Abstract
This study compare the effect of four scoring methods for the Center for Epidemiologic Studies Depression Scale (CES-D) on score reliability, concurrent validity, cut points, sensibility, specificity and classification reliability of the scale. The CES-D was scored using the conventional “ordinal” method, two binary methods (“presence” and “persistence” of symptoms) and a new “weekly” scoring system. On the basis of both psychometric analysis and receiver operating characteristic (ROC) analysis, performed on normative (n=1143) and clinical (n=44) samples, it was found that the “ordinal” and “weekly” methods performed best in detecting depression. On the other hand, the “persistence” of symptoms method resulted in worse performance. Overall, the results indicate that the scoring method has an effect on the reliability and validity of the CES-D.
