The psychological effect of electoral systems in founding elections

Abstract

In this research note, I revisit the conventional wisdom about when the psychological effect of electoral systems is observed. I rely on data from the first presidential and legislative elections in 45 third- and fourth-wave presidential and semi-presidential regimes to show that the psychological effect of electoral systems manifests itself in the first election. The effective number of legislative parties is significantly higher than the effective number of presidential candidates in the first election when the electoral system in legislative elections is more permissive than in presidential elections.

Keywords

coattails electoral law number of parties psychological effect

Since Duverger (1951), electoral system scholarship has established that electoral systems have mechanical and psychological effects. The former refers to their under-representation of small parties and the latter refers to the strategic behaviour of politicians and voters before the election: if the mechanical impact of the electoral system is anticipated, (some) non-viable parties may not run candidates or make alliances with other parties, while some voters will desert hopeless parties in favour of major parties (Blais and Carty, 1991; Blais et al., 2011). The interplay of the two effects reduces the number of parliamentary parties through the mechanical effect and the number of electoral parties through the psychological effect (Taagepera and Shugart, 1989: 65). In Rae’s (1971) terminology, the mechanical effect is the proximal effect of electoral systems and the psychological effect is the distal one.

In this article, I revisit the conventional wisdom about when the psychological effect is observed. The basic insight is that the mechanical effect occurs during a single election but the psychological effect takes at least two elections to manifest itself (Taagepera and Shugart, 1989: 65). The tabula rasa perspective in postcommunist countries, for instance, presents ‘an image of institutional experimentation that echoes the chaotic environment of the transition [and then] stresses the newness of the democratic experience and the propensity to form weak and fluid party configurations’ (Bielasiak, 2002: 192 and 189). According to this perspective, political actors have no knowledge regarding the political landscape and the electoral strength of competitors in founding elections. Founding elections are dominated by a considerable uncertainty regarding three crucial variables of electoral democracies: the number of parties competing in the election, the vote shares of the different parties and the consequences of the electoral rules. These dimensions are interdependent, in the sense that the uncertainty attached to any of them makes prediction of the others impossible (Andrews and Jackman, 2005: 67; Gallego et al., 2012). Tavits (2007: 116–117), for instance, argues that:

when there is very little or no electoral history,…it is not clear at the time of entry which parties will be viable on election day. Given such uncertainty, every potential entrant is perceived as having as good a chance of winning as any other, and voters can only vote sincerely.

A less radical perspective, however, would argue that the psychological effect of electoral systems is not completely absent in founding elections. This effect is weaker in founding elections due to limited information, but information is never completely absent, especially in the third- and fourth-wave democracies, given the existence of opinion polls, news analyses, candidates statements (Cox, 1997: 78) or a previous democratic experience before the current democratic period (Lago and Martínez, 2012).

The empirical evidence about the psychological effect of electoral systems mainly comes from studies focused on time series modelling of election dynamics. As a result of a learning process about parties and candidates and their electoral prospects, the strategic behaviour of parties and voters increases over time after the founding election. Based on results from 2007 districts in 183 lower chamber elections across 21 countries, Crisp et al. (2012: 151–152) find evidence of a very steep learning curve: the first couple of elections show strong evidence of coordination issues, which subside quickly until the fifth election, when coordination becomes indistinguishable from that in more experienced settings. The duration of democracy is associated, for instance, with a significant decrease in the number of votes for non-viable candidates or the effective number of electoral parties (Best, 2010; Crisp et al., 2012; Duch and Palmer, 2002; Tavits, 2007; Tavits and Annus, 2006).

Clearly, existing research strongly supports the developmental argument of the psychological effect: coordination increases over time after the founding election. However, from this empirical evidence, it is not possible to infer whether political actors were behaving sincerely in the founding election or whether they were already behaving strategically. The existing research about electoral systems has been mostly silent about how important is the psychological effect in founding elections. In other words, there is robust empirical evidence showing that the slope of the psychological effect is positive (i.e. β₁ > 0), but the hypothesis that β₀ = 0 in the first election (i.e. a constant equal or almost equal to 0 and no strategic behaviour at all or scarce) is still an assumption which to the best of my knowledge has never been tested. The counterfactual that moving from a restrictive to a permissive electoral system does not make any difference to the behaviour of politicians and voters in the founding election is crying out for an empirical test. This is the gap filled by this research note.

