Analisis Karakteristik dan Kualitas Butir Soal Informatika Menggunakan Model Rasch pada Siswa SMP

Abstract

This study aims to evaluate the quality of 20 multiple-choice Informatics test items administered to Grade VII students of SMP Negeri 2 Kedungwuni using the Rasch Model with JMETRIX 4.1.1. The research employs a quantitative approach to analyze item difficulty, item fit, reliability, and the alignment between students’ abilities and item difficulty levels. Responses from 33 students were analyzed using dichotomous Rasch modeling. The results show that most items meet the fit criteria, with INFIT and OUTFIT MNSQ values within the acceptable range of 0.5–1.5. Two items require revision due to indications of underfit and overfit, while difficulty estimates range from -1.99 to 2.05 logits, reflecting adequate variation across items. Item reliability (0.6454) and person reliability (0.3104) fall into the low category, suggesting that the instrument has limited stability in distinguishing item difficulty and student ability. The person–item map reveals a moderate alignment between the distribution of student abilities and item difficulty, although several items are positioned far above or below the students’ ability range. The correlation between raw scores and theta ability estimates is very strong (r = 0.9428), indicating that raw scores adequately represent student performance, although theta provides a more accurate measurement. Overall, the instrument is feasible for use but requires refinement in terms of misfit items and enhancement in terms of difficulty distribution. These results highlight the importance of Rasch-based evaluation in improving the quality and fairness of Informatics assessments at the junior high school level.