ESTIMASI PARAMETER TES DENGAN PENSKORAN POLITOMUS MENGGUNAKAN GRADED RESPONSE MODEL PADA SAMPEL KECIL

Abstract

Multiple choice tests with dichotomous scores are very often used by educators because it is easy to apply. Unfortunately, this scoring lacks an opportunity for educators to diagnose students' mistakes. This study aimed to analyze the items test using graded responses model (GRM). The data in this study was the response of the economics test that have been tested to 96 first-year students of economic education. The test form was multiple choice item using politomous scoring with ordinal scale 1-2-3-4-5. Test scoring use polytomous scores. Data was analyzed with descriptive quantitative techniques using PARSCALE application. The results showed that there are 50% items test that fit with the GRM model. Furthremore, the application of GRM model in this study seems less suitable because only 20% that have good quality. Big sample size is recommended when using the GRM model to obtain accurate estimation.