Perceived political polarization and political participation

Political polarization, defined as the increasing ideological divide between different groups or individuals within a society, often resulting in intense partisanship, gridlock, and an inability to compromise on key issues (Iyengar, Sood, and Lelkes Reference Iyengar, Sood and Lelkes2012; Prior Reference Prior2013; and many others). In recent years, many countries have experienced a significant rise in political polarization, with citizens increasingly identifying with and supporting political parties and candidates that align with their values and beliefs.

As one of the major topics of extensive recent research, there is a growing body of literature examining the causes and consequences of political polarization. Scholars have identified several factors contributing to political polarization, including the role of media bias and echo chambers (Garrett Reference Garrett2009; Levendusky Reference Levendusky2013), the rise of social media and digital communication (Barberá et al. Reference Barberá, Jost, Nagler, Tucker and Bonneau2015; Lee, Rojas, and Yamamoto Reference Lee, Rojas and Yamamoto2022; Tufekci and Wilson Reference Tufekci and Wilson2012), and the effects of economic inequality and economic globalization (Akdede Reference Akdede2012; Voorheis, McCarty, and Shor Reference Voorheis, McCarty and Shor2015). In addition, some scholars have argued that political polarization is driven by deeper structural and cultural divisions within society, such as racial and ethnic tensions, and the decline of traditional social institutions (Fukuyama Reference Fukuyama2018). These divisions create a sense of identity politics and group-based animosity that fuels public perceptions of political polarization.

In terms of the behavioral consequences of polarization, some research has indicated that polarization leads to increased political engagement and participation among citizens who are strongly aligned with a particular ideology or party (Enders and Armaly Reference Enders and Armaly2019; Roblain and Green Reference Roblain and Green2021). However, other studies have suggested that polarization has negative effects on political behavior, such as reduced willingness to compromise, increased hostility towards political opponents, and decreased political trust and social trust (Hetherington and Rudolph Reference Hetherington, Rudolph and Uslaner2018; Lee Reference Lee2022; Theiss-Morse, Barton, and Wagner Reference Theiss-Morse, Barton, Wagner, Bornstein and Tomkins2015). In addition, scholars have argued that polarization may contribute to stalemates and inefficiencies in the policymaking process, as politicians are increasingly unable to work together to pass legislation (Mason Reference Mason2018). Overall, the effects of political polarization on political behavior are complex and multifaceted, and they depend on a variety of contextual factors, such as the political environment, issue salience, and individual psychological characteristics.

Political polarization has also been a growing concern in Japan, especially in these last few years with the rise of nationalist and conservative tendencies in society. In particular, since the return of Shinzo Abe as prime minister in December 2012, the Abe administration's rightward policy shift has led to widespread concern that it will accelerate the ideological rightward tilting in the Japanese public. Indeed, research has shown that LDP candidates running in national elections have shifted markedly to the right on issues such as lifting the ban on collective self-defense, strengthening defense capabilities, and pressuring North Korea (Taniguchi Reference Taniguchi2015). Accordingly, conservative values such as nationalism, patriotism, ethnic purity, and anti-immigrant sentiment were found to play an ever-increasing supportive role in voting for the LDP (Yoneda Reference Yoneda and Tanabe2019). Thus, the heightened ideological conflict within the political sphere may have led to an increase in public perceptions of political polarization. In addition, the influence of social media and online communication has been shown to facilitate the formation of ideologically homogeneous groups and echo chambers, further exacerbating political polarization among the Japanese public (Lyu Reference Lyu2020; Takikawa and Nagayoshi Reference Takikawa and Nagayoshi2017).

On the other hand, the extent and impact of polarization in Japan remains a topic of debate, with some scholars suggesting that the emergence of political cleavages in Japan is less severe than in other advanced democracies. Numerous studies have shown that the distribution of ideological orientations is not significantly skewed to the right and that the right-wing policies of the Abe administration were not shared by the majority of Japanese voters (Jou, Endo, and Takenaka Reference Jou, Endo and Takenaka2017; Miwa Reference Miwa2018; Taniguchi Reference Taniguchi2015). In other words, it is possible that the degree of perceived political polarization in Japan has been overestimated.

