Math 121 Elementary Statistics Fall 2000

Special Assignments:

One of the assignments is for you to do a descriptive analysis of the actual vote totals by county, and the other is for you to perform inference procedures based on exit poll surveys.  You will randomly be assigned the two states that you will analyze.  Please make sure that at least one of the states has at least fifteen counties and an election for Senator or Governor as well as for President.  The data that we will use are from CNN's web site at: www.cnn.com/ELECTION/2000/results/.  You will use Minitab to carry out some of the analyses.

Please do not hesitate to ask questions of me if you have any trouble accessing the data or using Minitab.

Your assignments will be graded for their clarity of presentation and exposition as well as for the thoroughness and accuracy of your analysis.  Both assignments are due by Friday, December 15.

Specific instructions follow:

1. Descriptive analysis of county-by-county results
1. Preliminaries:
1. Choose a state with at least fifteen counties and with a Senator or Governor's race in 2000.
2. Decide which of the two major candidates (Gore or Bush) you will study.
3. If there was both a Senator and a Governor election, decide which race you will study.
4. Gather and organize data on the percentage who voted for this Presidential candidate in each county and the percentage who voted for the Senator or Governor candidate of the same party as the Presidential candidate.  You are strongly encouraged to enter the data into a Minitab worksheet and to save the file to a disk.
2. Univariate analysis: Examine the county-by-county percentages voting for the Presidential candidate you chose.
1. Display these percentages with a dotplot, histogram, and boxplot.  (Be sure to label your graphs clearly.)
2. Calculate descriptive statistics (mean, standard deviation, five-number summary) of these percentages.
3. Write a paragraph describing the distribution, concentrating both on general patterns and on deviations from the patterns.  (Be sure to relate your comments to the context.)
3. Bivariate analysis: Analyze the relationship across counties between percentages voting for the Presidential candidate and the Senator/Governor candidate of the same party.
1. Construct a scatterplot with the Presidential candidate's percentage of the vote on the vertical axis and the Senator/Governor's percentage on the horizontal axis.  (Be sure that the graph is labeled clearly.)
2. Write a paragraph describing the association between these two variables.  Comment on general patterns and on departures from them.  Also mention on whether most of the counties' values fall above or below a "y=x" line and comment on the relevance of this question.
3. Determine the least squares line for predicting the percentage voting for your Presidential candidate based on the percentage voting for the Senator/Governor candidate of his same party.  Report the equation of the line and also the value of r^2.  Also comment on how well the line appears to model the relationship between these two variables.
2. Inferential analysis of exit poll data
1. Preliminaries:
1. Identify the state for which you will analyze exit poll results.
2. Report how many people were sampled in the exit polls of that state.
3. Decide which of the two major candidates (Gore or Bush) you will study.
4. Two of the variables that you will study are gender and whether the voter has a college education.  Look through the list of questions asked in the exit poll and select one additional question, for which there are only two possible responses, to study.
5. For these three variables, record the marginal distributions for each.  (For example, what proportion of the respondents were women and what proportion were men, what proportion were college educated and what proportion were not, and so on.)
6. For these three variables, record the conditional percentages of voting for your candidate.  (For example, what proportion of  women voted for your candidate and what proportion of men voted for your candidate, and so on.)
2. Univariate analyses:
1. Use the exit poll data as a sample from which to conduct a test of whether the sample data provide strong evidence that the actual proportion of voters in the state who were women differs from one-half.  Report the hypotheses, test statistic, and p-value.
2. Use the exit poll data as a sample from which to form a confidence interval to estimate the actual proportion of voters in the state who were women.  Choose whatever (reasonable) confidence level you like.
3. Write a paragraph summarizing your findings.
4. Repeat the previous three questions for the "college educated?" variable and for the other variable that you select.
3. Comparative analyses:
1. Conduct a test of whether the proportions voting for your candidate differ significantly (at the .05 level) between women and men voters.  Report the hypotheses, test statistic, and p-value.
2. Form a 95% confidence interval to estimate the difference in proportions voting for your candidate between men and women.
3. Write a paragraph summarizing your findings.
4. Repeat the previous three questions for the "college educated?" variable and for the other variable that you select.