This project is a meta-analysis of active learning versus lecture-based instruction in higher education using multi-level methods to estimate both the weighted average effect size and its associated variance from experimental and comparative undergraduate STEM education research studies. It also models the heterogeneity in these effect sizes by including coded predictor variables such as: effectiveness, gender differences, ethnic differences, outcome measures, and explanatory moderator variables (achievement, persistence, and attitudes). Selected studies and their analyses include a focus on each discipline and the grouping of all STEM disciplines. Publication-bias is performed via forest plots and funnel plots and a sensitivity analysis is done by comparing two or more meta-analyses calculated using different assumptions regarding the primary studies.