We have both pre-test and post-test measures available for the study participants we can use difference-in-differences as a strategy to get causal estimates of both the ITT effect (the effect of being encouraged to watch Sesame Street) and the effect of treatment received (the effect of being a regular viewer of Sesame Street, or the "as-treated" analysis).

We first created variable representing the change in scores between the post-test (in letter recognition?postlet) and pretest (in letter recognition--prelet) and call it chglet. Here are the summary statistics for the post-test (postlet) and pre-test (prelet) scores on letter recognition (in test score points), and the change score, or ?Y (chglet), which ranges from -22 (student did much worse on post-test than pre-test) to 41 points (student did much better on post-test than pretest):