How Voters Use Grade Scales in Evaluative Voting

Abstract

Abstract:During the first round of the 2012 French presidential election, participants in an in situ experiment were invited to vote according to "evaluative voting", which involves rating the candidates using a numerical scale. Various scales were used: (0,1), (-1,0,1), (0,1,2), and (0,1,...,20). The paper studies scale calibration effects, i.e., how individual voters adapt to the scale, leading to possibly different election outcomes. The data show that scales are not linearly equivalent, even if individual ordinal preferences are not inconsistent. Scale matters, notably because of the symbolic power of negative grades, which does not affect all candidates uniformly. Keywords:Evaluative Abstract During the first round of the 2012 French presidential election, participants in an in situ experiment were invited to vote according to "evaluative voting", which involves rating the candidates using a numerical scale. Various scales were used: (0,1), (-1,0,1), (0,1,2), and (0,1,...,20). The paper studies scale calibration effects, i.e., how individual voters adapt to the scale, leading to possibly different election outcomes. The data show that scales are not linearly equivalent, even if individual ordinal preferences are not inconsistent. Scale matters, notably because of the symbolic power of negative grades, which does not affect all candidates uniformly. Classification JEL: D72, C93

A Characterization of Approval Voting Without the Approval Balloting Assumption

“…other types of considerations that can be invoked for the practical purpose of designing a voting procedure. For instance, experimental evidence suggests that the labeling of ballots has a psychological meaning to voters that can affect the output of a voting procedure, and that voters value the possibility of expressing non-dichotomous opinions even though the additional information is not exploited when determining the winner of the election (see for instance Baujard et al (2018)). Similarly, legal or political considerations may also point at the direction of enhancing the expressive freedom of voters, e.g.…”

Designing preference voting

“…Note however that none of them did so in in the context of PR with preference voting. Baujard et al (2017) for example asked a sample of French voters to evaluate the candidates to the 2012 presidential election using various scales including those covered in the present study such as the one in which voters can give a negative point. They show that although these scales affect the way subjects evaluate some of the candidates, the aggregate result remains unchanged.…”

“…To address this issue, we include all of the observable personal characteristics of the candidates in the same regression, as to estimate the unique effect of each of them. 14 Note that unlike other voting experiments of the same kind, each subject had to vote according to one of the three scales, instead of all of them one after the other(Baujard et al 2017; Darmann, Grundner, and Klamler 2017). We believe our design is superior as it shuts down possible contamination effects between treatments.…”

Some regrettable grading scale effects under different versions of evaluative voting

“…But under evaluative voting the consequences are not so clear, because it is difficult to know which grades best reflect voters' opinions: this raises a problem of representation in the sense of the theory of measurement (Narens 1985). For evaluative voting, this issue has been termed the calibration problem by Baujard et al (2018).…”

“…Voters prefer relatively longer scales, and they prefer scales with negative and positive grades as compared to scales with positive grades only. One might suppose that, from among a range of different scales, evaluative voting rules differ only as a result of a direct or indirect labeling effect: the results or relative scores of candidates vary due to the specific ways in which the labels of grades are interpreted by voters (Baujard et al 2018). Up to now, the state of the art has thus focused on the labeling effect, suggesting that length would only count through its indirect consequences on the labels of grades, and protocols thus far have not been able to go beyond this vague assertion.…”