Introduction

How do mixed legislative systems shape voter behavior? Moreover do voters understand the complexity enough to vote according to their interests? Despite the seemingly obvious impact of electoral institutions, individual-level research in mixed legislative systems is rare. The lack of emphasis on individuals leaves unexamined the extent to which citizens understand their mixed systems and whether electoral incentives alone largely predict behavior.

Mixed legislative systems elect legislators in both single member districts (SMDs) and to party seats determined by proportional representation (PR), with most allowing voters to cast separate ballots for each seat type. Since 1990 mixed systems have proliferated globally. Meanwhile a small literature suggests a convergence of Asian electoral institutions.¹ Whether the East Asian mixed systems—Japan, South Korea, and Taiwan—systematically diverge from broader theoretical expectations remains poorly understood. If the institutional effects of mixed systems are poorly understood in stable democracies, we must begin to question the value in encouraging younger democracies to enact similar systems. In contrast, if a quick adaption to electoral incentives is evident, even if citizens question their understanding of the system, this suggests promise for mixed systems in younger democracies.

This article analyzes voter decision-making in all three East Asian countries to identify to what degree citizens comprehend their mixed system and whether voting patterns seen elsewhere are mirrored in this region. Through this analysis I begin to disentangle what appears as a paradox: a sizable portion of the populace claiming not to understand their mixed system, yet minimal evidence of an inability to vote according to one’s preferences. This paper first briefly reviews voting expectations in mixed legislative systems. Next is an initial unpacking of what is meant by understanding a mixed system through various aggregate sources. Individual level survey data for all three East Asian mixed systems follows, which suggests voter sophistication that contrasts with claims of misunderstanding the system. Finally I suggest future avenues of research to tie voting sophistication and expectations within mixed systems.

The Expectation of Electoral Institutions

One of the most empirically robust findings in political science has been Duverger’s Law. Duverger (1954) argues that the mechanical and psychological effects of single-member districts (SMDs) with plurality rule lead to two-party competition whereas larger district magnitude, usually associated with proportional representation (PR), produces a multi-party system.² Mixed member legislative systems provide a new institutional context to test Duverger. While Germany was the only stable democracy to consistently use a mixed system prior to 1990, established and democratizing countries increasingly chose a mixed structure over the next twenty years—from Eastern Europe (e.g. Bulgaria, Albania, Ukraine) to Latin America (e.g. Bolivia) to East and Southeast Asia (Japan, Korea, Taiwan, Thailand, and Philippines).

Early proponents saw these systems as “the best of both worlds”³, combining the accountability and local constituency focus of SMDs with the diverse representation and national policy focus of PR. The expectation is for both tiers to act independently, with two-party competition in districts and multiparty competition for party list seats.⁴ However, critics argue that the two sets of electoral rules for the same legislature create interdependence (contamination thesis), the most common of which is the proliferation of non-viable district candidates.⁵

The East Asian cases provide a unique test for theories on mixed systems. Japan enacted a two ballot system in 1994 (the first election in 1996), with Korea and Taiwan following in 2004 and 2008 respectively.⁶ With Japan as the first adopter in the region, later converts implicitly if not explicitly considered the Japanese model and expected similar outcomes over time.⁷ Whereas most mixed systems see a proliferation of parties in the party list tier if not both seat types, all three East Asian cases saw minor reductions in the effective number of parties postreform. While the complexity of mixed systems possibly influences understanding and voter behavior, with fewer entrants voters should be able to identify viable versus non-viable district candidates. Similarly, small parties with limited resources in East Asia appear more likely to focus their efforts on the party list tier. As such, the East Asian cases should identify more clearly with the “best of both worlds” thesis than the contamination thesis and have less difficulty in adapting to the electoral reforms.

