Exploring undue advantage of differential item functioning in high-stakes assessments: Implications on sustainable development goal 4

Abstract

In ensuring equity of measurement between subgroups at the item level, Differential Item Functioning (DIF) analysis is essential. However, discounting DIF affects validity, thereby making Sustainable Development goal 4 (SDG4) challenging to attain. DIF occurs when test items behave differently for distinct subgroups, potentially leading to biased outcomes and undue advantages to one group over the other. The study aims to examine DIF in high-stakes assessment among the demographic variables (such as gender, school-type, and region). The study investigates DIF in the National Examination Council (NECO) mathematics items, focusing on gender, location and school-type. The expo facto research design employed a proportionate sample size to select 14,936 responses from the 1,034,629 senior secondary school students who participated in the NECO mathematics examination. The NECO mathematics items serve as the research instrument while unveiling Cronbach alpha of 0.84 coefficient. The study revealed significant DIF across gender, location and school-type in mathematics items. The findings feature the importance of addressing DIF in high-stakes assessments to ensure equal opportunities for all students, regardless of their demographics. The study concludes that careful review should be done in developing high-stakes assessments to eliminate potential biases that may disadvantage certain groups of students. This research contributes to the ongoing discourse on assessment validity and fairness, offering insight for practitioners and policymakers to enhance the integrity of high-stakes assessments.

Keywords

• High-stakes assessment
• Bias
• Differential item functioning
• Psychometric characteristics
• SDG4