Why does peer instruction improve student satisfaction more than student performance? A randomized experiment

Studies consistently show that active learning improves student satisfaction, but results on achievement are less conclusive. We address this puzzle by comparing an active learning classroom that employs peer instruction with a traditional lecture using an experimental design in a Chilean Economics and Business school. Students in the treatment group are more satisfied with the course and have better grades, but the treatment effect varies as the semester progresses: small at the beginning, larger in the middle, and nonexistent in the final exam. Our hypothesis is that the treatment changed student effort. Students performed better under peer instruction and may have decided to decrease their effort toward the end of the semester. We present evidence consistent with this hypothesis. Additionally, students in the treatment group perceived that the most important part of the treatment was the greater interaction with the professor, and they reported studying less with their peers outside of class.