Question 1 [50 points]

In this question, you will study crime recidivism. You should download the dataset needed for this question, “recid.csv” from Brightspace (please find the description there as well).

(a)  [5 points] Construct the variable crime by adding person and property .

(b)  [25 points] Estimate the probability of committing a crime against person or property (after a first offense), the variable from (a), on the covariates:

1. alcohol

2. drugs

3. married

4. educ

5. and a constant

With the linear probability model, the Logit model, and the Probit model. Report the regression tables (you may use the summary command and put all three of them) and comment on the sign of the estimates across the regressions.

(b)  [20 points] For each of the three models, estimate the probability of committing a crime against person or property (after a first offense), the variable from (a), for an individual who

1. has problems with alcohol, alcohol = 1

2. has a drug abuse history, drugs = 1

3. is not married, married = 0

4. has no formal education educ = 0

Question 2 [50 points]

University departments often consider teaching evaluations when hiring and promoting lecturers and pro- fessors.  However, teaching evaluations may not necessarily be fair in the sense that the exact same course could receive different evaluations based on who taught it. Among other concerns, the University is worried about discrimination against women and racial minorities.  Hamermesh and Parker (2005) collect data on course evaluations and characteristics of their instructors. Please find the data and the dataset description

on Brightspace.

(a)  [15 points] Run the following regression: