[The 21st Presidential Election and Korean Democracy: Crisis, Division, and Realignment] ① The 2025 Presidential Election and the Ideological Landscape of Progressivism and Conservatism

I. Introduction

Voter choices in elections are determined by various factors. Among these, ideology has been particularly noted as a key factor explaining voting choices. In Korean politics, the effect of ideology on voters' voting choices has been consistently confirmed over the past two decades. Kang Won-taek's (2003) <Korean Electoral Politics: Ideology, Region, Generation, and Media>, a seminal work on the ideology and voting behavior of Korean voters since democratization, revealed that even in the 1997 presidential election, which was dominated by regionalism, the ideology variable had a certain influence on voters' choices. Since then, numerous subsequent studies have established ideology as a significant factor in explaining Korean election outcomes (Kim Sung-yeon 2022, 2023; Kim, Choi, and Cho 2008; Lee Nae-young 2009; Kim Jeong 2022; Kang Won-taek 2013; Kang 2008).

Then, how did ideology divide voters' choices in the June 3rd by-election, held after the nation was extremely divided over the impeachment of President Yoon Suk-yeol in the first half of 2025? In this election, the two major parties, the Democratic Party of Korea and the People Power Party, nominated candidates with strong support bases, and candidates from a new party derived from the People Power Party and a progressive party outside parliament also competed. Moreover, Korean society at the time faced a political situation where a far-right group, calling itself 'liberal right,' was emerging, supporting conspiracy theories such as election fraud and directly engaging in violence against institutions like the judiciary and the National Assembly, which are the foundations of democracy. Meanwhile, even before the state of emergency in December 2024, ideological confrontation between progressives and conservatives had been identified as the most serious source of conflict in Korean society (Lee Hye-in 2025), and various other data and studies indirectly or directly support the amplification of ideological conflict in Korean society (Lee Dong-han 2025; Lim Jae-hyung 2024; Ga Sang-jun and Yoo Sung-jin 2023). Therefore, despite the specific issues of impeachment and emergency rule dominating the election, it is possible that the ideological confrontation between progressives and conservatives still had some influence on voters' voting choices.

This study aims to reveal the limitations of the widely used single progressive-conservative ideological spectrum in capturing the current ideological landscape of Korean voters and to explore a multidimensional ideological space to compensate for these limitations. To this end, we will utilize data from the 2025 EAI East Asian Survey conducted immediately after the presidential election to extract two independent latent ideological dimensions from multiple policy attitude items and visually present the ideological differences between the support bases of each candidate. In particular, we will show that the support base for candidate Lee Jun-seok, who evaluated themselves as being close to the center on the traditional one-dimensional progressive-conservative spectrum, is by no means centrist in a multidimensional ideological space. Furthermore, we will analyze the independent effects of ideological factors and other variables on voters' voting choices through regression analysis. Through this, we will examine the factors that differentiated the voting choices between candidates Kim Moon-soo, who shares organizational roots and conservative ideology, and Lee Jun-seok, and the factors that determined voter choices between Lee Jae-myung and Lee Jun-seok, who both advocated for moderate conservatism and shared the common ground of supporting Yoon Suk-yeol's impeachment.

II. Data and Variables

1. Ideology and Cleavage

Ideology refers to a vision of 'what constitutes a good society' (Downs 1957; Hinich and Munger 1994; Heywood 2021). The most common and oldest way of distinguishing ideologies is the left-right spectrum, which emerged after the French Revolution in the late 18th century and became a kind of 'political Esperanto' for structuring complex realities simply and universally (Laponce 1981). Traditionally, the left-right distinction is centered on economic issues, such as the degree of state intervention in the free market or the level of state support for socio-economic underdogs. Downs's (1957) median voter theory also presupposes a single left-right ideological axis, assuming that parties strategically adjust their ideological positions based on the distribution of voter ideologies on this dimension.

However, the reality of contemporary politics is multidimensional, and criticism has been raised that a single left-right spectrum is insufficient to fully explain complex ideological structures. In fact, liberal and conservative parties may both take right-wing stances on economic issues, but show opposing attitudes on socio-cultural issues such as abortion or LGBTQ+ rights (Benoit and Laver 2006). Furthermore, far-right parties may be culturally very conservative but advocate for active state intervention and welfare expansion economically, thus exhibiting left-wing characteristics (Bornschier 2010).

