Performance of Exact Versus Approximate Methods in Evaluating Longitudinal Measurement Invariance

• In Event: 29.082 - Roundtable Session 9
  In Roundtable Session: 29.082-14 - Linking, Equating, and Multidimensional Item Response Theory

Abstract

The use of exact methods (i.e., GFIs) in testing Measurement Invariance (MI) between independent groups has been studied. Cut-off values with Type I error and power were recommended. MI across independent groups is analogous to MI with repeated measures. To date, there is no study that has examined the function of GFIs in testing L-MI, neither by themselves nor with their performance in comparison to the approximate methods. The purpose of this study is to fill the research gap by assessing the performance of the approximate methods and some commonly used GFIs under varying configurations of L-MI, with the hope of finding the guideline for their use in longitudinal data from the perspective of Type I error and power rates.

Author

• Caiyan Zhang, The College Board