Rasch Modeling: A Multiple Choice Chemistry Test

Abstract

The study aimed to reveal the difficulty level of items and the suitability of items of Chemistry test with the Rasch model. In addition to detecting this item quality, the Rasch model shows the student's answer pattern as well, so that the assessment can imply the quality of the instrument as an assessment of chemical learning. As many as 20 numbers of multiple-choice questions in chemical bonding material were analyzed by using WINSTEPS 3.73. The samples consisted of 200 senior high school students in Banjarmasin Indonesia. The results revealed that the average item measure was 0.00 with items (Measure Order = 4.64) which has the highest difficulty level. The Q10 was the item that has a level of conformity with the model, and outliers or misfit in Rasch were MNSQ=+0.97, ZSTD=-0.2, Pt Mean Corr=+0.58. In other words, assessment of learning with test techniques such as multiple choice based on Rasch model analysis was an effective way for teachers to review the progress of students in the learning process, guidelines for designing chemical learning strategies, and identifying students' understanding of chemical material.