Cleavage theory provides a classic starting point for understanding this multidimensional political competition. Lipset and Rokkan (1967) argued that party systems reflect the fundamental conflict structures of society, positing that the four major cleavages that emerged during the formation of modern states and industrialization—center-periphery, church-state, urban-rural, and capital-labor—formed the basis of party conflict and 'froze' Western party systems for a long time.

However, regarding cleavage theory, it has been argued that social cleavages do not directly translate into party competition; rather, political cleavages are formed when political elites and parties strategically mobilize certain conflicts (Sartori 1969, 1990; Evans 1999). Another perspective suggests that even if voter attitudes are multidimensional, actual party competition converges on a single left-right framework centered on economic policy (Sartori 1976; Mair 2007). Knutsen (1995) also argued that all political cleavages are fundamentally linked to the left-right framework and that cleavages are based on psychological characteristics rather than deep-rooted social interests.

In addition, the phenomenon of dealignment, where the link between parties and voters has gradually weakened, has been reported (Pedersen 1979; Dalton et al. 1984). Dealignment refers to the phenomenon where voter party identification weakens, leading to increased volatility in voting choices. Knutsen and Scarborough (1995) pointed out that individual value orientations are becoming more important in voting decisions than social cleavages and group identities, and Dalton (1984, 2000) explained this phenomenon as an increase in 'cognitive mobilization,' where individuals' judgment and decision-making abilities are enhanced. In other words, the weakening of homogeneity within social groups, a key premise of cleavage theory, due to the spread of social liberalism and the emphasis on individual values, has led to the personalization of voter choices (Kriesi 2010).

Meanwhile, some perspectives interpret these changes not as a weakening of existing cleavages but as a realignment due to the emergence of new value-based cleavages. Inglehart (1977, 1990) argued that the rise of post-materialist values is shifting the structure of political cleavages from traditional economic cleavages to value and cultural conflicts. Recent studies have focused on how European party systems have changed due to external shocks such as the Euro crisis and the refugee crisis. In particular, Hooghe and Marks (2018) argue that issues related to globalization, combined with issues of national sovereignty and cultural identity, have created new political cleavages. That is, as voters' political attitudes diverge based on the benefits and losses from globalization, a new dividing line of transnational cleavage has emerged. At the two poles of this cleavage are far-right parties emphasizing national sovereignty and ethnic identity, and political forces such as Green parties advocating for openness and post-materialist values. This suggests a shift in the focus of cleavage theory from a traditional socio-structural approach to a cognitive approach emphasizing values and attitudes.

These discussions demonstrate that it is difficult to capture ideology in contemporary politics solely through a one-dimensional left-right spectrum. In the past, certain social groups were considered to have homogeneous voting tendencies based on common socio-economic positions and experiences; however, today, as individual values and experiences become more diverse, political choices are also becoming personalized. Therefore, to properly understand Korean political ideology and cleavages, it is necessary to attempt a multidimensional, value-centered approach that can encompass voters' diverse value orientations and social experiences.

2. Ideological Cleavages and Measurement Methods in Korea

Since the 15th presidential election in 1997, the influence of ideology in Korean politics began to be highlighted (Kang Won-taek 1998). Since then, in various elections in the 2000s, the fact that progressive-conservative ideology in voting choices has had a certain influence on voter choices has been repeatedly confirmed (Jeong Jin-min 2003; Lee Nae-young 2009; Kang Won-taek 2003; Kim, Choi, and Cho 2008; Lee Kab-yoon and Lee Hyun-woo 2008). Now, along with regionalism, progressive-conservative ideology has become a major variable that cannot be ignored in election analysis.

However, there are still differences of opinion in academia regarding the influence of ideology in elections. In particular, questions arise as to whether ideology still functions as a meaningful variable when certain short-term issues overwhelmingly dominate an election. For example, Kim Sung-yeon (2022) reports that the effect of ideology did not appear in the 19th by-election held after the impeachment of President Park Geun-hye, and furthermore, that ideology significantly affected voting for Lee Jae-myung in the 20th presidential election but did not affect voting for Yoon Suk-yeol. This suggests that there is not yet sufficient firm evidence that ideology consistently affects voting choices at all times.

