If democracy is to function effectively, voters and their representatives must have access to accurate information and engage in rational discourse on policy challenges. However, this democratic premise is undermined by opinion polarization, where extreme positions on either end of the opinion spectrum tend to dominate discourse, especially when it is strongly associated with demographic attributes such as race and religion. The hardening of attitudes (intensification), in combination with increased sorting, can lead to opinion polarization, as illustrated by “echo chamber” and “filter bubble” effects when individuals have a relatively narrow range of biased information sources either by choice or by algorithm. In the age of social media, disinformation, or “fake news,” can further exacerbate opinion polarization. Ultimately, opinion polarization among the elite and among the general public may interact with each other to increase the risk of political gridlock and social conflict (Chapter 1).

The United States provides a useful example in this regard. Elite polarization, as measured by the distance between the ideal points of Democrats and Republicans in the U.S. Congress, appears to have taken place primarily due to three factors: (1) political realignment in the wake of the civil rights movement in the 1960s; (2) neoliberal ascendancy in the Republican Party since the 1970s; and (3) increasing economic inequality and identity-based politics since the 1980s. In turn, voter (popular) polarization has accelerated since the 2000s, especially between groups with different party preferences. The disproportionate influence of political activists and large donors, amplified by the media, has tended to drive opinion polarization in the U.S (Chapter 1).

This study looks at the characteristics and drivers of opinion polarization in Korea and draws policy implications. Unlike in the U.S., it is difficult to find empirical evidence in support of voter (popular) polarization in Korea. However, elite polarization is at a high level, and some of the factors that have exacerbated opinion polarization in the U.S. are present in Korea, including the role of social media. Accordingly, it would be advisable to craft policy responses to curb opinion polarization.

World Values Survey (WVS) and Korean General Social Survey (KGSS) results show that it is difficult to find empirical support for a trend toward voter polarization in Korea, even when responses from progressives and conservatives are compared with each other, although there is a slight increase in differences between progressives and conservatives since 2010 with respect to the role of the government in addressing inequality (Chapter 2). Accrording to KGSS results, self-reported political preferences do not show a polarizing trend over the 2003-2018 period.

Empirical results from surveys and experiments conducted in 2019 are consistent with the KGSS results. On a Likert scale from 0 (the most progressive) to 10 (the most conservative), self-reported political preferences had a mean of 4.59 and a standard deviation of 1.93, with the neutral position of 5 accounting for 45% of the 1,000 respondents. Political preferences deduced from responses to 25 policy questions, ranging from national security to social welfare, also exhibited a single-peaked, weakly normal distribution. Two-thirds of the respondents were between 4 (moderately progressive) and 6 (moderately conservative), and those who took extreme positions were rare. Self-reported political preferences were statistically associated with respondents’ innate preferences. For example, respondents with a higher level of inequality aversion or risk aversion were more likely to be progressive; whereas, those with a stronger preference for competition, more likely to be conservative. Also, self-reported political preferences showed a statistically significant difference from deduced political preferences, depending on respondents’ demographic characteristics and innate preferences. For example, men compared with women, and the old compared with the young, tended to overestimate their own conservativeness (Chapter 3). Overall, ideological preferences among the general public exhibit a weakly normal distribution in Korea, not a dispersed bimodal distribution typical of opinion polarization.

By contrast, ideological preferences among the elite show a significant level of polarization. According to Han et al. (2018), elite polarization in Korea increased from 0.7 during the 17th National Assembly (2004-2008) to 0.9 during the first half of the 20th (2016-May 2018), as measured by the distance between the ideal points of the two largest political parties. Since the ideal points of lawmakers range from 1 (the most progressive) to +1 (the most conservative), the distance of 0.9 represents 45% of the maximum value and is comparable in magitude to elite polarization in the U.S. around 2010 (Chapter 1).

Moreover, some of the factors that have contributed to opinion polarization in the U.S. are present in Korea. In particular, people with strong political preferences tend to wield a disproportionately large influece in Korea due to to their high level of participation in opinion formation activities, including political demonstrations and social media (Chapter 2).

In fact, empirical results suggest that internet media (social media and internet-only news sites) can drive opinion polarization by providing biased information and influencing popular views. When the degree of information bias is measured by the frequency with which an internet media channel of a certain ideological inclination uses expressions preferred by the opposite ideological camp, it turns out that internet media are more biased than members of the National Assembly. An empirical analysis of the impact of internet media on consumers’ political preferences, based on media panel data in 2012 and 2016, shows that consumers exposed to social media and internet-only news sites tend to become more progressive and more conservative, respectively. In addition, it shows that those who do not prefer news (presumed to have a lower level of news literacy) tend to be affected more by internet media (Chapter 4).

Policy responses to address opinion polarization should be based on the understanding that opinion polarization stems from socioeconomic factors, amplified by the disproportionate influence of political activists and the media. At a fundamental level, because opinion polarization is exacerbated by group polarization, it is important to craft an inclusive identity and prohibit discrimination based on socioeconomic characteristics such as race and religion. It is also important to address economic inequality, which may provide the background for identity-based politics. In order to reduce the disproportionate influence of political activists, election and political financing rules should be amended to better reflect the concerns and aspirations of the general public and rely less on those with strong ideological preferences. Last but not least, in order to reduce information bias, including disinformation, media literacy should be improved through education.

Preface
Executive Summary

Chapter 1 Mechanisms of Opinion Polarization and Policy Implications
　Section 1 Theoretical Background
　Section 2 Opinion Polarization in the United States
　Section 3 Opinion Polarization in South Korea
　Section 4 Policy Implications
　References

Chapter 2 Trends and Patterns of Public Opinion Polarization in Korea through Survey Data Analysis
　Section 1 Introduction
　Section 2 Limitations of the Median Voter Theorem: Implications from U.S. Studies
　Section 3 Trends and Patterns of Opinion Polarization in Korea
　Section 4 Implications of the Analysis Results

Chapter 3 Determinants of Political Preferences: Economic Incentives and Innate Preferences
　Section 1 Introduction
　Section 2 Theoretical Background
　Section 3 Experimental Design
　Section 4 Experimental Research Findings
　Section 5 Conclusion
　References
　Appendix

Chapter 4 Internet Media and Opinion Polarization
　Section 1 Research Overview
　Section 2 Literature Review
　Section 3 Characteristics of Information Supply by Internet Media
　Section 4 Changes in User Preferences and Political Inclinations
　Section 5 Conclusion
　References
　Appendix

ABSTRACT