The Impact of Active Learning on Student Course Performance in STEM Varies by Type and Intensity: A Meta-Analysis

• In Event: Illuminating Instruction and Assessment to Support Student Success in Higher Education: International Perspectives

Abstract

Research interrogating active learning in undergraduate STEM education has found significant variation in efficacy (Andrews et al., 2011; Authors, Year; Freeman et al., 2014). Some cite fidelity of implementation as the cause for this variation (Andrews & Lemons, 2015; Auerbach et al., 2018; Borrego et al., 2013), but an alternative hypothesis is that the type or context of active learning impacts efficacy. Our meta-analysis aims to update the current knowledge of undergraduate STEM active learning and examine how specific active learning practices and classroom contexts impact efficacy.

Research that contrasts lecturing with active learning is based on distinct models of learning: transmissivist theory (lecturing) contends that transmission of knowledge from an expert to a novice is how humans learn (Brockliss 1996). Conversely, active learning is grounded in constructivist theory, contending that learning happens by actively using new information to construct knowledge (Wittrock, 1992). Active learning describes approaches that engage students in the learning process through in-class activities, with an emphasis on higher-order thinking and group work (Chi & Wylie, 2014; Freeman et al., 2014).

Methods and Results
We systematically reviewed studies conducted or published from 2010-2016 that contrasted active learning and lecturing and reported on undergraduate student performance (Table 1). This date range begins where Freeman et al., (2014) ended, included roughly the same number of studies, and avoided the impacts of the COVID-19 pandemic. Included studies contrasted student outcomes on identical assessments from the same undergraduate STEM course (Table 2). We sought to understand if aspects of active learning (e.g., type, intensity) or context (e.g., discipline, class size, course-level) impacted the efficacy (Table 3). Each study was coded independently by two coders who came to consensus and evaluated additional information to minimize within- and across-study bias (Moher et al., 2009; Figure 1).

Students in active learning classes performed higher than students in traditional lecturing. There was a positive impact of all types of active learning (except studio and worksheets), with notable differences. Similarly, there were differences across intensity levels: all but Low Intensity active learning had a positive effect, meaning that dedicating more class time to active learning is correlated with higher performance on assessments. Active learning is effective in Small, Medium, and Large classes, with slightly larger effects in smaller classes. Marginally higher effects were observed in Upper Division classes, though effects were also positive in Introductory classes. See Table 4 for effect sizes and Figure 2 for a visual representation of results.

Significance
Our results are consistent with earlier meta-analyses (Authors, Year; Freeman et al., 2014; Ruiz-Primo et al., 2011) and additionally demonstrate that active learning is effective regardless of type and context, especially when used for more than one third of the class time. This reifies the “start-small conundrum,” however: most faculty developers recommend starting with a few changes before gradually increasing the time spent on student-centered teaching. Getting past low intensity active learning, where students benefit is minimal, will require individual persistence paired with institutional support (NASEM, 2025).

Authors

• Elli Theobald, University of Washington

• Shangmou Xu, University of Washington