Another issue is the problem of measuring ideological orientation. In domestic election research, a method of asking voters to rate their own ideology on an 11-point scale (0-10) has been widely used. However, this subjective ideology scale has been consistently criticized for its lack of absolute meaning and limitations in reliability due to differing standards among respondents (Lee Nae-young 2009; Lee Kab-yoon and Lee Hyun-woo 2008). Moreover, this measurement method only shows the intensity and direction on a one-dimensional scale, and it is also pointed out that it does not capture the content or consistency of policy preferences (Ryu Jae-seong 2013). Therefore, the argument has been made that alternative methods, such as using specific policy preference items, should be considered for measuring ideology. As a result, election studies that verify the effects of policy preferences by using comprehensive scores on policies or issues in conjunction with subjective conservative-progressive ideological positions are increasing.

Generally, there is no complete consensus on what specific conflict factors are reflected in the ideology of progressives and conservatives. However, most studies to date have considered attitudes related to North Korea relations and security issues as the most important factors distinguishing conservatives and progressives in Korea. For example, Kang Won-taek (2005), analyzing survey data from political elites such as National Assembly members and ordinary citizens, reported that the issues that most sharply reveal differences in views between groups are related to attitudes toward anti-communist ideology. The study by Yoon Sung-yi and Lee Min-gyu (2014) also identifies attitudes toward North Korea policy as a factor closely related to subjective ideology.

Kang Won-taek (2011), using Lipset and Rokkan's (1967) cleavage theory, analyzed the patterns and origins of Korean political cleavages, emphasizing the conflict surrounding anti-communist ideology. He argued that the conflict between progressives and conservatives, centered on anti-communist ideology, developed into party competition during the nation-building process following liberation and the Korean War, and has become politically more prominent since the 2000s. Studies that have delved into the meaning of subjective ideological evaluations have mostly viewed North Korea and security issues as the core components.

However, in recent years, an increasing number of studies have supported the view that attitudes toward economic policies, taxes, and welfare issues are also important components of ideology. Yoon Sung-yi and Lee Min-gyu (2014) found that for the youth generation, attitudes toward economic distribution and growth are the most important factors constituting subjective ideology. Jou and Koo (2019) observed a stronger correlation than in the past between attitudes toward economic issues and subjective ideological evaluations among citizens of democratic countries in Asia, including Korea. Meanwhile, although Kang Won-taek (2011) assessed the conflict surrounding post-materialist values as still minimal, recent studies report ideological differentiation among voters not only in the political-economic sphere but also in social and post-materialist-related areas, as well as in gender-related issues (Koo Bon-sang 2024; Park Young-deuk and Kim Han-na 2022).

Studies focusing on policy attitudes have generally attempted to move beyond the premise that ideological cleavages are one-dimensional. They also focus on the impact of attitudes toward specific issues or policies on voting choices in elections (Kim Sung-yeon, Kim Jun-seok, and Gil Jeong-ah 2013; Lee Kab-yoon and Lee Hyun-woo 2008; Koo Bon-sang 2024). However, existing studies, while using multiple policy attitude items as alternative ideological indicators, have mostly summarized different policy areas into a single ideological spectrum or grouped items with similar themes within each area to provide summaries for each field. In particular, they used simple averaging of response values using classical test theory. While this has the advantage of being easy to interpret and intuitive, it is difficult to generalize because item characteristics such as difficulty and discrimination can vary depending on the characteristics of the respondent group, as they depend on the respondent's ability (or attitude). As a complementary study, Choi Hyo-no (2022) applied item response theory (IRT) to measure policy attitudes, but this also had the limitation of synthesizing multiple policy attitude items into a single dimension.