Citizens and Institutions

Theories of democracy generally assume citizen competence in terms of a basic knowledge of politics and an ability to participate, with the literature consistently showing that the more informed are also more likely to participate in politics.⁸ The public must at some level understand the political process to not only be engaged in politics but to pursue personal interests within a representative democracy. Similarly, voters require enough political knowledge and awareness to understand the effects of electoral rules and their ability to maximize their interests.⁹ Without this minimal foundation, expecting societal preferences to manifest through elections is unrealistic as voters fail to maximize the likelihood of their preferences being translated to policy. How much knowledge is necessary to participate effectively remains unclear. More broadly, whether citizens meaningfully grasp mixed systems is debatable as no systematic effort to gauge competency across multiple systems exists.

The ballot box is the quintessential democratic institution and remains the focus of citizen involvement in politics. Citizen competence is logically an accumulative process, based on various factors from formal education, attention to media, partisan attachments, elite discourse and experience under the political system—most notably previous experience voting.¹⁰ Interest in politics and education also drive political knowledge of electoral reforms in general and mixed systems in particular.¹¹

While the literature focuses on what impacts perceptions of systems more generally, similar arguments are relevant to the impact of mixed system reforms. Public knowledge is limited throughout the stages of reform, leaving much of the burden of understanding the new system on voters learning through the trial and error of actual voting. Elections function ultimately as experiments as to whether voters understand the consequences of their ballot choices on the formation of majorities or coalitions within governments. The introduction of a two-ballot system in particular requires voters to not only simultaneously digest campaign messages from both tiers (potentially exposing intraparty contradictions), but also reevaluate factors such as electoral viability and strategic voting. Confusion by the two sets of rules also potentially leads to lower levels of participation and results in contrast to society’s aggregate preferences.¹²

At this point parties can play a vital role. At the most explicit, parties directly convey the new rules as a means to encourage supporters to maximize party representation or party interests. The implementation of MMP in New Zealand increased participation and perceptions of efficacy among smaller party supporters, in part due to smaller party activities to educate voters.¹³ Similarly parties focus campaign attention on explaining party mergers or coalitional agreements. For example small coalitional partners in Germany traditionally attempt to educate voters to vote strategically in the coalition’s interests.¹⁴

In Taiwan’s first mixed system election in 2008, members of the People’s First Party (PFP) campaigned under the Kuomintang (KMT) banner in constituency seats (with 4 PFP members also appearing on the KMT party list¹⁵). This required additional efforts to ensure support for the co-nominated candidates. In contrast, new system threatened the New Party (NP) and the Taiwan Solidarity Union (TSU) with extinction even if their supporters understood the mechanics of the new system, making a public information campaign all the more crucial. In Korea, the Democratic Labor Party (DLP) bank on party list support and split ticketing to survive.¹⁶ Small parties in Japan similarly focus on the party list seats, preferring to implicitly if not explicitly support traditional coalition partners in the constituency seats rather than being locked out of power. Meanwhile larger parties—in most cases the largest two—have little incentive to explain the new system beyond generally encouraging their supporters to vote. With straight ticket voting usually in their favor and easier for voters as well, one does not expect supporters to feel the need to learn more about the mixed system compared to smaller party supporters.

Understanding the Voting Process

Self-reported rates of understanding mixed systems among the East Asian countries appear disappointing. In an original poll conducted in collaboration with the Election Study Center (ESC) at National Chengchi University in 2009, 58.88% of respondents stated they did not understand the new electoral system, with a significantly higher rate when reduced to those who did not vote in 2008 (69.48%).¹⁷ While abstention and understanding appear correlated, this conclusion still does not explain why a majority of voters claimed ignorance of the system. A 2009 survey through Keio University in Japan found that 67.2% of respondents did not find the electoral system easy to understand, presenting a major concern in that Japan conducted four elections under a mixed system by the time of the survey.¹⁸ Similar rates are evident in Korea. In a 2008 poll by the National Election Commission (www.nec.go.kr) 49.8 percent of respondents claimed they did not understand the mixed system, two elections into the two-vote system. Another NEC survey just days before the election found this rate to drop to 36.4%, still alarming if misunderstanding the system leads to representative democracy not effectively addressing the issues and concerns of voters.. Interviews of legislators and researchers in all three countries further suggest as many as half of the populations in each of these nations do not adequately understand the system.

