Investigating Differential Item Functioning in DINA Model

OpenAlex İngilizce

In this study, it is aimed to investigate the effects of various factors on the performance of the methods used in the determination of differential item functioning (DIF) in the DINA model included in the Cognitive Diagnosis Models. The current study is limited with Logistic Regression and Wald test methods which were used to determine the differential item functioning in DINA model. The Type I error and power rates of these methods in certain conditions were investigated to evaluate their performances. In the simulation study for the Type I error rates, four variables were manipulated: sample sizes, the number of attributes, correlations between attributes and reference group s and g parameter values. In the determination of the power rates of the methods, additionally, the variables that were manipulated in the Type I error study, DIF sizes and percentages of DIF items were manipulated, too. As a result, it was observed that especially in all cases where reference group’ s and g parameter values are low, both methods yielded a good control of Type I error rates. In addition, according to the results, it was observed that both DIF size and sample size affect the power rates of both methods.