Michael S. Hanson Department of Economics

HOME

TEACHING

RESEARCH

LINKS

OTHER

Economics 300: Problem Set #10

The following problem requires the use of EViews, a statistical package available on the PCs in the PAC Lab and most other labs on campus. If you need help with EViews, you should ask the TAs for assistance, consult the built-in help from within EViews, and/or browse one of the "tutorials" listed on the Econ 300 Resources Page. I also may be able to answer questions during office hours. Determinants of GPA Load the workfile GPA2.WF1 (or the zipped version) into EViews. The data are for a sample of several hundred students at a large midwestern university, and the variables are defined as follows: colgpa = Cumulative college GPA at end of the current semester sat = Combined SAT score tothrs = Total credit hours as of the beginning of the current semester hsrank = Graduating class rank in high school hssize = Total number of students in high school graduating class (in 100s) Create a new variable called hspct that corresponds to a student's percentile ranking in their graduating class. That is, hspct should be 5 if a student's rank was the top 5% of their class. Why is this variable a better one to include in the regression than hsrank alone? Hint: This variable is a function of the hsrank and hssize variables. Estimate a multiple regression equation for GPA that depends on the following factors: SAT score, total number of credit hours completed, and percentile ranking in high school. Print out the "estimated output" from EViews. Briefly explain the signs (positive or negative) of each estimated coefficient. Do any of the signs surprise you? Why or why not? Briefly explain whether you would expect, a priori, any of these variables to not belong in the regression. Which explanatory variables (regressors) are statistically insignificant, if any? Do your answers to these two questions coincide? Explain. Now regress college GPA on SAT score alone, and print the EViews output. Is this coefficient biased? Explain. Hint: What are the sign and magnitude of the indirect effect of SAT scores on GPA? If you could include one additional variable not currently in this workfile, what would it be? Explain why you think this variable belongs in the regression, what you hypothesize the sign of its coefficient to be, and how it might affect the estimated coefficients on the other variables. Comments: You may discuss this problem set with your classmates, but do your own work. Submitting any part of another student's work as your own is a violation of the Honor Code and is not acceptable. Print your name and Wes ID number on each page you submit. You should place all of the written answers on one page. (If you want to place more than one graph on a page, you may do that as well.)