Date of Graduation


Document Type


Degree Name

Doctor of Education (Ed.D.)


School of Education


Learning and Instruction


Learning & Instruction EdD

First Advisor

Patricia Busk

Second Advisor

Xornam Apedoe

Third Advisor

Sarah Capitelli


This research study conducted a meta-analysis to integrate the studies that investigated the effectiveness of project-based learning on academic achievement based on a worldwide higher-education population. The researcher explored the variables that previous meta-analysis studies used to evaluate academic achievements, such as student grade point average (GPA), course grades, and test scores, to explore the use of these variables in project-based learning. There were 17 studies in total included in the meta-analysis evaluating the effect size of academic achievement, which was influenced in moderator variable by subject-matter areas, school location, hours of instruction, technology support, and group size. The researcher calculated Cohen's d to measure the effect sizes to investigate the standardized mean difference between the means of two groups and converted it to Hedges's g for an unbiased effect size estimate. The average effect size of 17 articles is 1.64, with a standard error of 0.42. The 95% confidence interval (CI) is (1.56, 1.72). Hence zero is not in the CI, indicating that the average effect size is statistically significant at the .05 level. The research result concluded that project-based learning had a very large effect size on improving student academic achievement in higher education worldwide. Moderator analyses were conducted for Study Design, Conditions, and Institution Type. These analyses were different with statistically significant at .05 level, contributing to the heterogeneity of effect sizes. As the five moderator variables, technology support had a positive effect on improving student academic achievement. There was a difference between an individual or 2-5 students group size and STEM or non-STEM subject-matter areas, effecting project-based learning. Studies from the Southeastern countries had the most effective results of using project-based learning in higher education. To analyze the effect of the instructional hours of project-based learning per week, the dissertation researcher needed more data. Two or fewer instructional hours, however, were more effective than the other groups. In addition to the moderator analyses, additional analyses were performed, investigating the year of publication, study limitations, and sample sizes. All three of these variables contributed to the heterogeneity of the effect sizes as they were statistically significant. Two groupings of the year of publication were assessed for differences using the analysis of variance assuming a fixed effects model. The two groupings of the year of publication, 2011 to 2017 and 2018 to 2020, yielded statistically significant differences. 2011 to 2017 had fewer studies and a very large average effect size. Study limitations were coded as one, two, three, four, or more. Three and four were combined, as there was only one study with four or more limitations. The largest average effect size was for those studies with two limitations, followed by the one-limitation and the two-limitations category. The largest average effect size was for the largest sample-size grouping and that grouping’s confidence interval did not overlap with the two smaller sample-size groupings.