How influential is ballot design in elections?

ABSTRACT

We exploit an original dataset from a referendum in Peru to study the influence of voting “arrangements” on electoral outcomes. The relative importance of these arrangements (e.g. ballot design) with respect to the fundamentals (e.g. ideology, candidates’ quality) has not been measured. After controlling for a comprehensive set of politicians’ characteristics, we estimate unbiased ballot order effects making use of the within party variation in outcomes. We estimate a non-linear proba-bility model and we create counterfactuals to conclude that ballot design not only may have changed the electoral results but also has a greater importance than candidates’ ideology, education, experience and party affiliation.

Acknowledgements

We thank the comments from Z. Hessami, M. Machado, R. Pique, P. Riera, J. Millan, I. Ortuno-Ort'in, I. Clots, A. Loeper. D. Lougee and B. Schmidpeter. All errors are ours. Casas acknowledges the funding from ECO2017-85763-R (AEI/FEDER, UE).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Notes

1 This system of “ballot propositions” or “ballot measures” is extensively used in California, US. Typically, the propositions imply some sort of legislation or policy reform, presented in the form of referendums.

2 Alvarez, Sinclair, and Hasen (2006), Augenblick and Nicholson (2016), Ho and Imai (2006, 2008), Hessami (2016), Chen et al. (2014), Matsusaka (2016) among others. This literature finds that candidates at more prominent locations in the ballot tend to get more votes just because of the ballot design. We call this the ballot order effect.

3 In line with Selb (2008), this counterfactual estimate shows that voters are either prevented from translating their preferences into actual choices or simply cannot lead with the cognitive and/or non-cognitive cost of doing it.

4 Unless otherwise noted, we will use the term legislators for all forty politicians, including the mayor.

5 See Online Appendix for the individual characteristics of each of the 40 legislators.

6 Campbell (2005) uses the term election fundamentals but for US presidential elections.

7 All the following hypotheses should be read ceteris paribus.

8 As such, the paper is also related to the literature on political institutions in developing countries and their abuse (see for example Gutiérrez-Romero (2013) and references therein).

9 Figure 1b shows the arrangement of the coalitions, identifying each legislator’s party membership. The order of legislators in the largest coalition illustrates our point about the bargaining between the parties in the coalition. See, for example, that the FS legislators in positions 3, 8, 15 and 22, belonged to party Movimiento Teirra y Libertad, and that between them there were legislators coming from other parties.

10 Notice that while the fixed effects control for all possible district-level confounding variables, observed and unobserved, the Party Share in 2010 elections captures the district level support to each party, separately.

11 All sums and differences are statistically significant at the 95% level.

12 We only report the counterfactuals in the current version, for economy and space. The full results can be provided immediately upon request.

13 We perform the analysis by district for the 43 districts, we collect the estimated coefficients and, taking into account the districts size, we perform our counterfactual exercises.

14 Figure A2 in the Online Appendix.

15 See the online appendix for the details of the simulation.

16 PPC vote share in 2010: 37.6%. PPC vote share in 2013: 29.6%. It should be noted that in 2010 the PPC-UN coalition included party Perù Possible and one the legislators elected with PPC-UN came from Perù Possible. However, in 2013 Perù Possible ran on its own and got 11%. Since Perù Possible ran on its own in 2013 we do not include its single legislator elected with PPC-UN in PPC-UN’s 2013 numbers.

17 We thank the anonymous referee for highlighting this issue. Additionally, there is a lack of data that prevents us from analyzing the short-spanned untainted performance of the incumbents.

Additional information

Funding

This work was supported by ECO2017-85763-R (AEI/FEDER, UE). AEI stands for “Agencia Estatal de Investigacion” while FEDER stands for “Fondo Europeo de Desarrollo Regional”.