Abstract. We investigate student-chosen, multi-level homework in our Integrated Learning Environment for Mechanics [1] built using the LON-CAPA [2] open-source learning system. Multi-level refers to problems categorized as easy, medium, and hard. Problem levels were determined a priori based on the knowledge needed to solve them [3]. We analyze these problems using three measures: time-per-problem, LON-CAPA difficulty, and item difficulty measured by item response theory. Our analysis of student behavior in this environment suggests that time-per-problem is strongly dependent on problem category, unlike either score-based measures. We also found trends in student choice of problems, overall effort, and efficiency across the student population. Allowing students choice in problem solving seems to improve their motivation; 70% of students worked additional problems for which no credit was given.
Abstract. Item response theory is useful in both the development and evaluation of assessments and in computing standardized measures of student performance. In item response theory, individual parameters (difficulty, discrimination) for each item or question are fit by item response models. These parameters provide a means for evaluating a test and offer a better measure of student skill than a raw test score, because each skill calculation considers not only the number of questions answered correctly, but the individual properties of all questions answered. Here, we present the results from an analysis of the Mechanics Baseline Test given at MIT during [2005][2006][2007][2008][2009][2010]. Using the item parameters, we identify questions on the Mechanics Baseline Test that are not effective in discriminating between MIT students of different abilities. We show that a limited subset of the highest quality questions on the Mechanics Baseline Test returns accurate measures of student skill. We compare student skills as determined by item response theory to the more traditional measurement of the raw score and show that a comparable measure of learning gain can be computed.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with đź’™ for researchers
Part of the Research Solutions Family.