Rule formation in simulation-based discovery learning: optimized clustering based on Levenshtein edit distance

This study investigates how tools for formulating rules affect learning in a simulation of series electric circuits. Computer simulations can enhance exploratory learning but pose challenges to testing hypotheses, designing experiments, and interpreting data. I analyzed students' engineering tactics and search strategies in a simulation supplemented with tools guiding how to formulate rules. Participants were randomly assigned to a control or one of two experimental groups. Detectable strategy differences between groups were observed. Sequence analysis, leveraging Levenshtein edit distance, K-means clusters, silhouette coefficient, and generalized median method, revealed unique learning paths labeled Reinforced Confirmers, Dual-mode Strategy Diversifiers, Multi-strategy Jugglers, Self-regulated Revisers, and Methodical Integrators. This research contributes insights about effective instructional strategies for discovery learning in simulations, particularly how to improve knowledge integration and self- regulated learning in complex scientific domains.