This study aims to determine the expected effect size of intervention studies in science lessons through meta-analysis. Intervention studies were collected from education-center websites in every Japanese prefecture to calculate the average effect size and examine the moderation effect. The results of the quantitative analysis showed that the mean effect size of multi-valued items was g=0.594 [0.557, 0.630] (k=626, N=9122). The moderator analysis revealed relatively low effect sizes for learning in the geology domain, and differences in effect size for various types of academic indicators. In addition, we provided basic statistics to help determine the sample size needed for future studies.
