POLI 272 BAYESIAN METHODS
Fourth Assignment
Due 29 October 2009


All homeworks must be neatly typed with Microsoft Word or its equivalent with all R and WINBUGS code shown in an Appendix.

  1. Follow the instructions in Running Regressions in STATA and WINBUGS and run regressions using 3 of the Excel spreadsheets posted on Excel Spreadsheet Download Page. In particular, pick 3 that are at least 20 years apart but none before H50.XLS (1887-88 U.S. House of Representatives)! For example, H60.XLS, H70.XLS, and H80.XLS are each 20 years apart.

    1. Show the results from STATA and the corresponding results from WINBUGS in the fashion explained in Running Regressions in STATA and WINBUGS and shown in the handout Reagan Presidential Support Example for WINBUGS (PDF).

    2. Show the statistics for the Betas and Sigma with updates at 1000, 10,000, and 100,000. How do the results change?

    3. Show the Density plots for the Betas and Sigma. Do these make sense?

    4. Show the Auto Correlation plots for the Betas and Sigma. Do these make sense?

    5. Show the History plots for the Betas and Sigma. According to the WINBUGS Manual, "Checking Convergence" (near the end), do the chains look like they have converged?

  2. For one House File (Not all three!), run the following Probit in both STATA and WINBUGS. Treat partydum as the dependent variable and run:

    probit partydum x2 southdum

    Report the STATA results and repeat 1a) - 1e) for this problem as well.

    h105_BUGS_probit_model.txt -- WINBUGS model illustrating how to use Probit