Impact of differential item functioning on item model fit using concurrent equating method

Abstract

Abstract This study aimed to evaluate the item model fit by equating the forms of the PISA 2018 mathematics subtest with concurrent common items equating in samples from Türkiye, the UK, and Italy. The answers given in mathematics subtest Forms 2, 8, and 12 were used in this context. Analyzes were performed using the Dichotomous Rasch Model in the WINSTEPS program. Concurrent common item equating was used for the equating process in two ways for each country: keeping and removing items showing DIF. Item model fits obtained as a result of concurrent equating analyses were compared between equated form pairs and between selected countries. As a result, when the DIF status of the items was examined, it was observed that the DIF status of the common items varied between forms. In addition, it was concluded that items displaying DIF impacted the functioning of the common items. Therefore, it is recommended that the DIF status of items be taken into account in concurrent equating, as DIF can influence the performance of common items. However, it was observed that the presence of DIF did not significantly impact the equating results when common items that did not function equivalently across forms were removed. When item fit to the model was evaluated in concurrent estimations across countries, the UK emerged as the sample where items best fit the model for Form 2. For Form 8, the best item fit was observed in the Italy sample, while for Form 12, items in both the Italy and UK samples demonstrated better model fit.

Keywords

• Rasch measurement model
• Concurrent common item equating
• Differential item functioning