Quantitative Methods in Policy Analysis

Fall 2007

Problem Set #5

Use the UMCPsalaries.xls dataset to explore the difference between the salaries of male and female professors at the University of Maryland College Park (as of 1997).

A.

Calculate the mean salary of male professors as well as a 95% confidence interval for this mean.

NOTE: Although these professors represent the entire population of male professors at UMCP in 1997, they can be considered a sample of all of the possible male professors at UMCP in 1997, just as flipping a coin can be considered a sample of all possible coin flips.

B.

Repeat part A for female professors.

D.

Is this hard evidence of gender discrimination in salaries on the part of UMCP? What other variables (both those present in the dataset and those not present) might account for some of this difference?

2.

Use the welfare.xls dataset to answer the following questions about the income of participants in the Arkansas welfare-to-work experiment:

A.

What was the average income of control-group participants in the second year after random assignment? Give a 95% confidence interval.

HINT: Create a new column/variable called earn58 = (earn5+earn6+earn7+earn8). Then calculate the mean and standard deviation of earn58 for those in the control group, and use these to construct the confidence interval. Note that you can do this very quickly with a pivot table.

B.

Repeat for the experimental group. Compare your answer to part A. What do you conclude?

C.

Did the experimental group have a higher average income than the control group in the second year? How much higher? Give a 95% confidence interval for the difference in income.

D.

Your answer to part C is an estimate of the effect of the experiment on the average income of welfare recipients. Describe how you could use the earnings in the year prior to random assignment (pearn1+pearn2+pearn3+pearn4) to improve this estimate.

E.

Extra credit: Do the calculation described in part D.

Note: Make sure that you do these calculations only for participants in the Arkansas experiment.