Investigating Population Heterogeneity and Interaction Effects of Covariates: The Case of a Large-Scale Assessment for Teacher Licensure in Saudi Arabia

摘要

This article investigates the population heterogeneity of test data for the case of teacher licensure assessments in Saudi Arabia. The results from factor mixture modeling of the data (N = 15,962) on the construct of promoting learning revealed the presence of two latent classes of examinees based on their performance profiles across test items, with noninvariant parameters on a one-factor model of the construct across the two latent classes. Model covariates, such as gender, teaching experience, and age, demonstrated interaction effects in the formation of the two latent classes of examinees. The findings and methodology provided with this article can be useful to researchers in different contexts of educational and psychological assessment.