A meta-analysis of teaching and learning computer programming: Effective instructional approaches and conditions. (August 2020)

• Record Type: Journal Article
• Authors: Scherer, Ronny; Siddiq, Fazilat; Sánchez Viveros, Bárbara
• Is Part Of: Computers in human behavior. Volume 109(2020)
• Publication Date: 2020-08
• Subjects: Computational thinking -- Computer programming -- Intervention studies -- Multilevel meta-analysis -- Scratch programming; Interactive computer systems -- Periodicals; Man-machine systems -- Periodicals; 004.019
• DOI: 10.1016/j.chb.2020.106349
• Languages: English
• ISSNs: 0747-5632

Abstract

Abstract: This meta-analysis maps the evidence on the effectiveness of instructional approaches and conditions for learning computer programming under three study conditions: (a) Studies focusing on the effectiveness of programming interventions per se, (b) studies focusing on the effectiveness of visualization and physicality, and (c) studies focusing on the effectiveness of dominant instructional approaches . Utilizing the data from 139 interventions and 375 effect sizes, we found (a) a strong effect of learning computer programming per se (Hedges' g ‾ = 0.81, 95% CI [0.42, 1.21]), (b) moderate to large effect sizes of visualization ( g ‾ = 0.44, 95% CI [0.29, 0.58]) and physicality interventions ( g ‾ = 0.72, 95% CI [0.23, 1.21]), and (c) moderate to large effect sizes for studies focusing on dominant instructional approaches ( g ‾ s = 0.49–1.02). Moderator analyses indicated that the effect sizes differed only marginally between the instructional approaches and conditions—however, collaboration in metacognition instruction, problem solving instruction outside of regular lessons, short-term interventions focusing on physicality, and interventions focusing on visualization through Scratch were especially effective. Our meta-analysis synthesizes the existing research evidence on the effectiveness of computer programming instruction and, ultimately, provides references with which the effects of future studies could be compared. Highlights: Computer programming interventions show moderate to large effect sizes per se. Studies focusing on physicality are especially effective. Instructional approaches exhibit moderate to large effects. Study design features only explain marginal variation in the effects. Visualizing programming with Scratch is especially effective.