Most of the existing research has therefore focused on the current situation and various factors contributing to the rise or staleness of political polarization in Japan, but studies on the relationship between political polarization and political participation in the Japanese context are limited compared to other countries such as the United States and European countries. Therefore, more research is needed to better understand the dynamics of political polarization in Japan and its effects on political behavior.

Policy divergence as an indicator of perceived political polarization

Measuring perceptions of political polarization is a complex issue that has been approached using different types of data, such as surveys, voting records, media coverage, and online discussions. In this study, I use perceived policy divergence between the major political parties as an indicator of public perceptions of political polarization for the following reasons.

First, it is a relatively efficient way to measure perceived political polarization at the individual level through survey questions that ask respondents to rate the policy differences between political parties. Second, perceived policy divergence may be a useful proxy for measuring the level of perceived political polarization and its relationship to political participation, as individuals’ perceptions of policy differences between political parties are strongly associated with their attitudes toward politics and their willingness to engage in politics. Existing studies have shown that awareness of policy differences motivates individuals to vote for candidates who align with their policy preferences, and shapes broader forms of political expression and engagement, such as attending protests or contacting politicians (Abramowitz and Saunders Reference Abramowitz and Saunders2008). Limited research on this issue in Japan has found a similar tendency. For instance, in an experiment in which participants were asked to read the policies of the LDP and the Democratic Party of Japan (DPJ), Horiuchi, Imai, and Taniguchi (Reference Horiuchi, Imai and Taniguchi2005) found that reading the policies of one party alone did not promote voting, but reading the policies of both parties did. Moreover, the more people perceive significant policy differences between two parties, the more motivated they are to vote (Kobayashi Reference Kobayashi1988).

Third, although there is a possibility that individuals may overestimate the degree of policy differences between parties, especially on issues that are highly salient or emotionally charged, in the Japanese context individuals tend to have more moderate views and less extreme perceptions of policy differences between parties, because the limited success of opposition parties in challenging LDP dominance has led to fewer opportunities to present alternative policy proposals. In light of this, I believe that measuring perceived policy divergence is an effective way to quantify the level of perceived polarization among the Japanese public.

In order to capture a more complete picture of how individuals conceive of the policy divergence between political parties in Japan, I present a multidimensional framework for the analysis of perceived policy differences that incorporates social, economic, and international policy dimensions (Sinno et al. Reference Sinno, Oviedo, Atwell, Alikhani and Li2022). I then annotate each dimension along liberal versus conservative lines.

The social dimension of conservatism is characterized by a preference for tradition, a fear of uncertainty and change, a desire for order and structure, and a prioritization of personal and national security. In contrast, social liberalism emphasizes a belief in individual freedom and equality, a willingness to challenge traditional beliefs and social structures, a commitment to social justice and the promotion of diversity, and an emphasis on progressive change and social reform. On the economic dimension, conservatism is associated with a focus on social rewards, power, and prestige. This can manifest itself in support for deregulation, lower taxes, privatization, and a balanced budget, while opposing deficit spending. Liberalism, on the other hand, values the role of markets in promoting economic growth and prosperity, and emphasizes the importance of property rights and individual autonomy in achieving these goals, which leads to advocating for higher taxes on the wealthy, redistribution of wealth, and other measures to address economic inequality. Finally, conservatism and liberalism have different priorities and values in the international political realm. While conservatism emphasizes national sovereignty, a strong military, and a realist view of international relations, liberalism prioritizes international cooperation and the promotion of idealistic goals such as democracy and human rights. With these theoretical considerations in mind, I present measures of polarization in the next section.

