Now consider a case where the students with low test-taking ability and high test-taking ability are equally likely to get tutoring. Below is a subgroup breakdown consistent with such a scenario. Tutor = 0 Tutor=1 Total (control group) (treatment group) Ability = 0 5 7 12 (bad test takers) Ability 5 7 12 (good test takers) = 1 Total 10 14 24 Tutor = 0 (control group) Tutor=1 (treatment group) Total 2 10 12 Ability = 0 (bad test takers) Ability = 1 (good test takers) 8 4 12 Total 10 14 24 Once grades are finalized, we use our data to estimate the following simple linear regression: Grade; = a + BTutori + ei
