STAT Question

EXERCISE 8

Purpose: We are going to be focusing on using the (1) Restricted Regression adjustment model and the (2) Unrestricted Regression adjustment model for measuring treatment effects in non-RCTs. This exercise is to be handed in on Thursday, April 15 at 5:00 pm CT on Canvas. The exercise should be executed in R with you turning in your code with this exercise. The R program Exercise C17 Chapter 7 in Wooldridge 7th ed.R and the notes Program Evaluation and Regression Adjustment.pdf should contain all the information your need to complete this exercise. In this exercise we are going to be examining the effect of a training program on the subsequent earning of workers. The dependent variable is earning98, the earnings of the worker in 1998. The variable train is the treatment dummy equal to 1 if the person received the job training and 0 otherwise. The treatment adjustment variables are earn96, educ, age, and married.

(a) Report the RCT equation that you would use in evaluating the treatment effect of the training program. Report the equation in standard form. (Hopefully everyone remembers what this means.) Does this treatment effect make economic sense? Explain your answer. If the result is contrary to what you think it should be, explain why you think it is causing the illogical result.

(b) Report the Restricted Regression adjustment equation. Report it in standard form. Does the estimated treatment effect make economic sense? Explain your reasoning. What is different between the results in part (a) and this part of the exercise?

(c) Conduct an F-test of which of the two models you should prefer – the RCT model or the Restricted Regression adjustment model. What is your conclusion? Explain your reasoning.

(d) Report the Unrestricted Regression adjustment equation. Report it in standard form. Does the estimated treatment effect make economic sense? Explain your reasoning. What is different between the results in part (b) and this part of the exercise?

(e) Conduct an F-Test that allows you to decide which of the two models, Restricted Regression adjustment equation or Unrestricted Regression adjustment equation, you should prefer. Correspondingly, which treatment effect would you prefer to report, given the information you have? Explain your answer.

(f) Report the R program that you used to complete this exercise.