Electoral fraud and the (strategic) use of information: an experimental study

Abstract

Alberto Simpser distinguished between true, actual and reported turnout, where actual turnout is the share of voters who came to voting booths, the reported turnout is the share declared by the government, and the true turnout is the counterfactual notion: the share of individuals who would vote if they did not expect the falsifications. A gap between true and actual turnouts appears because the rational voter anticipating that his/her vote would be stolen, prefers to abstain. Still, many politicians find attractive the strategy to announce the forthcoming electoral fraud.
Since it is little known how the expectations of electoral fraud affect intentions of voters the aim of this paper is to answer two following research questions: does the extra publicly available information of the electoral fraud decrease the chance of the opposition to win? Do the voters for governing party and opposition react differently on the additional information about the forthcoming expected electoral fraud?
A game-theoretical model shows that the expectations of the fraud suppresses the voters' turnout, but the willingness to go to the ballot boxes is more suppressed by the fraud expectations among pro-government electorate. The proposed experimental design is aimed to test the veracity of this model in the lab.

Introduction

Does electoral manipulation reduce voter turnout? Alberto Simpser who asked this question, claimed that the previous research was inconclusive. By measuring the turnout at the district level before and after electoral reform in Mexico, he found that the electoral manipulation discouraged the voters to participate in elections. Still the results found by Simpser are far from convincing because of too many confounders that can interfere. In his model developed in *, he distinguished between true, actual and reported turnout, where actual turnout is the share of voters who came to ballot boxes, the reported turnout is the share declared by the government, and the true turnout is the counterfactual notion: the share of individuals who would vote if they did not expect the falsifications.

The idea is that there is a gap between true and actual turnouts because the rational voter anticipating that his/her vote would be 'stolen' can deduce that it is better to abstain. But despite the risk of lowering turnout among their own electorate, many politicians find attractive the strategy to announce the forthcoming electoral fraud. There can be plenty of reasons why they do it. For instance, in the recent editorial in the NYT “The Success of the Voter Fraud Myth” they wrote:
...Republican lawmakers, ... have for years pushed a fake story about voter fraud, and thus the necessity of voter ID laws, in an effort to reduce voting among specific groups of Democratic-leaning voters. Those groups — mainly minorities, the poor and students — are less likely to have the required forms of identification. Behind closed doors, some Republicans freely admit that stoking false fears of electoral fraud is part of their political strategy.

It cannot be the universal explanation though, since the voter ID strategy is rather specific for the US context, and moreover the ability of the party in opposition to push for the amendments in legislation is usually rather limited or non-existent in those countries where the electoral fraud is common.

Since it is very little known how the expectations of electoral fraud affect intentions of voters we suggest to state two (almost identical) research questions and to provide the answers by using either agent-based modelling and lab experiments:

Two possible research questions (for further discussion):

  • Can the extra publicly available information of the electoral fraud decrease the chance of the opposition to win?
    OR (which is the same but more general)
  • How the additional information about the forthcoming (expected) electoral fraud affects turnout of voters for both governing party and opposition?

Literature Review

The role of information on electoral turnout has been experimentally and theoretically investigated by considering different perspectives. Taylor et al. (2010) present a theoretical model where elections are more likely to be close (i.e. the majority is less likely to win), and voter turnout is more likely to be high when citizens possess better public information about the composition of the electorate (i.e. the distribution of political preferences is common knowledge) with respect to a situation where citizens are ignorant about the preference distribution. Bouton et al. (2016) focus on multicandidate elections under plurality rule and provide experimental evidence that giving ex ante (imprecise) aggregate information about the distribution of preferences in the electorate decreases the amount of sincere voting than when such information is not provided. Großer and Seebauer (2016) examine compulsory and voluntary majoritarian voting when individuals decide or not whether to become informed, and suggest that severe payoff losses might be associated with the curse of uninformed voting. In particular, in their experiment, each individual has to decide whether to buy a costly signal about the (randomly selected) true state of the world (A or B), before casting a vote for one of the two.

However, in their experiment the decision to be informed about the true state of the world is endogenously determined, with the fundamental assumption that information is always available to voters (conditionally on paying for that), independently on their preferences. Differently, in an election with fraud, information about the forthcoming fraud is not necessarily common knowledge and easily accessible to the entire electorate. Moreover, the majority/minority party, the one possibly advantaged by the electoral fraud, may strategically use this information in order to influence (specific) individuals' willingness to vote.

Previous empirical literature provides mixed evidence about the role of (imprecise) information about electoral fraud on voter turnout. In order to overcome the complexity of the voting environment in the real world and the unobservability of electoral fraud we use a controlled laboratory experiment. To the best of our knowledge, there is just one paper that previously examined the role of electoral fraud in the lab. Baghdasaryan et al. (2016) examine the role of (full information about) different levels of electoral fraud on voter turnout and find that i) differently than expected, turnout may increase rather than decrease as a result of a ballot box stuffing, and ii) whether the experiment is framed as an election or in neutral terms affect voters` behavior. Our experiment allows to investigate the role of aggregate uncertainty about electoral fraud on turnout. Moreover, we will investigate whether information is used strategically by a player holding superior private information about the true state of the world (i.e. whether a forthcoming election is characterized by fraud or not).