Experimental Design
This study adopted a stratified, three-arm, cluster randomized control trial. In selecting the target schools for this study, we first selected 60 public elementary schools with relatively low performance for each of the 5 regional education offices in the prefecture based on data from the national learning achievement survey in the 2022 school year, and whose 4th grade class sizes were 4 or less in the 2023 school year. This selection criterion followed the request of the board of education of the prefecture. Then, among the selected schools, we randomly selected 4 schools per regional education office each that would be assigned math teachers("1st treatment group"), 4 schools that were assigned science teachers ("2nd treatment group"), and 4 schools that were assigned neither teachers ("control group"), for a total of 12 schools per regional education office.