In the literature, estimation results of the determinants of academic achievement are controversial. There may be several reasons for these controversial results, including sample or cultural differences. Conversely, these results may arise from ignoring certain important facts, such as an interaction effect. Some studies do not consider interactions among students, and some studies may not use effective models. Surprisingly, very few studies have focused on student academic achievement using spatial models, which may be one of the most suitable models for testing interaction effects. In this study, we estimated student achievement using spatial models and a data sets from Turkey. We observed an interaction between students who live in the same neighbourhood and found evidence of an interaction among students in terms of their achievement based on a spatial error model.