Comparing Long- Versus Short-Forms of Depression Scales in an Omnibus Longitudinal Survey

Abstract

Many large scale omnibus surveys have adopted shorter versions of the original lengthy psychological scales for efficiency. Evidence is lacking on the impact of different forms on respondent behaviors and data analysis. This study focuses on the widely used Center for Epidemiologic Studies Depression Scale (CES-D) as a case in point, by utilizing data from the nationally representative China Family Panel Studies. It compares CES-D20 with CES-D8 as they have been applied to two separate and equivalent groups. Our findings suggest that reducing the test length from CES-D20 to CES-D8 does not lead to a notable reduction to their psychometric qualities. Respondent behaviors as measured by survey perceptions in its concurrent wave, and attrition rates two years later are not affected by the forms. Analysis by combining data from the two forms is possible after proper conversion of the scores to be on the same metric without worrying about the form effect. In conclusion, the findings provide supporting evidence for a possible smooth transition between long and short forms of scales under proper survey designs.