In: Statistics and Probability
The following regression was run using a sample of 656 municipalities in Brazil, where the national government allocated personnel randomly to permit National Government Audit Teams to audit and publish local government finances (the taxes and spending of municipalities). The variable audit# is the number of accountants assigned to be part of the team (the average team was 4 people, but some teams were 0 and large teams were 10) and voteshare is the vote share received by the incumbent mayor in the subsequent election. Standard errors for coefficients are below each coefficient in parenthesis.
voteshare = 0.65 - 0.17audit# R^2=0.15
(1.1) (0.03)
a. What is the estimated coefficient on audit#? What does it mean in words?
b. Suppose one municipality in the sample, XX, had a team of 5
auditors and the mayor received vote share of .65 in the election.
Calculate the regression residual for that observation.
c. Test the null hypothesis that an additional audit # is not
associated with a change in vote share. What is your
conclusion?
d. Calculate and write the 95% confidence interval for the
slope.
e. Explain in words what the R^2 tells us about this
regression.