POL221
Methods and Statistics in Political Science
November 21, 2005

Problem Set #5
(UPDATED, 12/2/05, with additions or changes underlined)

1. The Stata data set called 2004nes.dta contains information about Americans’ attitudes during the 2004 presidential election.  The codebook for the data set contains more information about the survey and the data set more generally.  After reviewing the codebook, access the data set in Stata, and use it to answer the following questions.

   1. Using the dataset, propose a regression model explaining President George Bush's approval rating. Use two independent variables in your model, and explain the causal mechanism that (you believe) links each independent variable to the dependent variable (i.e., why are they related). Then, estimate your model, using Stata. Try to use two independent variables that produce significant coefficients.

   2. Interpret all parts of the Stata regression printout that we have discussed in class.  For each independent variable, make sure that you follow Ritchey's six steps of hypothesis testing.  You can combine some parts of the hypothesis test for all variables.  For example, you could include the null and alternative hypotheses for all independent variables in the same section of your answer.

   3. With this model, use the first differences method to calculate (by hand or in Excel) predicted values of the dependent variable (Bush’s approval rating) associated with relevant independent variables.  Again, make sure that you show the steps in your calculations, to ensure getting partial credit. Also, include a one-sentence interpretation of each pair of predicted values, in which you decribe in words the pattern suggested by the predicted values.

   4. Reestimate the linear regression model, dropping one of the independent variables.
              1.    How do the results of the model change from the previous model?
              2.    Exactly why does the remaining independent variable produce a different coefficient in the second model?  
                          a.    Section 5.2.1 in KKV (especially equation 5.3 and note 9) can help explain the logic behind the change.

Make sure that your answer includes the printouts that you obtained from Stata.  You can answer all questions by working together with others.  You should type your answers, using 1-inch margins and a font no smaller than 10 point.  Unlike previous problem sets, this assignment does not have a page limit.  The first page of your paper should still only contain identifying information (your name, etc.).  The text of your answers should begin on the second page.  The paper should contain appropriate in-text citations for any sources used (including KKV and any other commonly used sources).  The detailed information for each citation should then appear in a bibliography on a separate page.  The problem set is due at 9:30 a.m. or 10:30 a.m. (depending on your section) on December 5, in the box on my office door (CHA2039). Papers are late if turned in more than 10 minutes after this time.  Paper grades are lowered by ten points (out of 100 total points) for each 24-hour period (after the start of class) that they are late.  Finally, the Honor Code binds all answers; make sure that you pledge your work.