Building a new Rasch-based self-report inventory of depression

Abstract

This paper illustrates a sequential item development process to create a new self-report instrument of depression refined with Rasch analysis from a larger pool of potential diagnostic items elicited through a consensus approach by clinical experts according to the latest edition of the Diagnostic and Statistical Manual of Mental Disorders criteria for major depression. A 51-item pool was administered to a sample of 529 subjects (300 healthy community-dwelling adults and 229 psychiatric outpatients). Item selection resulted in a 21-item set, named the Teate Depression Inventory, with an excellent Person Separation Index and no evidence of bias due to an item–trait interaction ( χ ²=147.71; df =168; P=0.48). Additional support for the unidimensionality, local independence, appropriateness of the response format, and discrimination ability between clinical and nonclinical subjects was provided. No substantial differential item functioning by sex was observed. The Teate Depression Inventory shows considerable promise as a unidimensional tool for the screening of depression. Finally, advantages and disadvantages of this methodology will be discussed in terms of subsequent possible mathematical analyses, statistical tests, and implications for clinical investigations.