Analysis of biology multiple-choice questions using classical test theory and Rasch model

Quality education requires accurate assessment to measure students' learning outcomes. Multiple-choice tests play a central role in biology education, but their psychometric properties require careful evaluation to ensure valid and reliable measurement. This study aimed to evaluate the quality of Biology multiple-choice test items by comparing the results of Classical Test Theory (CTT) and the Rasch model. A descriptive quantitative design involved 70 Grade 12 students and selected through purposive sampling. Data were collected using a 30-item Biology end-of-semester multiple-choice test. Students' responses were analyzed using CTT in Microsoft Excel and the Rasch model in Winsteps. CTT identified 25 valid items and five invalid items, whereas the Rasch model classified 14 items as fit and 16 as misfit. Both approaches demonstrated high reliability. CTT categorized item difficulty into two levels, whereas the Rasch model identified three levels, as supported by the Wright map and scalogram analyses. The study evaluated item discrimination using score distributions in CTT, whereas the Rasch model distinguished item difficulty from person ability using bubble charts. Only the Rasch model detected Differential Item Functioning (DIF), providing additional evidence of item fairness. These findings highlight the complementary strengths of CTT and the Rasch model in improving the quality of educational assessment.