To reiterate, the purpose of this study is to examine the effects of the three dimensions of perceived policy divergence, which represent the degree of political polarization, on various forms of political participation in Japan. The analysis is exploratory, so I have not developed a specific hypothesis. However, if the relationship between perceived political polarization and political participation in Japan is similar to that found in Western societies, we can expect that higher levels of perceived polarization would encourage Japanese people to engage in political activities, including voting and other forms of political engagement.

Data

The objective of this empirical analysis is to examine the impact of the perceptions of political polarization on the political behavior of the Japanese public. To this end, the analysis draws on data from the JES, a repeated cross-sectional survey that covers a wide range of topics, including voting behavior, attitudes toward political issues and candidates, party identification, and political participation. I chose this survey primarily because no other individual-level national election survey in Japan measures respondents’ perceptions of parties’ positions on a common set of policy issues. The JES III (Wave 8), the JES IV (Waves 2–3), the JES V (Wave 1 and Waves 10–11), and the JES VI (Waves 1–3) were used to test the hypotheses in this research because identical questions measuring political attitudes and behaviors were asked in these waves. The surveys were conducted either before or after the 2005–2017 House of Representatives elections and the 2019 House of Councilors election. Utilizing a multistage stratified random sampling strategy, the JES surveyed a random sample of members aged 20 and above (JES III–V), and 18 and above (JES VI), reflecting the adjustment made in response to the lowering of the minimum voting age in national elections from 20 to 18 since 2016. After dropping observations with missing data on the variables used in this analysis, I obtained a pooled sample of 17,874 individuals.

Variables

Dependent variables. The dependent variables measure political participation through a series of questions. Besides voting in an election, respondents were asked whether they had been involved in the following 13 types of political events in the past five years: (1) active in a civic or neighborhood association, (2) contacted influential people in their hometown when needed, (3) contacted politicians or bureaucrats when needed, (4) petitioned or lobbied Congress or government agencies, (5) attended a political event related to their own interests, (6) assisted in an election campaign (e.g. supporting a candidate), (7) participated in a civic or community movement, (8) signed a petition, (9) attended a fundraising event, (10) attended a demonstration, (11) expressed an opinion on the Internet, (12) expressed an opinion by contacting, writing to, or appearing in the media, (13) purchased or refused to purchase a particular product for environmental, political, or ethical reasons.

The first dependent variable measures whether or not the respondents have voted in the last five years. I then conducted an exploratory factor analysis of the thirteen items in order to identify the major categories of these political behaviors. Two major factors emerged from the analysis, each dealing with a category of political participation: political participation within the personal network size, such as contacting politicians directly (hereafter referred to as “private participation”); and political participation resorting to publicly accessible networks, such as joining with others to carry out a demonstration or sit-in (hereafter referred to as “public participation”). To refine the factor structure, items 1 (active in a civic or neighborhood association), 5 (attended a political event related to their own interests), and 12 (expressed an opinion by contacting, writing to, or appearing in the media) were deemed less contributive due to their factor loadings falling below the threshold of 0.3. The detailed results of the factor analysis, including factor loadings and the refined factor structure, are presented in Table 1.

Table 1. Factor analysis of political participation items

Note: Factor loadings of .30 or larger are considered significant after the varimax rotation.

While private and public forms of political participation are not mutually exclusive and can overlap, there are important distinctions between them. Private participation refers to actions taken by individuals outside of formal political institutions to advance their interests, often with the goal of securing material benefits. It is thus self-interested, rather than contributing to the public good or the larger political community. Public participation, on the other hand, is often motivated by a desire to promote the common good or advance a particular ideological agenda, and thus this type of political behavior is more likely to be associated with one's political values and beliefs, such as support for social justice and equality (Chen Reference Chen2013).

The visibility and patterns of private and public political participation also differ. Public participation takes place within established political institutions and processes, is highly disciplined by rules and regulations, and provides transparent and structured opportunities for individuals to engage in the political process. In contrast, private participation tends to take the form of contacting elected officials, lobbying policymakers, or making campaign contributions through personal networks or in small groups, and it is therefore potentially less legitimate than public participation.

