Investigation of affective traits affecting mathematics achievement with SEM and MARS methods

The aim of the study is to examine the affective characteristics that affect mathematics achievement with Structural Equation Modeling and MARS data mining methods. Structural Equation Modeling is a model that requires very intense assumptions. This model, which is a rooted analysis family, is very popular for social sciences due to its various advantages. Could there be an alternative to this analysis, a more understandable and flexible structure? It is intended to seek an answer to this question. In parallel with this aim, firstly, SEM was established with the R program with the affective data selected from the TIMMS 2019 data and the success variable. Later, the MARS model was established with SPM. The results were compared on the goodness of fit and coefficients. As a result, it was concluded that the MARS model could be an alternative analysis at certain points. In particular, the fact that regression analysis is the basis of both statistical analyzes increases this possibility. It has been seen that the MARS model gives slightly better results than SEM in terms of prediction, and the MARS model gives more consistent results with the literature.