How to Tamper with a Parliament: Strategic Campaigns in Apportionment Elections

TL;DR

This paper investigates the computational complexity of strategic voter campaigns in apportionment elections, analyzing how small vote changes can influence outcomes, and proposes a second-chance voting mode as an alternative to thresholds.

Contribution

It provides a comprehensive complexity analysis of bribery strategies in apportionment systems and introduces a second-chance voting mode to mitigate threshold effects.

Findings

Optimal campaigns can significantly influence election outcomes.

Heuristic strategies are less effective than optimal campaigns.

Second-chance voting alters the complexity landscape.

Abstract

In parliamentary elections, parties compete for a limited, typically fixed number of seats. Most parliaments are assembled using apportionment methods that distribute the seats based on the parties' vote counts. Common apportionment methods include divisor sequence methods (like D'Hondt or Sainte-Lagu\"e), the largest-remainder method, and first-past-the-post. In many countries, an electoral threshold is implemented to prevent very small parties from entering the parliament. Further, several countries have apportionment systems that incorporate multiple districts. We study how computationally hard it is to change the election outcome (i.e., to increase or limit the influence of a distinguished party) by convincing a limited number of voters to change their vote. We refer to these bribery-style attacks as \emph{strategic campaigns} and study the corresponding problems in terms of their computational (both classical and parameterized) complexity. We also run extensive experiments on real-world election data and study the effectiveness of optimal campaigns, in particular as opposed to using heuristic bribing strategies and with respect to the influence of the threshold and the influence of the number of districts. For apportionment elections with threshold, finally, we propose -- as an alternative to the standard top-choice mode -- the second-chance mode where voters of parties below the threshold receive a second chance to vote for another party, and we establish computational complexity results also in this setting.