ANALYSIS OF COGNITIVE DIAGNOSTIC INSTRUMENT USING ITEM RESPONSE THEORY ANALYSIS FOR PHYSICS LEARNING Duden Saepuzaman *, Rizki Zakwandi, Lina Aviyanti
Physics Education Program, Universitas Pendidikan Indonesia
Abstract
Scientific conceptions are crucial in learning physics. Many important findings indicate that students experience mismatches between their conceptions and scientific conceptions. A robust instrument to measure and address this need is needed. This study aims to analyze the quality of students^ cognitive diagnostic instruments using item response theory analysis. This study uses a quantitative approach. The respondents involved were 98 high school students in the provinces of West Java and Banten. The cognitive diagnostic instrument consists of 20 items and Two-Tier Multiple-Choice (TTMC) in form with a polytomous score of 4 categories (1,2,3 and 4). Data analysis includes two stages: the IRT assumption test and then continued with IRT analysis to determine item and ability parameters. The results show that the data are unidimensional and local independence based on empirical data so that the IRT assumption is fulfilled. The IRT analysis used is the generalized partial credit model (GPCM). The results of the item parameter analysis show that all items have good discriminatory power parameters (0.304 < ai < 1.254) and are classified as good. The difficulty level analysis showed that almost all items had good step parameters ( b1 < b2 < b3), with the mean of difficulty index (b mean) in the range of -0.322 to 0.123 and categorized as moderate..This study provides an overview of the analysis of test quality using the polytomous IRT analysis for the benefit of developing the test and developing further studies.
