The Dif Identification in Constructed Response Items Using Partial Credit Model

Abstract

The study was to identify the load, the type andthe significance of differential item functioning (DIF) in constructed responseitem using the partial credit model (PCM). The data in the study were thestudents’ instruments and the students’ responses toward the PISA-like testitems that had been completed by 386 ninth grade students and 460 tenth gradestudents who had been about 15 years old in the Province of Yogyakarta SpecialRegion in Indonesia. The analysis toward the item characteristics through thestudent categorization based on their class was conducted toward the PCM usingCONQUEST software. Furthermore, by applying these items characteristics, theresearcher draw the category response function (CRF) graphic in order toidentify whether the type of DIF content had been in uniform or non-uniform.The significance of DIF was identified by comparing the discrepancy between thedifficulty level parameter and the error in the CONQUEST output results. Theresults of the analysis showed that from 18 items that had been analyzed therewere 4 items which had not been identified load DIF, there were 5 items thathad been identified containing DIF but not statistically significant and therewere 9 items that had been identified containing DIF significantly. The causesof items containing DIF were discussed.

Keywords

• dif
• polytomous data
• partial credit model