The mixed Rasch model (MRM) is an extension of the standard Rasch model, which identifies latent classes of respondents for whom differential item functioning (DIF) exists. MRM helps in detecting qualitative differences among test takers and in finding out how individual differences affect test performance by decomposing the population into meaningful subpopulations that differ qualitatively. The aim of this chapter is to demonstrate an application of MRM in language testing research. For illustration purposes, we applied the MRM to a high-stakes reading comprehension test in English as a foreign language and identified two latent classes of respondents. In our example, examination of class-specific item profiles revealed that the identified latent classes might be related to individual differences in working memory capacity, which is essential in integrating propositions formed from the text to create a coherent picture.