Experimental Impact Evaluation Design – Quiz 4

1. Suppose that you have data on a key covariate at baseline and you want to maximize balance across treatment groups on that covariate in your random assignment. What is the best way to maximize balance?

2. Using the treatment assignment that you generated in Quiz 2, do a balance test on total_bl. What is the mean difference in baseline test scores between treatment and control groups? Round to the nearest 0.01.

3. Now do a balance test on all of the baseline test score variables (total_bl, hindi_bl, math_bl, english_bl). Be sure to do the joint test rather than four independent tests. What is the p-value on the F-statistic from testing the joint likelihood that the coefficients on all 4 variables are 0? Round to the nearest 0.01.