Improving STEM Learning: Experimental Evidence from Urban Primary Schools in India.
We study the impact of a program that aimed to improve math and science learning in Indian public schools in the city of Pune. The program recruited college students who were majoring in STEM subjects to serve as science and math teaching fellows for the entire academic year. Because the fellows had little to no teaching experience, the program relied on scripted lesson plans to deliver instruction. Relative to regular public school teachers, teaching fellows had higher scores on written assessments of instruction, understanding of student errors, and content knowledge. However, both types of instructors had similar absence rates and other measures of effort. At the end of the school year, students who were randomly assigned to a fellow performed .27-.35 SD better than control peers in math, .17-.20 SD in science. In addition, we find evidence of positive spillovers to language learning. The improvements in math and science persisted through the summer break; students in the treatment group performed .29-.34 SD better in math and .08-.10 SD better in science at the beginning of the following year. Despite these gains, the program did not improve student attitudes towards math and science or alter their plans to pursue these subjects in secondary school, higher education, or careers.