An Application of Latent Class Analysis for TIMSS 2015 Data: Detecting Heterogeneous Subgroups

Abstract

This study aimed to investigate the heterogeneity of the TIMSS 2015 data from Turkey and the USA 8th grade math. Latent Class Analysis (LCA) was used to determine the latent classes that cause heterogeneity in the data by using categorical observed variables. As a result of the LCA, supporting absolute and relative model fit indices through AvePP and entropy values, it was concluded that the data obtained from both countries fit the three-class model. The latent class probabilities and conditional response probabilities were examined for homogeneity and degree of segregation of the classes from each other. Based on the findings, it is recommended that the assumption of homogeneity in international evaluations be evaluated empirically with LCA. With this article, an example of the application of LCA is provided, and it is believed to be useful for researchers in the context of education and psychological evaluation.