Improving Estimates in Likely Voter Models: How the Feeling of Duty and the Feeling of Guilt Perform in a Voter Turnout Index?
Abstrakt
Research problem: A correct estimate of voter turnout is one of the assumptions of a good likely voter model (Sturgis et al., 2017). The basic method to determine likely voters is a question about turnout intention.
In addition to this question, other variables such as past voting behavior, socio-demographic characteristics, or interest in politics are used to identify voters and non-voters. However, as various studies show (e.g.
Dimock et al., 2001), these variables do not necessarily improve the overall turnout index, and they only extend the questionnaire and burden the respondent rather than contribute to the accuracy of the turnout estimate. Therefore, it is necessary to carefully choose which variables are included in the index and what the assigned turnout weights should be.
With the development of behavioral sciences, discussion about the influence of emotions on electoral behavior and voter turnout has begun. This paper aims to examine whether the inclusion of questions about the feeling of duty to vote and the feeling of guilt due to abstention of voting can improve the estimate of who will turn out to vote and produce a more accurate estimate in the likely voter model for political parties.
Data and methodology: Our paper uses data from the Making Electoral Democracy Work project. These datasets are from Swiss, German, Spanish, French, and Canadian national and federal elections (N = 17,305), and each consists of a pre-election and post-election wave.
We used logistic regression to estimate the model of voter turnout, which includes the feeling of duty to vote and the feeling of guilt due to the abstention of voting. We compare this extended model of voter turnout with the basic model and see how the results of these models differ from the election results in case of particular political parties.