Political Science 399
Fall 1995
ASSIGNMENT #5
Bivariate data analysis:
Using the “vote92.sav” dataset on the diskette I gave you, please answer the following questions. If you have difficulty accessing the data or performing any of the functions, feel free to talk to me outside of class and/or in the computer lab (ACD 205, Friday, Nov. 17, from 12:00-1:30).
1. Present (i.e., create and print out) a crosstab of respondent interest in the campaign by how much attention they paid to newspaper articles about the campaign. (Note that I am not giving you the variable numbers; this means, I guess, that you’ll have to look them up.)
Judging from this table, what is the relationship between these two variables?
2. To get a more precise estimate of the relationship between newspaper reading and interest in the campaign, choose an appropriate bivariate measure of association, report its value, and interpret it.
Hopefully you selected one of the following statistics: gamma, Kendall’s tau-b, or Kendall’s tau-c. Which of this is/are appropriate for this relationship, and why?
3. Now let’s assume that the pre-and post-election feeling thermometers for the three major candidates (variables 29 through 34) are interval-level measures. Under the “statistics” menu, highlight the “correlate” and then highlight the “bivariate” choice. The highlight the “Pearson’s” box, and put all 6 variables (v29 through v34) into the “variables” box. Then go into the “options” area and click on the “means and standard deviations” box. Now click on “OK.”
This would give you the means and standard deviations of the candidates’ feeling thermometer scores, plus a matrix of the correlation between these variables.
OVER
Look first at the mean feeling thermometer scores: these means tell a story about the fortunes of the various candidates over the course of the campaign(remember, each candidate has two feeling thermometer
ratings: one taken prior to the election , and one taken afterward). What is that story?
Now look at the correlation matrix. How strongly is each candidate’s pre-election thermometer rating correlated with his own post-election rating?
What would explain the differences between the strength of these relationships?
What is the correlation between the Bush and Clinton post-election thermometer scores?
How would you interpret this result?
What is the correlation between the Bush and Perot post-election thermometer scores?
How would you interpret this?
What is the correlation between the Clinton and Perot scores?
How would you interpret this?
Please turn in this sheet plus all printouts by Monday, Nov. 20.