Empirical analysis

First elections in third- and fourth-wave presidential and semi-presidential regimes provide an ideal scenario to ascertain the timing of the psychological effect of electoral systems. Presidential elections are held under a majoritarian system, plurality or majority, while elections for the lower house can take place under a majoritarian, Proportional Representation (PR) or mixed-member electoral system. Therefore, in some countries, the carrying capacity of the presidential and the legislative electoral systems is similar, while in other countries (especially those using PR), the legislative electoral system is more permissive than the presidential one.¹ If parties and voters do not respond to the different incentives provided by electoral rules and do not behave strategically in the election, the effective number of electoral parties (ENEP) and the effective number of presidential candidates in the first round (ENPRES) should be similar even when the carrying capacity of the electoral systems is different.² This is the null hypothesis. However, if the psychological effect manifests itself in the first election, ENEP should be greater than ENPRES when the carrying capacity of the legislative electoral system is greater than that of the presidential electoral system. This is the alternative hypothesis.

The sample of presidential and semi-presidential countries employed here includes 45 third- and fourth-wave democracies in which voters cast different votes for the presidency and the legislature (see the Online Appendix for a description of the sample). Given that the psychological effect crucially depends on the availability of public and shared information, the focus on third- and fourth-wave democracies allows us to control for the role of the mass media. As Mainwaring and Zoco (2007: 167) argue, ‘in countries with weak democratic heritages, television became a mass phenomenon before parties were deeply entrenched in society. Candidates for executive office can get their messages across on television without the need to rely on well-developed party organizations.’ Additionally, I select the first presidential and parliamentary election when the country becomes a democracy. If data about the number of parties and candidates in the first democratic parliamentary and presidential elections are not available, the country is dropped from the sample. A regime is classified as a democracy at the time of an election if (i) the chief executive is elected, (ii) the legislature is elected, (iii) there is more than one party competing in the elections and (iv) an alternation under identical electoral rules has taken place (Cheibub et al., 2010). The source for all the data used in this article is Bormann and Golder (2013).

It can be argued that this research design is challenged by reverse causality and presidential coattails. First, it might be the case that countries that are more diverse (linguistically, culturally or socially) tend to adopt more than others PR systems in legislative elections, which are supposedly more favourable to minority interests. However, this is not an issue for the empirical analysis. Given the limited or the lack of information about the carrying capacity of presidential and legislative in founding elections, diversity should equally affect the fragmentation of presidential and legislative party systems. Second, if coattails running from presidential to legislative elections already exist in founding elections, fragmentation in both elections should converge. However, coattails play in favour of the null hypothesis of no effect of electoral systems.

In the sample, there are 20 presidential regimes, 24 semi-presidential regimes and 1 parliamentary regime (Moldova) but where presidential elections are also held. Only in six countries, the electoral formula used in the presidential election is plurality. The numbers of presidential candidates in the first presidential election in countries using plurality and majority are quite similar, 2.57 and 2.92, respectively. Not surprisingly given the low number of countries using plurality, the difference is not statistically significant (t = 0.60). In the legislative elections in the sample, 9 countries use plurality, 10 use a mixed-member electoral system and 26 use PR.

The difference between ENEP and ENPRES in the sample of 45 countries is shown in Figure 1. As can be seen, there are substantial differences between ENEP and ENPRES in most of the countries, and there are no big differences between countries with and without simultaneous presidential and legislative elections. There are three interesting results. First, the mean of ENEP is 4.42 and the mean of ENPRES is 2.86. The (paired) difference is statistically significant at the 0.01% level (t = 4.49). Second, when we focus exclusively on the 15 countries with simultaneous presidential and legislative elections, the pattern is similar: the mean of ENEP is 4.23, while the mean of ENPRES is 2.93. The (paired) difference is statistically significant at the 0.05% level (t = 2.89). Finally, in the 30 countries with non-simultaneous elections, the mean of ENEP is 4.52 and the mean of ENPRES is 2.82. The (paired) difference is statistically significant at the 0.01% level (t = 3.57). In sum, the results clearly show that fragmentation is greater in the first legislative election than in the first presidential election and thus the null hypothesis of no coordination is rejected: the psychological effect already manifests itself in the first election.

