- Consider A Scenario Where You Know The Following Information Fraction Moved Fraction Arrested 0 20 0 70 Mto Treatment M 1 (34.35 KiB) Viewed 29 times
- Consider A Scenario Where You Know The Following Information Fraction Moved Fraction Arrested 0 20 0 70 Mto Treatment M 2 (19.83 KiB) Viewed 29 times
- Consider A Scenario Where You Know The Following Information Fraction Moved Fraction Arrested 0 20 0 70 Mto Treatment M 3 (29.7 KiB) Viewed 29 times
Consider a scenario where you know the following information. Fraction Moved Fraction Arrested 0.20 0.70 MTO Treatment MTO Control 0.20 0.30 Consider running potentially three sets of linear regressions. Your first regression is looking at moving, a binary outcome. What is the coefficient on a dummy variable for being in the treatment group (or beta from the equation below)? movingi = Bo + B. Treatment;
You could also estimate a regression with arrest likelihood as an outcome when the dummy variable for treatment as the regressor. (This regression estimates what is commonly called the intent to treat effect). What would the coefficient on treatment be for that regression? arrest; = Q0 + a treatment;
Take your estimates for the last two questions. You could estimate the causal effect of moving on being arrested. This is the samething as using MTO as an instrument for moving. While you estimate this in R, you don't have to. You can also estimate it with the last coefficients from the last two equations (which you found by hand). Simply take Hint: (In other words, if only half the people move because of treatment, you should double the intent to treat estimate to recover the causal effect of moving).