Predictively Improving the Problem Solving of Science Students

University Of California-Los Angeles, Los Angeles CA

Investigators

Abstract

This proposed three-year study will investigate how undergraduate and high school students make and use strategic choices as they engage in complex problem solving in chemistry. The researchers will then use this to develop effective collaborative approaches for modifying unproductive strategies. IMMEX software will sequentially capture the students' actions while they perform a series of related chemistry cases, and extract common strategies using artificial neural network technologies. These strategies will be aggregated into strategy types using evidence of the quality of student understanding, related to student ability, by Item Response Theory analysis. These analyses will suggest directed case delivery sequences for audiences with different abilities and provide an organizing framework for linking ability, and preferred problem-solving approaches. Steady state models of the development and persistent usage of particular strategies and strategy types within classrooms of students with different abilities (i.e. regular high school, AP, undergraduate) will then be developed through Hidden Markov Modeling. Here, additional student performances are predicted to have little change on the distribution of strategies used previously within a particular classroom. These models will provide baseline probabilities that students will transit from one strategy type to another on a subsequent series of IMMEX cases. These models are baseline in the sense that no particular interventions will be suggested or tested. Collaborative learning activities will then be constructed around these models to perturb these steady states. The most refined interventions will use the Intelligent Collaborative Learning System that will relate the interaction sequence details of student's communication behaviors/skills with the effectiveness of these activities in modifying strategic approaches. The research will be used to develop practical, yet effective classroom interventions that teachers can use in conjunction with IMMEX performance data to accelerate the acquisition of their students' problem-solving skills. While targeted to chemistry, the studies may be applicable to many scientific educational activities.