Curtailment and Stochastic Curtailment to Shorten the CES-D

Abstract

The Center for Epidemiologic Studies–Depression (CES-D) scale is a well-known self-report instrument that is used to measure depressive symptomatology. Respondents who take the full-length version of the CES-D are administered a total of 20 items. This article investigates the use of curtailment and stochastic curtailment (SC), two sequential analysis methods that have recently been proposed for health questionnaires, to reduce the respondent burden associated with taking the CES-D. A post hoc simulation based on 1,392 adolescents’ responses to the CES-D was used to compare these methods with a previously proposed computerized adaptive testing (CAT) approach. Curtailment lowered average test lengths by as much as 22% while always matching the classification decision of the full-length CES-D. SC and CAT achieved further reductions in average test length, with SC’s classifications exhibiting more concordance with the full-length CES-D than do CAT’s. Advantages and disadvantages of each method are discussed.