Such low levels of understanding are surprising, but it is difficult to determine whether this is common, as few analyses examine whether citizens in mixed systems understand the system. Mixed systems experts suggest that no study directly addresses this question. While most would-be voters are exposed at minimum to public education campaigns though election commissions often supplemented by party efforts, there is scant evidence from citizens themselves.

In only a few other mixed systems do surveys ask whether people understand their system. In New Zealand, surveys asking about understanding MMP only started in 2005, with consistently between one-fifth to one-third of respondents claiming the system is difficult to understand. According to the 1996 New Zealand Election Survey (NZES), two thirds (66.2%) stated they found the system easy to understand compared to only 12.4% indicating it was hard. By 2005 only 41.89% stated the system was easy to understand compared to 28.96% identifying it as hard.¹⁹ Surveys in the Philippines suggest the public has little information about the party lists, with the high number of invalid and spoiled ballots under Thailand’s mixed system also suggesting a lack of understanding. The New Zealand Business Roundtable (BRT) argued that their system was too complex, referencing low public understanding in Germany as well.²⁰ Survey centers in other countries specifically stated they knew of no such surveys related to their system.²¹

Other disconnected pieces of information suggest areas of public confusion on technical aspects of the mixed systems. Pre-election polls in Germany in 1990 showed that a majority of respondents did not know that the party list vote determined the overall number of seats and thus was more important than the district vote, despite the system being used in West Germany for nearly forty years.²² Pre-election survey data in 2009 from the German Longitudinal Election Study (GLES) found that only 46% identified the party list seat as determining the total proportion of seats, with two-thirds (66.5%) knowing about the five percent threshold. Similarly, NZES surveys from 1996 and 2002 asked respondents to identify which vote is most important in determining the overall number of seats, and the percentage identifying correctly decreased from 67.35% in the 1996 survey to 55.61% in the 2002 survey. In all of the surveys conducted by or for the Electoral Commission, none have shown voters to correctly answer this question regarding tier importance reach eighty percent, with some as late as post-surveys in 2005 with a slim majority of correct responses (54.6%). Additionally, multiple polls show less than half of respondents could identify how legislators not elected from districts are filled (by rank on party list).

Furthermore, many election commissions show little concern about public understanding. The Central Election Commission in Lithuania acknowledges that while debates on the most effective system for representation endure, “no substantial claims about the incomprehension” of the system have emerged.²³ The Armenian Central Election Commission echoes such confidence in the public understanding, yet admits no knowledge of any study directly asking about the two-vote system.²⁴ In Bulgaria, despite just recently returning to a mixed system for the first time since 1990 and with no relevant polls, experts do not perceive ignorance of the system. Nelly Gospodinova from the Center for the Study of Democracy in Bulgaria explains it succinctly:“…it is virtually impossible for those who enter the voting booth to miss the fact, that there are two types of electoral lists - the parties' lists and the majority vote candidates.”²⁵

While the mechanical complexity of mixed systems is debatable, other factors partially explain why some citizens state they do not understand the new system. In Taiwan the Central Election Commission (CEC) arguably did not clearly explain the two-vote system. While the CEC made available a film explaining the system, few seemed aware of this service, with a Ministry of Interior officer suggesting that as high as 70% of citizens did not understand the new system.²⁶ In personal correspondence, the CEC stated it distributed 10,000 posters, 600,000 leaflets and ran newspapers ads for two days, but whether this effort was enough for a country of 23 million is unclear. The TSU claimed that 60% of citizens did not understand the system just a month before the 2008 election, after the CEC had already spent nearly 19.2 million NT (roughly $650,000) on a public awareness campaign.²⁷ Little promotion of the Korean two-vote system occurred in 2004 due to a constitutional ruling shifting from a one-vote mixed system, with the expectation that voters would acclimate by the 2008 election.

