Analyzing the Practical Relevance of Voting Paradoxes via Ehrhart Theory, Computer Simulations, and Empirical Data

F. Brandt, C. Geist, M. Strobel

Election type Ordinal
Culture Impartial Anonymous Culture
Candidates [2-10]
Voters {50, 51}
Instances 1000000
Parameters None
The Mallows model parameter is selected to be a worst-case scenario based on the theoretical analysis presented earlier in the paper. The urn model assumes we put back exactly $\alpha$ copies of a drawn vote.
Notes The whole preflib at that time gave 314 meaningful elections. Additionally from the Netflix Prize dataset (nowadays also preflib) 11 million suitable elections derived by Mattei et al (An empiri- cal study of voting rules and manipulation with large datasets) were used.