Finally, the impact of private political participation depends on an individual's social and economic status and access to resources, and on the responsiveness of political institutions. For instance, individuals with abundant social capital may be more effective in lobbying politicians or shaping public opinion through the media, while those with limited resources may struggle to amplify their voices. In addition, the impact of private participation may be less predictable and harder to gauge than that of public participation. This is because while public demonstrations and protests may result in media coverage and attention from policymakers, the impact of private lobbying or communication with policymakers is less observable or measurable.

Overall, this classification of political participation allows for a more nuanced understanding of the different forms of political behavior in response to perceived political polarization. By distinguishing between private and public forms of political participation, it is possible to examine how individuals’ perceptions of policy divergence on different issue areas relate to the political actions they choose to take and, in turn, how these actions may affect the overall level and nature of political polarization.

The data show that voting in the elections was the most common political act performed by the Japanese respondents (75.8%). In contrast, a much smaller percentage participated in the public type of political activity (25.7%), and participation in the private type of political movement was the lowest of all (13.7%). Three dummy variables were created to indicate whether a respondent had voted or participated in “personal” and “public” types of political activities.

Independent variables. To explain Japanese people's patterns of political participation in response to their perceived policy divergence among the major parties, it is crucial to examine how they access differences in the parties’ positions on policy issues. The JES provides two opposing views (View A and View B) regarding issues such as fiscal reconstruction, consumption tax increase, public services development, nuclear power plant restart, etc., and asks respondents to indicate their preference for one of the two views by choosing from the following answers: “close to A,” “somewhat close to A,” “somewhat close to B,” and “close to B,” and then rate the positions of the major parties on each of the issues. For the analysis, I chose the issues of constitutional revision, public welfare and burden, and the exercise of the right of collective self-defense, because they are the only three issues that were asked in all the waves that I use.Footnote ¹ More importantly, the choice of these issues is consistent with the three dimensions of perceived political polarization described earlier: domestic social dynamics, economic conditions, and foreign affairs.

To analyze how ordinary people perceive policy divergence between major political parties, I selected the parties that have contested all national elections since 2005Footnote ² and divided them into left-wing parties, including the DPJ/center-leftFootnote ³ and the JCP, and right-wing parties, which includes only the LDP. The end of the Cold War, coupled with drastic changes in the party system, has led to inconsistency and confusion in citizens’ ideological orientations (Kabashima and Takenaka Reference Kabashima and Takenaka2012), and reduced the impact of ideology on voting choices compared to factors such as party support and evaluations of political performance (Miyake Reference Miyake1990; Nakamura Reference Nakamura2012; Takenaka et al. Reference Takenaka, Endo and Jou2015). Recent research, however, has called for a “return of ideology” to the study of political behavior due to the rightward shift in Japanese politics. In support of this line of argument, scholars have confirmed that the hosyu–kakushin (conservative–progressive) axis (Curtis Reference Curtis1988) remains the fundamental criterion by which the Japanese voters judge parties’ ideological positions (Hirano Reference Hirano2015; Jou and Endo Reference Jou and Endo2016). In a latest study, Asano (Reference Asano2022) also found a statistically significant correlation between greater ideological extremity and the propensity to vote based on survey data collected in the 2010s, which demonstrates the expanding effect of ideology on political participation. These findings justify the classification of parties according to traditional ideological lines.

