Validation of the Center for Epidemiologic Studies Depression Scale—Revised (CESD-R): Pragmatic depression assessment in the general population
Introduction
Major depression affects approximately 5% of the US population annually (Hasin et al., 2005). Although understanding of depression has increased dramatically in recent years, the most commonly used assessment scales are approximately 25 years old. A literature search in PSYCH ARTICLES suggests the most popular depression scales are the Beck Depression Inventory (BDI; Beck et al., 1961), the Center for Epidemiologic Studies Depression Scale (CES-D; Radloff, 1977), and the Hamilton Rating Scale for Depression (HAM-D; Hamilton, 1960). Psychometric flaws and clinician administration are major drawbacks to the use of the HAM-D (Bagby et al., 2004). In contrast, the BDI and CES-D are both self-administered, making them popular alternatives. Design differences (e.g., different targeted populations, response option formats, and emphasis on different aspects of depression) between the BDI and CES-D lead to measurement of different variations of depression (Skorikov and VanderVoort, 2003). Structural equation modeling analyses suggest a standardized factor loading of 0.51 and 0.63 for the BDI and CES-D, respectively, on a common hierarchical factor, suggesting that the latent traits for the BDI and CES-D are similar, but ultimately different (Skorikov and VanderVoort, 2003).
Neither the BDI nor the CES-D reflects current diagnostic criteria for major depression (American Psychiatric Association, 1994). While the revised form of the BDI (BDI-II; Beck et al., 1996) is growing in popularity, the revision of the CES-D (CESD-R; Eaton et al., 2004) is scarcely mentioned in the literature. The revision of the CES-D was undertaken to more reliably indicate general dysphoria and reflect the nine primary symptoms of a major depressive episode according to DSM-IV (Eaton et al., 2004). Among other aspects of the revision, an extra response category (nearly every day for two weeks) was added, two existing items were simplified, items predominantly unrelated to modern notions of depression were removed, and items reflecting anhedonia, psychomotor retardation/agitation, and suicidal ideation were added (see Eaton et al., 2004). In attempt to optimize the psychometric properties, items reflecting positive affect were eliminated. Some have argued that negative responses to positive items may reflect important aspects of depression (e.g., Wood et al., 2010). Several new items seem to directly reflect the inverse of old positive items. Thus, the important discriminative contribution of positive affect (e.g., Mineka et al., 1998) may potentially be compensated via better representation of dysphoria and anhedonia.
The present study aimed to determine the factor structure and to examine psychometric properties of the CESD-R across a large community sample and a smaller student sample. Comparison of base-rates of depression identified by the CESD-R in a large community sample to those identified by prior epidemiological research (e.g., Hasin et al., 2005) can provide important validation of the measure. Research has shown that the traditional CES-D cut-score, which also applies to the CESD-R (Eaton et al., 2004), lacks specificity (e.g., Santor et al., 1995). Therefore, an additional goal was to compare the original classification method (cut-score of 16) to an algorithmic method based on the DSM-IV approach. Correlates of the scale were also examined to explore convergent and divergent validity, as follows.
Depression and anxiety are known to be highly co-occurring conditions (Mineka et al., 1998). Accordingly, a measurement of depression should have a high correlation with a measure of anxiety. Similarly, but to a much lesser extent, schizotypy has shown positive correlations with emotional disorders, reflecting neurodevelopmental vulnerabilities for impaired cognitive and affective regulation. In particular, recent work suggests moderate relationships between positive dimensions of schizotypy and affective disorders (Lewandowski et al., 2006). Finally, while increased negative affect (NA) is common to emotional disorders (e.g., anxiety and depression), deficits in positive affect (PA) are specific to depression (Mineka et al., 1998). Since NA is common to emotional disorders, both anxiety and depression scales should significantly correlate with NA, even if common variance is controlled for (e.g., including both depression and anxiety simultaneously in the correlations). A valid measure of depression should also significantly correlate with PA after controlling for anxiety, but anxiety should not correlate with PA after controlling for depression.
Section snippets
Procedure
Participants either (a) responded to an email request issued to the National Organization for the Reformation of Marijuana Laws (NORML) listserv (Sample 1) or (b) completed questionnaires for undergraduate course credit at a state university in the northeastern United States (Sample 2). Upon completion of the survey, sample 1 participants were entered into a drawing for a $250 Amazon.com giftcard or 1 of 5 4 GB iPods. Willing participants from sample 1 forwarded the email to others who might be
CESD-R scale properties
In sample 1, 6971 participants completed all CESD-R items for a total score range of 0 to 77 (M = 10.3, S.D. = 11.7). The distribution of CESD-R scores had a large positive skew (2.15 ± 0.03) and was leptokurtic (kurtosis = 5.77 ± 0.06). Internal consistency was high (Cronbach's α = 0.923). In sample 2, 243 participants completed all CESD-R items for a total score range of 0 to 71 (M = 16.4, S.D. = 13.5). The distribution of CESD-R scores had a positive skew (1.49 ± 0.16) and was leptokurtic (kurtosis = 2.56 ±
General discussion
Exploratory and confirmatory factor analyses, assessment of internal consistency, and exploration of convergent and divergent validity all suggest the CESD-R has strong psychometric properties, making it a useful tool for assessing depression in the general population. The CESD-R also has theoretically consistent relations to measures of anxiety, negative affect, and positive affect (e.g., Mineka et al., 1998). These findings are particularly salient given that the majority gender is opposite
Acknowledgements
The authors thank Dick Haase and Gerhard Mels for technical advice and Gillinder Bedi and Kyle DeYoung for comments on earlier drafts.
References (26)
- et al.
Anxiety and depression symptoms in psychometrically identified schizotypy
Schizophrenia Research
(2006) - et al.
Does the CES-D measure a continuum from depression to happiness? Comparing substantive and artifactual models
Psychiatry Research
(2010) Diagnostic and Statistical Manual of Mental Disorders
(1994)- et al.
The Hamilton Depression Rating Scale: has the gold standard become a lead weight?
American Journal of Psychiatry
(2004) - et al.
Manual for the Beck Depression Inventory-II
(1996) - et al.
An inventory for measuring depression
Archives of General Psychiatry
(1961) Confirmatory Factor Analysis for Applied Research
(2006)- et al.
Center for Epidemiologic Studies Depression Scale: review and revision (CESD and CESD-R)
- et al.
Psychometric properties of the State-Trait Inventory for Cognitive and Somatic Anxiety (STICSA): comparison to the State-Trait Anxiety Inventory (STAI)
Psychological Assessment
(2007) A rating scale for depression
Journal of Neurology, Neurosurgery, and Psychiatry
(1960)