Figure 1.

Comparing the effective number of electoral parties in legislative elections and the effective number of presidential candidates in 45 countries.

In Figure 2, I examine how the difference between ENEP and ENPRES varies across electoral systems in legislative elections. According to my argument, and given that the carrying capacity of all the presidential electoral systems in the sample of countries is largely similar, I expect that the difference between ENEP and ENPRES should increase the more permissive the electoral system in legislative elections is. Therefore, the greatest difference between ENEP and ENPRES should be observed in those countries using a PR system. As expected, countries using PR systems in legislative elections have a greater difference between ENEP and ENPRES (the median is 1.24 and the mean is 1.87) than those using mixed-member (the median is 0.63 and the mean is 1.13) and majoritarian systems (the median is 0.04 and the mean is 1.17). The difference between ENEP and ENPRES is statistically significant at the 0.01% (t = 3.84) and 0.05% (t = 3.17) levels in those countries using PR and mixed-member electoral systems, respectively, but not in those countries using a majoritarian electoral system (t = 1.19).

Figure 2.

Comparing the effective number of electoral parties in legislative elections and the effective number of presidential candidates across electoral systems.

In Table 1, I estimate how the electoral system affects the difference between ENEP and ENPRES when controls for two standard institutional variables affecting the independence of presidential and legislative elections are added: the powers of the president and the temporal proximity between elections.³ According to existing research about coattails in presidential regimes (Golder, 2006; Hicken and Stoll, 2011; Stoll, 2015), the effect of presidential elections on legislative fragmentation increases the more powers the president has, and when presidential elections are held in temporal proximity to legislative elections, no matter if presidential elections are held subsequent or prior to a legislative election. The two variables are defined as follows. First, as presidents in presidential regimes tend to have more powers than in semi-presidential ones, I create a dichotomous variable, presidential regime, which takes the value 1 if the regime is presidential and 0 if the regime is semi-presidential or parliamentary. Second, following Stoll (2015), the temporal proximity of presidential and legislative elections is measured as the logarithm of the absolute number of days between the dates of the first legislative and presidential elections plus one. A value of zero indicates that the elections are simultaneous. The greater the value, the less temporally proximate the presidential and the legislative elections are. In the sample of countries, the variable ranges from 0 in 15 countries to 7.65 in Sri Lanka (the legislative election was held on 15 February 1989 and the presidential on 9 November 1994).

Table 1.

The determinants of the difference between the effective number of electoral parties in legislative elections and the effective number of presidential candidates.

	ENEP–ENPRES (1)	ENEP–ENPRES (2)	Log (ENEP/ENPRES) (3)	Log (ENEP/ENPRES) (4)	ENEP (5)	ENEP (6)
Legislative electoral system (refer majoritarian)
PR	1.384*** (0.434)	1.324* (0.696)	0.291** (0.136)	0.250 (0.179)	1.347*** (0.429)	1.366* (0.694)
Mixed	0.768* (0.449)	−0.114 (0.390)	0.160 (0.177)	−0.085 (0.161)	0.656 (0.404)	−0.161 (0.370)
(log) Number of days	−0.055 (0.088)	−0.124 (0.075)	0.002 (0.022)	−0.019 (0.041)	−0.059 (0.087)	−0.108 (0.066)
Presidential regime	−0.945* (0.551)	−0.927* (0.540)	−0.419*** (0.137)	−0.419*** (0.134)	−0.691 (0.516)	−0.669 (0.491)
PR × (log) Number of Days		0.027 (0.130)		0.012 (0.049)		0.003 (0.130)
Mixed × (log) Number of Days		0.238* (0.129)		0.064 (0.058)		0.218* (0.114)
ENPRES					0.738*** (0.228)	0.739*** (0.229)
Constant	0.861 (0.546)	1.071** (0.480)	0.375** (0.168)	0.450** (0.168)	1.528* (0.881)	1.700** (0.790)
N	43	43	45	45	43	43
R ²	0.178	0.206	0.263	0.286	0.309	0.333

Note: Estimation is by Ordinary Least Squares (OLS). Robust standard errors are in parentheses.

*p < 0.10; **p < 0.05; ***p < 0.01 (two-tailed).

