Rasch model item response theory (IRT) to analyze the quality of mathematics final semester exam test on system of linear equations in two variables (SLETV)

Abstract

A high-quality test has a balanced level of difficulty and can be completed by the respondent with their level of abilities. This study analyzed the test instrument used to measure students' mathematics abilities in the semester final exam on System of Linear Equations in Two-Variables. The purposive sampling technique was applied to select the respondent students (N=195). The test items were twenty multiple-choice questions. The researchers performed the data analysis using Rasch model Item Response Theory (IRT) approach with the QUEST program. The analysis revealed that the twenty items’ validity matched the Rasch model with a range of INFIT MNSQ values between 0.89 – 1.17. Items on the final semester exam can be used based on the estimated OUTFIT t-value less than equal to 2.00. The OUTFIT t analysis obtained nineteen qualified items and one unqualified item.