Evaluation findings
We attempted to measure the effect of attending the after-school all-star tutoring program on students’ final GPA. We know that some other variables may confound this relationship, so we started by closing these paths. Location, Siblings (as a measure of background) and GPA (as a measure of ability to learn) can affect the probability of a student to attend the afterschool program. In fact, students who are far from school may not be able to attend the afterschool program as it is in the school building, and they cannot stay late because of transportation issue. Also, students with many siblings will be more likely to attend the tutoring as they parents may not be able to work with every child. Finally, previous performances in class may lead one to enroll or not in the program. Students with lower GPA may lean toward enrolling in the program. Based on our confounders, we predicted the probability of every student in our dataset to enroll in the program or not. After predicting these probabilities (weights), we used them to estimate the effect of our program on the final GPA.
Our regression estimate is 0.754, meaning the program increased the final GPA by 0.75 points on average, which is the effect of the Afterschool tutoring program.
