Febri Yanto, Mawarni Sulistianingsih, Destia Erfani, Khairun Nisa, Nurul Aini, Dito
The research aimed to test the effectiveness of problem-based learning in improving science problem-solving skills through a meta-analysis of summary effect-size data. The sample in this study included 25 articles that discussed learners’ science problem-solving skills through problem-based learning (PBL) application at the primary, lower secondary, upper secondary, and university levels. The fields of study include integrated science, physics, chemistry, biology, and ICT. This study used the group contrast meta-analysis method. The research consisted of an effect size analysis of the experimental and control classes of articles. The analysis used JASP software to test heterogeneity, summary effect size, and publication bias. The results showed that the summary effect size using the random effect model was 1.33, with a lower limit interval of 0.58 and an upper limit interval of 2.08. This value is in the high category with Cohen's U₃ (%) 90,82%. Therefore, it can be concluded that problem-based learning is effective for improving science problem-solving skills. The overall moderator test shows that educational level (p > 0.05) and science discipline (p > 0.05) do not have significant differences. Thus, the level of education and scientific disciplines in the application of PBL does not affect problem-solving skills. Therefore, PBL can be applied at various levels of education and scientific disciplines. The novelty of this study is that there are significant differences between individual studies (p < 0.001), which indicates the possible influence of other factors or variables that occur during the implementation of PBL in the classroom. © 2026 Fırtına Akademi A.Ş. All rights reserved.
Faculty of Mathematics and Science, Universitas Negeri Padang, Indonesia
