Item Analysis Model for Mathematics Test Instruments Using the Item Response Theory Model
Year: 2026
Type: article
Abstract: This study aims to analyze the quality of test items in the Year-End Examination (Ujian Akhir Tahun/UAT) instrument for mathematics using an Item Response Theory (IRT) approach. The analysis focuses on identifying the best-fitting IRT model, item parameter estimation, item fit evaluation, examinee ability distribution, and the test information function. This study employed a descriptive-evaluative quantitative approach using dichotomous response data. The research subjects comprised 310 ninth-grade students from four public junior secondary schools in Watampone City during the 2025/2026 academic year, while the instrument under analysis consisted of 40 multiple-choice mathematics items. The analysis was conducted through testing the assumptions of unidimensionality and local independence, comparing the 1PL, 2PL, and 3PL models, estimating the difficulty, discrimination, and pseudo-guessing parameters, and analyzing the test information function. Data were collected using the documentation method and analyzed using R software. The results indicate that the 2PL model was the most appropriate model, as it provided the optimal balance between model fit and parameter interpretability. Based on the 2PL model fit test, 36 items were found to fit the model, while 4 items did not. Fitting and valid items are recommended for retention, whereas misfitting and invalid items should be replaced with new items, as they no longer meet the standard requirements for instrument development.
Language: English
Field: Decision Sciences
Domain: Social Sciences
SDG: Quality education
Open Access status: diamond
OpenAlex ID: W7220109804