As shown in Table 2, in terms of the issues of constitutional revision and the exercise of the constitutional right of collective self-defense, more than 80 percent of the respondents were able to accurately assess the parties’ issue positions by correctly locating the LDP as “close to A” or “somewhat close to A” and the left-wing parties as “close to B” or “somewhat close to B.” It has long been argued that citizens lack the incentive and ability to “calculate relative positioning of parties and candidates” (Carmines and Stimson Reference Carmines and Stimson1980, 82). In fact, most studies that employ empirical methods to examine public perceptions of policy divergence develop arguments positing that voters are uncertain about the differences between political parties (Enelow and Hinich Reference Enelow and Hinich1981; Bartels Reference Bartels1986; Goodman and Murray Reference Goodman and Murray2007). However, the results presented in Table 2 provide evidence that the general public is able to identify the policy positions of the major parties with a high degree of accuracy, especially on the issues that are “symbolic,” “dealing with policy ends than means,” and “long on the political agenda” (Carmines and Stimson Reference Carmines and Stimson1980, 80). Regarding the issue of public welfare and burden, however, while over 80 percent of the respondents placed the LDP in the right position, perceptions of the left parties’ position on this issue were less clear, with approximately half of the respondents placing the parties near View A, and the other half near View B. This result is understandable given the LDP's long history in power, whereby there tends to be greater familiarity with the party's policies because of the issue ownership association that links the LDP to the handling of economic affairs (Bélanger and Meguid Reference Bélanger and Meguid2008). In addition, the median voter theorem posits that policymakers tend to seek to adopt the positions most preferred by the median voters in order to maximize their share of the votes on policy issues that have long been of most interest and importance. Thus, with respect to the issue of public welfare and burden, which has received considerable attention and are of major concern to the general public, the various parties show such a high degree of similarity that people can hardly tell the difference.

Table 2. Perceived policy positions of left- and right-wing parties (2005–2019, in percent)

Next, I calculated the respondents’ perceptions of the differences between left- and right-wing parties on three policies: constitutional revision, public welfare and burden, and the exercise of the right of collective self-defense, which are used as independent variables. The specific calculation method is to take the average score of the two types of parties on each policy and subtract them from each other. The resulting value is then organized into a variable ranging from 0 to 3, with intervals of 1. A score of 0 represents no perceived difference between the two wings of the party, while a score of 3 represents a significant difference (see Figures 1–3).

Figure 1. Public perceptions of policy divergence over constitutional revision between left-and-right wing parties (2005-2019)

Figure 2. Public perceptions of policy divergence over the exercise of collective self-defense between left-and-right wing parties (2005-2019)

Figure 3. Public perceptions of policy divergence over public welfare and public burden between left-and-right wing parties (2005-2019)

As seen in Figures 1–3, there is a notable degree of fluctuation in respondents’ perceptions of differences between the policies of left- and right-wing parties. For example, regarding the issue of constitutional revision, the percentage of respondents who perceived significant differences between the two wings was 23.3 percent in 2005, which increased to 36.6 percent in 2017, before declining slightly to 33.8 percent in 2019. While the percentage of those who perceived medium to small levels of differences has remained relatively stable or decreased over time, the percentage of respondents who viewed no difference between the two wings has increased from 15.1 percent in 2005 to 23.3 percent in 2019.

Moreover, the data reveal that citizens perceived a higher level of policy divergence on constitutional revision and the conduct of collective self-defense compared to the issue of public welfare and burden. Specifically, with regard to constitutional revision, the perception of divergence has increased over the survey period. In addition, there has been a decline over time in the percentage of respondents with a perception of large differences between the two wings of the party on the exercise of collective self-defense, suggesting some convergence in the views of left- and right-wing parties on this issue. However, the exercise of collective self-defense continues to be the policy area in which the highest percentage of all response options is accounted for by “large differences.” For public welfare and burden policies, while there has been an apparent increase over time in the percentage of respondents who perceived large differences, those who perceived no to small differences have consistently comprised over 60 percent, indicating a sense of convergence between the parties on this issue.

In conclusion, while the results alone are not sufficient to draw a definitive conclusion as to whether the level of perceived political polarization has increased or not, public perceptions of policy differences between the left- and right-wing parties appear to revolve around national security and domestic policy, namely constitutional revision and the exercise of collective self-defense, which have been at the center of societal debate for decades. This is consistent with previous research that has highlighted the importance of these issues in shaping Japanese politics.

