An Empirical Study of Voting Rules and Manipulation with Large Datasets

The study of voting systems often takes place in the theoretical domain due to a lack of large samples of sincere, strictly ordered voting data. We derive several million elections (more than all the existing studies combined) from a publicly available data, the Netïflix Prize dataset. The Netïflix data is derived from millions of Netïflix users, who have an incentive to report sincere preferences, unlike random survey-takers. Using these elections, we investigate theoretically-possible phenomena, including different voting rules yielding different winners, and the possibility of irrational aggregated preferences. We investigate the occurrence of statistical patterns of interest, such as single- peakedness. In addition, we look at the practical manipulability of elections according to several recently-proposed manipulation algorithms. [This talk includes research developed jointly with Nicholas Mattei and James Forshee.]