Quantitative Impact of a Cognitive Modeling Intelligent Tutoring System on Student Performance in Balancing Chemical Equations

Abstract

The need for improved interactive tutoring capabilities in educational software for chemistry problem solving is an important one clearly articulated by teachers and students. To deliver the next generation of individualized interactive capabilities users demand, it is necessary to go beyond the conventional computer-assisted instruction methodology. The focus of this paper is the assessment with first-semester general chemistry students of a recently developed artificial intelligence (AI) tutor for balancing chemical equations. This is the first such assessment of an AI-based learning tool in chemistry. Students in CHEM 121 in the Fall 2001 semester at Duquesne University (N = 273) participated in the study. Students were divided into a test group that used the AI tutor as part of their study activities and a control group that did not use the tutor. It was found that the tutor improved the performance of the test group students to a statistically significant degree, helping the weakest students the most. This study establishes the feasibility of an AI-based approach to creating advanced new tutoring software for chemistry problem solving. Access to a Web-based demonstration of the equation-balancing tutor may be obtained by emailing the corresponding author.