Final Exam

POSC 202A

Prof. Collingwood

Instructions

Please answer the questions and show your work. Attach an R script to the end of the document or on a separate page. Please turn the exam into my box by 5:00pm on 12/17/14. Please do not share notes on this exam. But if you have any questions, you can email me, although my internet access sometimes is really spotty over the weekend…weird, I know.

Exam

This exam is similar, but more challenging than the midterm exam. Instead of simply answering the questions below, write up an analysis of the 2008 contest between Jay Nixon and Kenny Hulshof. Pretend you are a pollster/advisor to Jay Nixon, who is Missouri’s governor now but candidate in 2008. In your memorandum, include tables and figures/plots where relevant. Below are the basic topics you should cover. But broadly consider how well voters know Nixon, whether they are generally warm to him, and how likely he is to win the election. Which traits are favorable to him and among which voting blocs. In the end, where should Nixon place his resources and what should he say? The memo should be anywhere from 10-15 double-spaced pages not including graphics. You are free to do a little background research on the candidates and the state of Missouri to help place the data into context.

Read the dataset “mosw.dta” into R. This dataset is a survey of likely voters in Missouri during the 2008 elections. The survey inquires about horse-race questions, ratings of political and public figures, as well as message testing. In writing your memo, incorporate the following points, perhaps as sections, including the necessary tables and graphs. You can do more than what is asked of below if you feel like you need to buttress your memo/argument, but you do not necessarily have to do so.

Name Recognition

1. Is Nixon a well-known candidate, relative to other statewide candidates and nationally known politicians? Report the percentages of respondents who can identify Jay Nixon. What about Kenny Hulshof? You should also create a bar chart of Nixon stacked against other candidates and politicians (Bush, McCaskill, Blunt, Obama, Hulshof, and McCain). [NOTE: Assume that the order of THERM2_ in the dataset matches the order in the layout]

-In R, create a bar chart comparing the percentage of respondents who can identify Nixon (and not) to the same for Hulshof. Use the barplot() function, more about graphing here:

-In R, create another chart stacking politicians mentioned above against one another. Do this for their mean thermometer rating.

How personal of a vote does Nixon have?

2. How well do voters really know Jay Nixon? Conduct a correlation analysis on Obama’s therm and Nixon’s therm. Why would you do this to see if Nixon has a personal following? Be sure to consider how missing data will affect the correlation (hint: you will need the option use=”complete.obs” to the cor function if there is missing data). Report the results both in terms of the correlation coefficient but also a standard dot plot with a regression line to indicate the general trend. Include the plot’s correlation coefficient in the text on the plot.

The Vote

3. Which candidate, Jay Nixon or Kenny Hulshof is more popular among the electorate? Use the “govvot” and “govvote” variables to create a new total vote variable. Which candidate has deeper support (cert2)? Is the race more fluid when compared to the presidential contest (so you should compare the two contests)? That is, are voters still making up their mind?

In answering the above, examine how the vote is divided by gender, party identification (ptyid1), ideology (ideo1), race (white/non-white categories), and education (educ) – bin this last one into three categories: High School or less; Some College, College/Post-grad. Be sure to code missing data (don’t know/refused) to NA. Then, generate crosstabs using the CrossTable() function, where you should also include a chi-2 test (option chisq = T)

Candidate Traits

4. Thinking about the better job battery (BETJOB2), which candidate traits work in Nixon’s favor; which traits work in Hulshof’s favor? Offer an explanation based on your findings. Subset the data to independents voters, do your findings remain the same or do they change?

Statistical Analysis

5. Finally, conduct a regression analysis between Obama’s Therm and Nixon’s Therm ratings. How well does Obama’s Therm do in predicting how favorable people are to Nixon? How much variance in the dependent variable is explained (R2-squared)? Include an interpretation of the slope, y-intercept and whether the variable is statistically significant.