TIMSS – 2019 Bilgisayar Kullanımına İlişkin Maddelerin Ölçme Değişmezliği ve Değişen Madde Fonksiyonu Açısından İncelenmesi

Abstract

This study aimed to investigate the measurement invariance and bias of computer use-related items in TIMSS 2019 across different languages and cultures using the changing item function method. The analysis included data from Turkiye, England, and Qatar, and utilized MG-CFA and OLR with Poly-SIBTEST methods for measurement invariance testing. The study used R studio, SIBTEST, and SPSS for analysis and calculated Cronbach's alpha coefficient for reliability, as well as conducting exploratory and confirmatory factor analyses for validity. The results showed that structural invariance was achieved for items in the same culture (AR-ENG), and structural and metric invariance was achieved for items in different languages and cultures (TUR-ENG). Upon analysing the bias, it was found that the OLR method revealed four uniform differences in the same culture and language, while seven items were rated at the C level. Four items were rated at the B level, and four items showed a preference for English according to the Poly-SIBTEST method in comparing different languages in the same culture. In the OLR method analysis of different languages and cultures, five items exhibited uniform differences at the A level, while three items displayed non-uniform differences at the B level, and two items showed non-uniform differences at the C level. The Poly-SIBTEST method results for comparing different languages and cultures revealed that four items showed differences at level A, and seven items showed differences at level C. Of these, seven items showed a preference for Turkish, and four items showed a preference for English.