Control variables. I included a set of additional explanatory variables that have been found to be informative about political behavior. First, I included issue salience in the model because it has been found to play an important role in voting behavior (Bélanger and Meguid Reference Bélanger and Meguid2008; Edwards, Mitchell, and Welch Reference Edwards, Mitchell and Welch1995; Fournier et al. Reference Fournier, Blais, Nadeau, Gidengil and Nevitte2003; Kiousis, Strömbäck, and McDevitt Reference Kiousis, Strömbäck and McDevitt2015). Issue salience is determined by asking the respondents to rank the extent of importance of the political issues in question from 1 (not important at all) to 4 (pretty important). Specifically, approximately 70 percent of respondents find issues related to constitutional revision and the exercise of defense force important. Moreover, the issue of public welfare and burden emerges as particularly salient, with over 85 percent of respondents ranking them as important.

Second, in the realm of voting behavior research, the predominant focus has been on exploring the spatial model of voter choice, where individuals tend to align themselves with the party or candidate whose stances on key issues closely mirror their own (see e.g. Adams and Merrill Reference Adams and III2005; Downs Reference Downs1957). Adhering to the conventional proximity spatial model, the variable of issue proximity is formulated as the squared difference between a respondent's self-placement on pertinent issues and their perceived issue positions of the LDP, the DPJ, or the JCP.

The third control variable is partisan attitudes. Decades of political science research suggest that political behavior can be strongly conditioned by voters’ prior partisan tendencies (Bartels Reference Bartels2002; Evans and Pickup Reference Evans and Pickup2010; Kabashima and Imai Reference Kabashima and Imai2002; Miyake Reference Miyake1983). More importantly, people do not perceive the policy positions of political parties completely objectively, but more or less through the bias of “preferences.” Specifically, people tend to perceive a party's policies as closer to their own if they have a favorable opinion of the party (this psychological bias is referred to as projection or persuasion), and to relegate a party's policy position to the opposite side of the spectrum if they do not like the party (Brody and Page Reference Brody and Page1972; Markus and Converse Reference Markus and Converse1979; Page and Jones Reference Page and Jones1979; Taniguchi Reference Taniguchi2005). With this in mind, I added party support to control for the influence of affective attitudes toward specific parties. Party support is measured by the following close-ended question: “Putting this election aside, which party do you usually support?” Respondents who chose the LDP, the DPJ/center-left, or the JCP were coded as 1, and those who chose other parties or no party were coded as 0. Furthermore, to enhance the analytical rigor and minimize potential confounding effects, individuals with emotional warmth points (感情温度, scaled from 0 to 100) surpassing 90 were systematically excluded from the analysis. Emotional warmth points signify respondents’ affective attitudes towards political parties, and this exclusionary criterion was employed to ensure that individuals with exceptionally strong emotional attachments to a party did not unduly sway the findings.

Fourth, I selected four variables that assess respondents’ psychological involvement in politics, which have long been established as important determinants of political participation: political ideology (measured by self-placement on an 11-point scale from liberal (0) to conservative (10)), satisfaction with current politics (measured on a 4-point scale ranging from “pretty dissatisfied” to “pretty satisfied”), political interest (measured on a 4-point scale ranging from “not interested at all” to “very interested”), and political efficacy, which was measured by the following three items: “Since many people vote in elections, it doesn't matter whether I vote or not,” “I have no power to influence the actions of the government,” and “Politics and government are so complicated that sometimes I have no idea what is going on” (each statement was asked on a 5-point scale ranging from “strongly disagree” to “strongly agree”).

Finally, a set of socio-demographic variables are controlled for, including gender (where 1 indicates male), age, education level, employment (an employee, self-employed, housewife, student, and unemployed); and household income level (less than 4 million yen/year, 4–8 million yen/year, and more than 8 million yen/year). To control for shifts in compositional characteristics across the fifteen years of survey, I include a set of dummy variables: year 2005 (reference), year 2009, year 2012, year 2014, year 2017, and year 2019.