The information function of mixed-format test of physics learning outcomes measurement

Abstract

This study investigates the function of item information on mixed-format tests on the measurement of physics learning outcomes. The item information function is an important attribute in the item response theory approach that describes the accuracy of the measures taken. In this research, the mixed-format test used is a physics test consisting of 30 multiple choice format items and five constructed-response formats. The data analyzed in this study were responses from 300 junior high school students from six schools. The sample was selected using the Purposive Sampling Method. The mixed format test response data were analyzed using item response theory models, namely the Multiple-Choice Model and the Graded Response Model. Calibration is done in two ways: simultaneously and separately, then the results are compared to find out the value of higher information functions. The results of the analysis show that the information function of the items carried out simultaneously for the mixed format test is higher than the information function produced by the calibration done separately. Therefore, the combination of multiple-choice and graded response models is highly recommended for analyzing mixed-format tests; this combination increases the accuracy of capability measurement.