Eğilim Puanları Kullanılarak Abide Çalışmasındaki Maddelerin Değişen Madde Fonksiyonu Açısından İncelenmesi

Abstract

The purpose of this study is to examine whether the items in the Turkish, mathematics and science subtests of the ABIDE 2016 research show differential item functioning (DIF) according to the school type variable before and after the propensity score matching (PSM). In this context, while examining the effect of PSM on DIF and the steps of the PSM methodology are presented. DIF analyzes were performed before and after PSM using Mantel-Haenszel (MH) and logistic regression (LR) methods. The data of the study consisted of 6330 students who participated in the ABIDE 2016 application provided by the Ministry of National Education, General Directorate of Measurement, Evaluation and Examination Services. As a result of DMF analyzes performed on the whole group before PSM analyzes, it was determined that four items with the MH method and one item with the LR method showed DMF. PSM was carried out by the nearest neighborhood matching method. The variables of socioeconomic level, interest in family, education level of mother and father were used as covariate in calculating the propensity score. The matching balance was checked, and as a result of the matching it was determined that the balance was achieved in the public and private school groups in terms of covariates. As a result of the PSM analyzes, 489 students from the public and private school groups were matched, and DIF analyzes were carried out on 978 students. As a result of these analyzes, it was determined that an item showed DMF by the MH method and that there was no item showing DIF with the LR method. It was observed that there was a decrease in the number of substances showing DIF after PSM. It has been concluded that using PSM provides control over potential sources of DIF.