In contrast, efforts to explain the two vote systems in Germany and New Zealand were stronger and not limited to election commissions. For example, from the onset of serious election reform debate in New Zealand, an independent Electoral Referendum Panel informed the public of the pros and cons of not only the mixed proportional system but other systems under consideration. By the time of implementation, scholars, politicians and the public developed relatively clear expectations for New Zealand’s new system. Furthermore, in New Zealand and Germany, ballots include a statement that the party vote determines the overall representation in the legislature, logically eliminating some voter confusion. In Lesotho, the Independent Electoral Commission (IEC) continued to explain the system to would-be voters even at the voting precinct.²⁸ In Bulgaria, the Central Election Commission aired daily commercials and ran newspaper ads while parties engaged in similar public education campaigns. This effort partially explains the presumably higher rates of understanding in MMP systems.

A larger question still remains: what is meant by understanding the system? What Understanding is is itself ambiguous. On one hand, understanding is a general acknowledgement that a system exists or that a new system is being implemented and as such, scholars and reformers expect high rates of understanding considering government and party advertisement. On the other hand, if understanding the system means comprehending the minutia that connects an individual’s vote to the number of seats, low rates are more likely. Prof. Sung Ho Lim of Kyunghee University compared it to a soda machine. While virtually everyone knows how to operate one, they may not fully understand the mechanical operations involved.²⁹ Party officials estimated that only ten percent of the public in Korea, Japan, and Taiwan understand the mechanics behind the system, yet have little trouble in following the instructions at polling centers.³⁰ Alvaro Cabrera, head of Information and Methodology Unit for the Hungarian Parliament, stated that voters appear to “capture the essence” of the system, even if the mathematics behind it are unclear. Even in the oldest mixed system, Germany, a self-proclaimed lack of understanding the technical translation of seats pervades. In other words, the distinction between effective and technical knowledge potentially clouds assessments.³¹ The data thus presents a paradox: much of the public claims not to understand the mixed system, yet their actions are consistent with adapting to the electoral incentives of these systems.

If voters truly did not understand the system, we would expect to see evidence of misinterpreting the simple mechanics of voting. Scant evidence suggests mass confusion among voters. One way to potentially uncover confusion in two-ticket systems is the disparity in the number of total ballots in each tier.³² One would expect some voters to not have a preferred choice in both tiers, for example if the party does not run a candidate in the district or the district candidate is not aligned with a party presenting a list. Overall, scholars presume such instances are rare if voters understand the basic act of voting in a two-vote system. A cursory examination of aggregate tier deviation shows little variation.³³ For example, in Korea in 2004 there were 44,786 more district ballots submitted than party list seats in an election with over 21 million votes in each tier, a difference in turnout rates of less .12% between tiers. In the 2008 Korean election, district ballots outnumbered party list votes by 81,153 in an election with over 17 million votes in each tier. In Japan’s recent elections the difference was minimal: 255, 222 out of over 67 million in 2006 and 211,424 out of over 70 million in 2009. Aggregate survey data from 2006 also shows 99 percent of voters claimed to have voted in both tiers.³⁴ Lastly, the same pattern is apparent in Taiwan: out of over 10 million votes in both tiers, only 25,620 separated the tiers.

Similar evidence is seen outside of East Asia. In three consecutive Italian elections (1994, 1996, 2001) with over 37 million voters, the difference declined (215, 531; 144, 463; 64, 313). While the disparity increased over Germany’s last two elections, this still only amounted to less than 0.3% of votes in the larger tier. Meanwhile, in Macedonia’s 1998 and Lithuania’s 2008 elections, less than 200 votes (115 and 198 respectively) separated the tiers. Of the available data, only three examples show a discrepancy of over one percent: Nepal 2008 (2.5%), Armenia 2003 (slightly over 10%), and Philippines 2010 (almost 11%). These aggregate findings suggest that voters in most systems have little difficulty in the actual casting of two ballots.

