Modeling the Discrimination Power of Physics Items

Abstract

For the purposes of tailoring physics instruction in accordance with the needs and abilities of the students it is useful to explore the knowledge structure of students of different ability levels. In order to precisely differentiate the successive, characteristic states of student achievement it is necessary to use test items that possess appropriate discriminatory power. By identifying the cognitive factors, which account for differences or similarities between high achievers and low achievers, we can evaluate the efficacy of developing various aspects of physics competence within the physics instruction. Further, knowing the predictors of physics item discrimination power makes it possible to systematically modify physics items with the purpose of improving their psychometric characteristics. In this study, we conducted a secondary analysis of the data that came from two large-scale assessments of student physics achievement at the end of compulsory education in Bosnia and Herzegovina. Foremost, we performed a content analysis of 123 physics items that were included within abovementioned assessments. Thereafter, an item database was created. Items were mainly described by variables, which were supposed to reflect some basic cognitive domain characteristics of high and low achievers. For each of the items, we calculated the item discrimination power. Finally, a regression model of physics item discrimination power was created. It has been shown that 43,6 % of item discrimination power variance can be explained by factors which reflect the automaticity, complexity and modality of the knowledge structure that is relevant for generating the most probable correct solution, as well as by the constructs of cognitive load and retention. Interference effects between intuitive and formal physics knowledge structures proved to influence the item discrimination power, too.