A systematic review of the applications of mixture IRT models

Abstract Mixture item response theory (MixIRT) models, integrating latent class models with traditional IRT models, aim to uncover latent subgroups in the data, allowing an IRT model to hold in each identified latent class. These models’ potential for unraveling individuals’ cognitive processing has garnered scholars’ attention, providing deeper insights into their capabilities. The current study replicates Sen and Cohen (2019) review article to offer an updated exploration of MixIRT model applications by examining additional articles from the past five years and considering some extra features. Various applications of MixIRT models were highlighted, such as test speededness, heterogeneity, and differential item functioning. In addition, applications of these models were extended to examine psychometric quality, compare different MixIRT models for model-data fit, and develop more complex MixIRT models. Trends in the implementation of MixIRT models and study characteristics, along with relevant points and information, were reported.