You recently conducted a survey of French citizens about their vote intent in th

Question

You recently conducted a survey of French citizens about their vote intent in the second round of their presidential election. You ask them for whom they’ll vote, as well as some demographic and attitudinal questions. Your data is as follows: • vote: 1 = Macron; 2 = Le Pen; 3 = Undecided; 99 = Skipped/no response • sex: 1 = male; 2 = female; 99 = Skipped/no response • ideology: 1 = very liberal; 2 = liberal; 3 = somewhat liberal; 4 = neither liberal nor conservative; 5 = somewhat conservative; 6 = conservative; 7 = very conservative; 99 = Skipped/no response • eu: 1 = France should more strongly integrate with the European Union; 2 = France’s integration with the European Union is about right; 3 = France should withdraw from the European Union; 99 = Skipped/no response

1. You are interested in studying how sex, ideology, and EU skepticism drive presidential vote choice. Given the described data, what type of model (e.g., linear, Poisson, negative binomial, logistic, probit, et cetera. . . ) would you estimate? Why?

2. Before actually fitting your model, would you need to recode any of the variables? If so, what would you do to each? (Just describe the recoding; there’s no need to provide any code.)

3. Imagine that you hypothesize that EU skepticism (i.e., wanting France to leave the Union) would cause a voter to choose Marine Le Pen, even after controlling for respondent gender and ideology. How would you assess your theory using the model you chose above? Be as specific as possible. In particular, keep in mind the estimates your model would produce, which you would consider, and the criteria by which you would judge them in evaluating your theory

Explanation / Answer

1. You are interested in studying how sex, ideology, and EU skepticism drive presidential vote choice. Given the described data, what type of model (e.g., linear, Poisson, negative binomial, logistic, probit, et cetera. . . ) would you estimate? Why?

Since there are three options for the Presidential candidate, multi-nomial (logistic) regression would be appropriate here.

2. Before actually fitting your model, would you need to recode any of the variables? If so, what would you do to each? (Just describe the recoding; there’s no need to provide any code.)

Since most software allow specification of categorical variables, in regressand as well as the regressor, re-coding is not essential anymore.

3. Imagine that you hypothesize that EU skepticism (i.e., wanting France to leave the Union) would cause a voter to choose Marine Le Pen, even after controlling for respondent gender and ideology. How would you assess your theory using the model you chose above? Be as specific as possible. In particular, keep in mind the estimates your model would produce, which you would consider, and the criteria by which you would judge them in evaluating your theory

We would need to first create a variable about the respondent being skeptical about the EU and then look at the proportion among the two categories voting in favor of Le Pen or otherwise. Another alternative is the look at the sign of the coefficient of this factor variable in the fitted logistic regression model.

Related Questions

Navigate

• Browse (All)
• Browse Y
• Subjects

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.