Understanding at the Individual Level

Shifting to individual level data, the assumption remains that if voters understand the basic institutional constraints of a mixed system that they will act in accordance to tier incentives of mixed systems. Supporters of larger parties are expected to vote a straight ticket, while smaller parties have greater incentive to strategic voting in districts, but sincerely on the party list. Strategic voting is “voting not in accord with the voter’s true preference order, with the intent of bringing about a social choice more desired by the voter than the social choice that would result, other things being equal, from voting in accord with the voter’s true preference order”.³⁵ A prerequisite for strategic voting is an ability to form reasonable expectations from the electoral system.³⁶ Evidence in all three East Asian countries suggests an ability to maximize voting influence, consistent with assumptions that the incentives of electoral rules in each tier dominate voting behavior.

In the East Asia Institute (EAI) 2008 post-election survey, evidence of strategic voting is clear in Korea. The survey asks respondents to identify their choice in both the district and party vote. Straight-ticket voting dominated among respondents (59.25%) versus 40.75% voting for different parties across tiers, an increase in ticket-splitting from 20.81% in the 2004 post-election survey. Not surprisingly, almost all straight-ticket voters voted sincerely in the district tier (91.75%), choosing the same party as they identified as their party preference. Of split ticket voters, only a third voted sincerely with their district vote (31.38%). In addition, over a third of split ticket voters (38.71%) voted irrationally—that is crossing coalitional lines and voting for liberal or progressive candidates in the district and conservatives in the party list or vice versa.³⁷ This result suggests one or a combination of factors: a personal vote at the district level³⁸ while a sincere vote on the tier, a desire to create a “balance” in the legislature, a misunderstanding or ignorance of the policy placement of the individual parties, inconsistent personal views or a more general lack of understanding of the two vote system.

Regression analysis confirms voter sophistication among supporters of smaller parties (Table 1). To address split ticket voting, Model 1 employs a probit model with the dependent variable whether or not one split their ticket across parties. As small parties are traditionally vulnerable to split-ticket voting,³⁹ I include controls for each small party: the Democratic Labor Party (DLP), Liberty Forward Party (LFP), the Creative Korea Party (CKP), the New Progressive Party (NPP), and the Pro-Park Alliance (PPA). Demographic controls along with a measure of ideology are also included.

Either due to public perceptions of the system or party encouragement, the expectation is that supporters of smaller parties will split their votes by deserting their preferred party in district races. For example post-election evidence of the DLP, which performed well in the party list tier but poorly in districts, suggests strategic voting. Using the largest two parties as the base (GNP and DP), a preference for all but one of the smaller parties, the regionally-based Liberty Forward Party (LFP), was positively correlated with split-ticket voting at the .01 level. The positive coefficient on the PPA is also consistent with the perceptions of several legislators.⁴⁰ Demographic controls all failed to reach statistical significance.

Model 2 attempts to more explicitly uncover strategic voting. In this model, the dependent variable is a sincere district vote. Again leaving the largest two parties as the base, respondents whose party preference was any of the five smaller parties were all more likely to cast a strategic district vote, significant at the .01 level. In contrast, straight ticket voters (who have little reason to deviate from their preferred party) positively correlate with sincere voting at the same level. Once again, demographic and political controls failed to reach significance.⁴¹ Both models were run additionally with dummy variables for two regions, Yeongnam and Honam, traditionally associated with strong regional voting and one-party dominance (Models 1a and 2a). The results for split-ticket voting remained consistent with the original model, with only Honam negatively correlated and statistically significant at the .05 level. When adding the regional variables to sincere voting, neither reached significance. This suggests that regionalism is a secondary factor at best at explaining strategic voting. While recognizing the traditional bias in recalling one’s vote choice⁴², these findings suggest supporters of smaller parties comprehend the institutional incentives of mixed systems.

