Analysis of Pre-Olympic Middle School Mathematics Test Instruments Based on Item Response Theory

Abstract

Junior high school of mathematics Olympiad is one of the routine activities every year in Indonesia. Analysis of the items on the Mathematical Olympiad test instrument has been carried out by several researchers using the concept of classical test theory. The theory has many weaknesses when compared to the concept Item Response theory (IRT). The purpose of this article is to describe the analysis of pre-Olympiad mathematics test instrument for junior high school students in Pacitan East Java based on the IRT. This analysis was carried out using a quantitative descriptive method. IRT-based item analysis was performed using R software with the package irtawsi. The results of the analysis with the package show that the instruments used fit to 2PL model. From the 15 items used, there were two items that did not fit, namely items 4 and 9. Furthermore, from the 13 test items that were fit, there were 3 items that did not meet the quality of the discriminant parameter used. These three items are 1, 6 and 13 items, these three items must be dropped because they can provide biased information when used to estimate students' math olympiad abilities. Based on this information it can be concluded that there are 10 test items that have quality that meets the qualification parameters of difficulty and discriminant with a standard error of 0.25.