The causal-contrast approach is a new teaching method that recruits learners’ implicit causal discovery process to improve math learning by juxtaposing information critical to discovering the goal of each solution step. Students often memorize mathematical procedures and have difficulty transferring their knowledge to novel problems. By enabling learners to infer the goal of each step, the causal-contrast approach substantially improved high-school algebra problem solving compared to a traditional instructional control (Walker et al., 2014). The present study developed Walker et al.’s instructional materials into a computer-based teaching program and tested the new approach on community-college students. The study added two new conditions: a baseline that received no instruction and a condition using a video from Khan Academy, a well-regarded online educational website representative of the traditional approach. A delayed post-test indicated that the causal-contrast condition produced greater success in solving transfer problems than the other three conditions.