The Effect of Covariant Variable on Determination of Differential Item Functioning Using Mixture Rasch Model

• In Event: Advancements in Detecting Differential Item Functioning and Measurement Invariance

Abstract

The aim of this study is to investigate the effects of including a covariant variable in a mixture Rasch Model, on the model’s latent class structure and latent DIF determination. The results have shown that including a covariant variable in a MRM has lowered the model’s efficiency to define latent classes. However, when the inclusion of a covariant variable is made certain conditions were met, such as a larger DIF effect size, a strong relationship between covariant variable and the latent variable and the latent class size is equal, the accuracy of classification has reached the same level as MRM. When covariate include in MRM, power to determine latent DIF and percentage of correct non-DIF decisions was improved.

Authors

• Gozde Sirganci, Southern Methodist University

• ömay çokluk bökeoğlu