The key independent variable is the electoral system employed in the legislative election, legislative electoral system. This is a categorical variable that takes one of three values indicting the basic type of electoral system used in the election: (1) majoritarian, (2) PR and (3) mixed. The category of reference is majoritarian. Differences between dependent and independent mixed electoral systems are not explored in this note as there are only 10 countries using this type of electoral system.⁴

Apart from the difference between ENEP and ENPRES, I test the robustness of the results using two additional measures of the dependent variable. First, I use a relative measure of the dependent variable, the logarithm of (ENEP/ENPRES). When the ENEP/ENPRES is equal to 1, the fragmentation of presidential and legislative elections is similar; when the values of the ratio are greater than 1, legislative elections are more fragmented than presidential elections, and the other way around when the values of the ratio are lower than 1. The reason for using the logarithm is that when ENEP is greater than ENPRES, the ratio may vary between 1 and infinity. Hence, the resulting variable is highly skewed and the estimated effects are not additive. Second, ENEP is the dependent variable and ENPRES is included on the right side of the equation.

I run two models for each definition of the dependent variable: an additive model with the three independent variables and an interactive model in which an interaction term between legislative electoral system and number of days is added to the previous specification.⁵ The expectations is that the difference between ENEP and ENPRES should increase when using a permissive legislative electoral system and the less temporarily proximate presidential and legislative are:

\begin{array}{l} Y_{i} = β_{0} + β_{1} {legislative electoral system}_{i} \\ + β_{2} {number of days}_{i} + β_{3} {presidential regime}_{i} + ε_{i} \end{array}

\begin{array}{l} Y_{i} = β_{0} + β_{1} {legislative electoral system}_{i} + \\ + β_{2} {number of days}_{i} + β_{3} {presidential regime}_{i} + β_{4} legislative electoral system \\ \times n {umber of days}_{i} + ε_{i} \end{array}

When the dependent variable is ENEP, ENPRES is added to the two models as independent variable.

It is immediately clear in Figures 1 and 2 that two observations, Comoros in the sample of countries using a majoritarian system in legislative elections and Poland in the sample of countries using a PR system in legislative elections, are outliers. Whereas the difference between ENEP and ENPRES is 8.77 for Comoros and 10.01 for Poland, it is only 1.20 (the standard deviation is 1.62) for the remaining 43. When running the models with the difference between ENEP and ENPRES as the dependent variable, the Studentized residual for Comoros is 3.26 and for Poland it is 3.33, while all the residuals for the remaining observations do not exceed 1.93. Therefore, when the dependent variable is the difference between ENEP and ENPRES and ENEP, the models have been estimated after dropping the observations from Comoros and Poland. There are no influential outliers when the dependent variable is the logarithm of (ENEP/ENPRES).

As expected, the three additive models show that in countries using PR and mixed-member systems in legislative elections, the difference between ENEP and ENPRES is greater than in countries using a majoritarian system. The dummy for PR systems is statistically significant at the 0.05 or 0.01% levels: in the first and fifth models, the effective number of electoral parties in legislative elections is about 1.3 greater in countries using PR than the ENPRES; in the third model, the effective number of electoral parties in legislative elections is about 34% greater in countries using PR than the ENPRES. The dummy for mixed-member systems is only statistically significant (at the 0.1% level) in the first model. The three additive models show that the fragmentation of presidential and legislative elections is more similar in presidential regimes than in semi-presidential ones. The dummy for presidential regimes is statistically significant in the first and third model at the 0.1 or 0.01% levels, respectively. When ENEP is the dependent variable (fifth model), there are no differences between presidential and semi-presidential regimes. The reason is that part of the effect of the dummy for presidential regimes is captured by ENPRES. Finally, the temporal proximity between presidential and legislative elections does not significantly affect the difference between ENEP and ENPRES.

Interestingly, the interaction between using a mixed-member system in legislative elections and the temporal proximity between presidential and legislative elections has the expected positive sign and is statistically significant at the 0.1% in two of the models (the second and the sixth): the less temporally proximate presidential and legislative elections are, the greater the difference between ENEP and ENPRES when a mixed-member system is used in legislative elections. The effect of PR systems on party system fragmentation does not depend on the temporal proximity between presidential and legislative elections.

