Analysis of item difficulties and students' computational thinking skills assessment bias on electrolyte and non electrolyte solutions: An applications of Many Facets Rasch Model

Abstract

Skills are latent in nature, but all this time the measurement uses a ranking scale with certain criteria called ordinal data. Ordinal data is only counting, does not have a unit or distance between scores in a definite manner, and does not have zero absolute values as in the ratio data generated from physical measurements. Therefore, ordinal data is a raw score that cannot properly show one's skills. Ordinal data can be converted into ratio data using the Rasch model analysis. This study aims to analyze the difficulty of items and bias towards the assessment of students' Computational Thinking (CT) skills conducted with more than one observer on electrolyte and non electrolyte solutions. The method used was descriptive quantitative, with participants as many as 3 observers and 186 of tenth grade students in 3 high schools in Surakarta with high, medium, and low categories. The measurement uses observation assessment sheet at each meeting during the learning process which is then analyzed by the Many Facets Rasch Model (MFRM). The results obtained were that there was a bias or inconsistency of observers in assessing CT students' skills and there were differences in the number of students who could perform CT skills well based on the difficulty of items assessed by observers in the categories of high, medium and low schools. The results of this analysis can provide more accurate data on the assessment of students' CT skills.