Similar evidence emerges in Taiwan. Using Taiwan Election and Democracy Study 2008 dataset (TEDS2008L) from the Election Study Center (ESC), 76% of respondents declared voting a straight ticket. Less than a third of split ticket voters were rational in the traditional sense (31.31%), with more than half (54.04%) crossing coalitional lines.⁴³ Similar models were run as those for Korea (Table 2).⁴⁴ Support for the two most relevant small parties, the New Party (NP) and Taiwan Solidarity Union (TSU), were statistically correlated with split-ticketing (Model 3). Meanwhile supporters of the People’s First Party (PFP), a party which largely coordinated on district candidates and produced a unified party list with the KMT, was not statistically significant.⁴⁵ Only one control reached significance—age—positively correlated with ticket-splitting. Similarly, the same two parties were less likely to vote sincerely in the SMD, significant at the .01 level (Model 4). Additional models included dummies for political identity for Taiwanese and Chinese using respondents who identified as both as the base (Models 3a and 4a). While both were significant only in the sincere voting model, the addition had minimal effect on the accuracy of model classification.⁴⁶

Similar models for Japan on 2005 pre-election Japanese Election Survey (JESIII) data shows comparable results (Table 3).⁴⁷JESIII indicated 78.4% of respondents intended to vote a straight party ticket. Of split-ticketers, rates of those intending to vote within a coalition versus across coalitional lines were similar (42.25% vs. 44.13%).⁴⁸ The two largest of the minor parties, Komeito and Social Democratic Party (SDP), reached statistical significance at the .01 level in both models while the Japanese Communist Party (JCP) failed to reach significance.⁴⁹ Again, with the exception of the straight ticket variable in the second model, no controls reach significance.⁵⁰

Considering the level of strategic voting here and existing evidence elsewhere, what it means when citizens say they do not understand the system requires reevaluation. I contend that instead of just measuring direct knowledge of the system, polls inadvertently tap into expectations of election outcomes, which logically differ based on partisan and demographic factors. Elections create winners and losers. Expectations differ from technical or general knowledge of the system, just as trust in the electoral process differs from expectations.⁵¹ Individuals whose favored party benefitted from reform (or whose electoral viability at least remained the same) have reason to see the new system in a more positive light then those whose preferred party does poorly. Voters cannot fully anticipate the outcomes, especially when parties intentionally misrepresent their electoral chances. For example, in 2008 the DPP in Taiwan garnered a greater share of the vote than previous legislative elections, yet their percentage of seats declined. As DPP leadership articulated an expectation of a more favorable outcome, their poor showing likely left supporters less satisfied with the system in general. Thus a gap between expectations and results potentially emerges even after the citizens understand the mechanics of voting.⁵²

Conclusion

In substantive terms, a country becomes a democracy when institutions perform according to rules and become increasingly responsive to preferences of the citizenry.⁵³ Maintaining this link and public satisfaction more generally is crucial for stability in democracies new or old.⁵⁴ From early empirical studies, it is clear that the public’s political sophistication often does not meet theoretical expectations.⁵⁵ Yet here we seem to have evidence of the public being able to quickly adapt to the mixed system rules. If mixed systems create a substantial influence on misunderstanding the electoral system, this potentially undermines calls for their adoption elsewhere. Even if small percentages of voters misinterpret the mechanical effects of mixed systems, and these voters are not randomly distributed, such misunderstanding potentially sways the outcome of an election.

Despite self-identifying as not understanding the mixed system, how the public actually votes in all three East Asian mixed systems suggests enough of an understanding to vote according to one’s preferences. Understanding how to cast votes effectively is not necessarily synonymous with an ability to gauge likely electoral outcomes or with a general satisfaction with the results or the electoral system. Little evidence emerges of widespread misunderstanding of electoral institutions, but the existing literature fails to connect practical knowledge of mixed systems (such as evidence of strategic voting) with general perceptions of the electoral system. Future research should be able to elucidate this linkage and whether electoral success influences such perceptions. Evaluating success at both the national and district levels creates an additional possibility in extending the literature on losers consent⁵⁶ to more adequately capture the dynamics of mixed legislative systems.