Modeling Differential Item Functioning Using Generalized Linear and Nonlinear Mixed Models

• In Event: Marriott Roundtable Session 35
  In Roundtable Session: Differential Item Functioning

Abstract

Traditional approaches to detect differential item functioning (DIF) under the framework of generalized linear and nonlinear mixed models (GLNMMs) mainly focus on the use of generalized linear mixed models, for example, the logistic mixed model, which is a special case of GLNMMs, with an emphasis on uniform DIF only. This paper examines the feasibility of GLNMMs in modeling both uniform and nonuniform DIF simultaneously by performing comprehensive simulation studies, and the results demonstrate that GLNMMs are powerful in DIF detection.

Authors

• Jing Jiang, Boston College

• Zhushan Li, Boston College