Analysis of Differential Item Functioning on Students’ Mathematical Problem-Solving Ability Assessment Instruments

The current problem-solving ability assessment tool lacks specified criteria for fair instrument items to measure students’ abilities across genders and grade levels. To ensure score validity, unfair or biased assessment items should be removed from the test instrument. This study analyzed the quality of assessment items for mathematical problem-solving abilities, considering variables of gender and class. This research is an instrument development study with a quantitative descriptive approach. Data collection using an online survey obtained responses from 362 high school students to 10 instrument items. The data analysis technique utilized classical test theory, employing an item functioning analysis approach through the Mantel-Haenszel method. The item validity analysis indicated that not all items were classified as biased among gender groups. Two items were identified as biased according to grade level groups. The reliability estimate was 0.829, indicating a high level of consistency in the results obtained using different samples. The study’s findings validate the bias item analysis approach and provide a reliable assessment tool for evaluating high school students’ mathematics problem-solving abilities. This study also suggests using modern theoretical analysis to identify item features that distinguish students’ ability, guessing levels, and carelessness in test item responses.