Electoral system scholarship establishes that the psychological effect of electoral systems takes at least two elections to manifest itself or at least that it is weak in the first election. However, to the best of my knowledge, this assumption has been never tested. When comparing the first presidential and legislative elections in third- and fourth-wave presidential and semi-presidential regimes, I have found that the effective number of electoral parties is greater than the ENPRES when the electoral system employed in legislative elections is more permissive than in presidential elections. Both the mechanical and the psychological effects of electoral systems are then proximal. Finally, in order to know how much of the total amount of the psychological effect is already observed in the first elections, longitudinal analyses are crucial. As here I am only considering the first presidential and legislative elections, I have shown that there is a meaningful psychological effect from the very beginning, but it is not possible to determine whether this is the tip of the iceberg or most of it.

Supplemental material

Supplemental Material, PPQ706342_Appendix - The psychological effect of electoral systems in founding elections

Supplemental Material, PPQ706342_Appendix for The psychological effect of electoral systems in founding elections by Ignacio Lago in Party Politics

Footnotes

Acknowledgements

I thank Marina Costa Lobo and Agustí Bosch for very helpful comments on a previous draft. I gratefully acknowledge the Salvador de Madariaga fellowship provided by the Ministerio de Educación, Cultura y Deportes of Spain for a research stay at the Instituto de Ciências Sociais da Universidade de Lisboa, where this article was written.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Ministerio de Economía y Competitividad of Spain under the research project CSO2013-40723-C2-1-R.

Supplemental material

Supplementary material for this article is available online.

Notes

References

Andrews

Jackman

(2005) Strategic fools: electoral rule choice under extreme uncertainty. Electoral Studies 24: 65–84.

Best

(2010) Increasing irrationality? The equilibrium relationship between electoral and legislative party system size. Electoral Studies 29: 105–116.

Bielasiak

(2002) The institutionalization of party systems in postcommunist states. Comparative Politics 34: 189–210.

Blais

Carty

(1991) The psychological impact of electoral laws: measuring Duverger’s elusive factor. British Journal of Political Science 21: 79–93.

Blais

Lachat

Hino

Doray-Demers

. (2011) The mechanical and psychological effects of electoral systems: a quasi-experimental study. Comparative Political Studies 44: 1599–1621.

Bormann

Golder

(2013) Democratic electoral systems around the world, 1946-2011. Electoral Studies 32: 360–369.

Cox

(1997) Making Votes Count. Strategic Coordination in the World’s Electoral Systems. New York: Cambridge University Press.

Cheibub

Gandhi

Vreeland

(2010) Democracy and dictatorship revisited. Public Choice 143: 67–101.

Crisp

Olivella

Potter

(2012) Characteristics of electoral contexts that impede voter coordination. Electoral Studies 31: 143–158.

10.

Duch

Palmer

(2002) Strategic voting in post-communist democracy? British Journal of Political Science 32: 63–91.

11.

Duverger

(1951) Les Parties Politiques [Political Parties]. Paris: Seuil.

12.

Gallego

Rico

Anduiza

(2012) Disproportionality and voter turnout in new and old democracies. Electoral Studies 31: 159–169.

13.

Golder

(2006) Presidential coattails and legislative fragmentation. American Journal of Political Science 50: 34–48.

14.

Hicken

Stoll

(2011) Presidents and parties: how presidential elections shape coordination in legislative elections. Comparative Political Studies 44: 854–83.

15.

Lago

Martínez

(2012) Forgetting to make votes count: the role of previous democratic experience. Electoral Studies 31: 413–421.

16.

Mainwaring

Zoco

(2007) Political sequences and the stabilization of interparty competition: electoral volatility in old and new democracies. Party Politics 13: 155–178.

17.

Rae

(1971) The Political Consequences of Electoral Laws, 2nd ed. New Haven: Yale University Press.

18.

Stoll

(2015) Presidential coattails: a closer look. Party Politics 21: 417–427.

19.

Taagepera

Shugart

(1989) Seats and Votes. The Effects and Determinants of Electoral Systems. New Haven: Yale University Press.

20.

Tavits

(2007) Party systems in the making: the emergence and success of new parties in new democracies. British Journal of Political Science 38: 113–133.

21.

Tavits

Annus

(2006) Learning to make votes count: the role of democratic experience. Electoral Studies 25: 72–90.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.18 MB