Psychometric Evaluation of an Applied Mathematics Examination Instrument Based on The Rasch Model
Administering the same examination to students from different programs can result in uneven measurement of competence and potentially flawed academic decisions. In many vocational settings, psychometric reviews of examination instruments are uncommon, increasing the risk of inaccurate evaluation. This study investigates the Applied Mathematics Final Semester Examination (FSE) instruments in the Mechanical Engineering Department, Bali State Polytechnic, through the Rasch model. The data came from 206 students who answered 40 multiple-choice questions. The test was delivered online via the institutional e-learning system but taken in the classroom. Winsteps 5.9 was used to assess construct validity, reliability, separation indices, item difficulty in relation to ability, and Differential Item Functioning (DIF) across programs. Demonstrates that the instrument meets all psychometric standards. Nine components that didn’t fit nonetheless showed the planned construct. The separation indices were 3.46 and 6.18, and the person and item reliability were both over 0.90. The major dimension explained more than 63% of the variation, which indicated that it was one-dimensional. Usually, the difficulty levels were right for the students, but not at the extremes. Eight things suggested that there might be bias. The result is a one-of-a-kind Rasch evaluation of a vocational mathematics FSE that combines reliability, unidimensionality, and DIF analysis. It can be used as a model for other situations and to make vocational math tests fair, valid, and reliable.