th ed... (1994). Whitehead. & Young. the sample included in this study was a sample of convenience. & Wood.. 55. J. Given the widespread use of the BDI–II. Norris. & Mather. Spitzer. (1998). Journal of Clinical Psychology. M. A.. H.).. C. First. Encino. DC: American Psychological Association. Use of the Beck Depression Inventory—II with African American primary care patients... G. Boston: Harcourt Brace. Ball. Dimensions of the Beck Depression Inventory—II in clinical depressed outpatients. Dutton. C.. G. Dozois. Clark. 181–191. Cavanaugh. The use of psychological testing Received April 14. A step-by-step approach to using the SAS system for factor analysis and structural equation modeling. NJ: Erlbaum. B. Journal of Nervous and Mental Disease. Methodological issues in assessment research with ethnic minorities. M. DeGruy. Mahwah. P. 279 –284. E.. and applications (pp.. 3. D. A. 171. Hoyle (Ed. S. (2001). Gutierrez. 1749 –1756.. Washington. 561–571. 2004 Accepted September 14. (1961). pp. R. Grothe. 34. A. T. & Smith. NC: SAS Institute. Mock. this may be a result of sampling from a medical population attending physician appointments. Cary. E. BDI–II. R. F. 37–55).). Psychometric evaluation of the Beck Depression Inventory—II with primary care medical patients. Downs. (1999).1 Beta Build 70) [Computer software]. Further evidence for the construct validity of the Beck Depression Inventory—II with psychiatric outpatients. Beck. we experienced a 16% refusal rate for participation among our population. EQS Structural Equations Program for Windows (Version 6. B. In A. & Ramel. (1998). 20. 1060 –1073. (2001). R. 26. The core symptoms of depression in medical and psychiatric patients. 56. Kumar.. F. R. A psychometric evaluation of the Beck Depression Inventory—II. Clinical presentations of major depression by African Americans and Whites in primary medical care practice. B. J. W.. Estimates and tests in structural equation modeling. Mendelson.. 20. F. S. 263–281). & Bentler. 112–119. Factor structure and psychometric characteristics of the Beck Depression Inventory—II. H. T. W. D. M.. Ethical principles of psychologists and code of conduct. Thus. Steer. Steer. & Brantley.. DC: Author. Cognitive therapy of depression. C. 38. M. CA: Sage. Arnau and colleagues (2001) determined that the inclusion of somatic items on the BDI–II was appropriate for their medical sample. (1995). (2000). Psychological Assessment. Kopper. A.. Kazdin (Ed. V. M. K.. 111–124. D. (1998). The use of a more extensive psychiatric interview may have enhanced our attempt to establish criterion validity. Maruish (Ed. Structural equation modeling: Concepts. Lesser.. G. S. A.. Health Psychology. depression status (D.. 127–137. V. A. Use of the Beck Depression Inventory—II with depressed geriatric inpatients. Clark. A. Journal of Psychopathology and Behavioral Assessment. C. A.. W. B. D. & Watson. J. Hatcher. K. A. Finally. Several limitations to the present study should be noted. Rush. A. African American medical outpatients attending physician appointments. R. W. & Chiros. Williams. all items made salient contributions to the first-order factors. N.. Brown. J. 41. P. & Emery... M.. A. the PRIME-MD Mood Module is a brief measure designed for use in medical settings.114 BRIEF REPORT for treatment planning and assessment outcome (3rd ed. S. Ranieri. New York: Guilford Press. 2004 Ⅲ . (2000). 1983).. G. K.). Factor structure of the Beck Depression Inventory—Second Edition (BDI–II) in a student sample. Journal of the American Medical Association. 83– 89. IL: Riverside... H. L. F. Utility of a new procedure for diagnosing mental disorders in primary care: The PRIME-MD 1000 study. Steer. which was not the case in the present study.. Schulberg. 57. 272. A. Dutton. & Brown. K. C. In M.. WoodcockJohnson III: Tests of Achievement. Although patients were not queried directly as to why they refused to participate. issues. Okazaki. M. Perez. 80. American Psychological Association. Ball.. K. M. J.. Cavanaugh. Shaw. E.. C. General Hospital Psychiatry.. N. Chou. A. X. J. Clinical presentations of depression in African Americans and White outpatients. M. A comparison of depressive symptoms in African Americans and Caucasian Americans. & Beck.). Barrios. it is important that the measure is validated with other ethnic minority groups. L. (2002). et al.. Ranieri. R. Steer. The Beck Depression Inventory—II (BDI–II) and the Beck Depression Inventory—Primary Care (BDI–PC). In addition.. W. McGrew. B. P. Use of the Beck Depression Inventory—II with adolescent psychiatric outpatients. 545–551. & Beck. R. G. Beck. Second.. 4. Kendra. References American Psychiatric Association. 443– 446.. Psychological Reports. It would be expected that corrected item-total correlations for somatic items would be lower if these items were not good indicators of depression. Steer. All participants were low-income. D.. Kroenke. A. (1979). M. Ranieri.. Jones. Ward. Bentler. Wohl.. R. T. K. Behaviour Research and Therapy. & Madonia. W. (1994). (1996). & Ahnberg. (2002). 313–326). Cultural Diversity and Mental Health. C. Ayalon. P. K. P. & Erbaugh. & Bramson. R. 359 –376.. S. A. L. 309 –319. as many patients appeared ill or lethargic. 117–128. Our results demonstrate the same finding with the criteria they set forth. A.... Third. E. Itasca. & Beck. Journal of Affective Disorders. CA: Multivariate Software. M... pp. 10. American Psychologist. R. M. Beck... A. Psychological Assessment. J. Washington. J. T. W. R. Journal of Clinical Psychology. C. P. An inventory for measuring depression. Journal of Psychopathology and Behavioral Assessment. although the present study adds to the research examining psychometric properties of assessment instruments in minority populations. Woodcock. (1997). & Beck. Whisman.. E. G. J. 19. Meagher.. African American medical patients and should be replicated with African Americans of differing economic status as well as psychiatric inpatients and outpatients.. 7. T. & Gibbons. P. which argues for the inclusion of somatic items.. Arnau. L. Constructing validity: Basic issues in objective scale development. (1994). R. Methodological issues & strategies in clinical research (2nd ed. F. 311–318. R. In R. 437– 442. T. R. 2004 Revision received September 13. Osman. (1997).. Archives of General Psychiatry.). (1996). D. Linzer. R.