Abstract
In an exemplary test implementation, the items used should be fair and free from bias. This biased content threatens the validity that will affect the interpretation of the test results. This study aims to analyze the biased content of items and tests on the test device for the Commercial English Course, with the code ADBI4201, a course in the Business Administration Study Program at the Faculty of Law, Social, and Political Sciences, Open University (UT). It uses the quantitative approach. Data were collected through documentation in the form of grids and responses from final semester test participants in January 2024. Data analysis was carried out by (1) estimating item parameters using classical test theory and item response theory for tests, (2) estimating item parameters based on regional groups and based on gender, which are used to identify DIF content, (3) testing the significance of DIF with a maximum likelihood comparison. The study's findings showed 26 items contained DIF by gender and 25 items contained DIF by region. The results of the test bias detection showed that the test slightly favored the female group and the students from the Java region.