Using Item Response Theory (IRT) has the advantage that when measuring respondents' abilities, characteristics such as item difficulty and discrimination are less sensitive to specific respondent groups or item configurations and are maintained as relatively fixed parameters. Consequently, the estimated latent trait scores are expressed as generalizable scales, independent of specific item and sample characteristics, rather than relative positions within a particular sample. Furthermore, it allows for more precise measurement by comprehensively considering the response patterns to individual items (Hambleton, Swaminathan, and Rogers 1991).[1]

However, using unidimensional IRT to reduce heterogeneous policy items into a single dimension or to set separate unidimensional scales for each policy area has the following limitations. First, forcibly integrating items from different policy areas such as economy, security, and socio-culture into a single dimension dilutes the unique ideological content of each area, reducing the accuracy of interpretation (Treier and Hillygus 2009). Second, even within a specific policy area, if items inherently reflect multiple latent dimensions, forcing them into a single dimension violates the unidimensionality assumption of the model, leading to a decrease in model fit and reliability (Embretson and Reise 2000). Ultimately, to accurately capture the multidimensional ideological structure of reality, it is more appropriate to use multidimensional IRT (MIRT), which will be discussed later.

III. Data and Research Methods

We will examine how candidate selection based on ideology was aligned in the 21st presidential election and what patterns of partisan alignment emerged based on policy attitudes, distinct from subjective self-ideology. The data used for the analysis are from the 2025 EAI East Asian Survey. This survey was conducted via web survey for two days (June 4-5, 2025) immediately after the 21st presidential election, and a total of 1509 responses were collected through stratified sampling by region, gender, and age (response rate 22.5%). Among these, 1451 respondents reported having voted. The vote distribution by candidate was as follows: Lee Jae-myung 48.7%, Kim Moon-soo 33.9%, Lee Jun-seok 6.2%, and Kwon Young-guk 1.5%. Additionally, 9.1% responded that they did not wish to disclose their voting target, and 0.7% included other candidates and don't know/no response.

The main independent variable, ideology, is measured in two ways. First, we use respondents' subjective self-assessment of their ideological position on a one-dimensional scale from progressive (0) to conservative (10), the most common method used in previous research. By comparing the distribution of subjective ideologies across the support bases of each candidate, we will assess whether this election was indeed one where voters were polarized into progressives and conservatives and aligned in a partisan manner. Of particular note is that in this presidential election, candidate Kim Moon-soo advocated for conservatism, while Lee Jae-myung and Lee Jun-seok both advocated for moderate conservatism. Did the nominal ideologies of the parties and candidates reflect in the ideological distribution of each candidate's support base? Or was the social perception of severe ideological conflict in recent years more reflected in the actual ideological distribution of voters?

Next, we examine the ideological differences in policy attitudes, separate from subjective self-ideology. We will explore multidimensional ideological cleavages using policy attitude items related to North Korea security and economic policies, which have formed the main axis of progressive-conservative ideology in Korean politics since democratization, as well as items related to minority rights, gender, and climate crisis, which have emerged as new social conflict issues in recent years. To overcome the limitations of classical test theory or unidimensional IRT, this study applies MIRT, which simultaneously estimates multiple latent dimensions and effectively reflects the structural relationships between items. In the resulting two-dimensional ideological space, we will visually compare the distribution of support bases for each candidate to reveal how votes were divided by ideology. This suggests not only that voters held different ideological positions depending on the candidate they chose, but also that post-materialist values, such as social justice, gender equality, and the environment, which previous studies have relatively overlooked or underestimated, have now emerged as undeniable dimensions of political cleavage.

Finally, we will statistically verify whether ideology still influences candidate selection while controlling for other variables, using logistic regression analysis. We will examine how the probability of voting for a particular candidate changes according to voters' ideological positions on each of the socio-cultural and economic-security ideological dimensions. In particular, we will statistically analyze the factors that differentiated the voting choices of voters who voted for Lee Jun-seok, whose chances of winning were slim and who was expected to be a 'wasted vote,' and why they voted for Lee Jun-seok rather than Kim Moon-soo or Lee Jae-myung.

IV. Analysis Results

1. A Polarized Presidential Election between Progressives and Conservatives?

<Figures 1(a)-(d)> visualize the distribution of subjective ideologies measured on an 11-point scale, divided by the support base for each candidate. These results clearly show that voters are distinctly divided ideologically according to their candidate choice.

First, the subjective ideology distribution of Lee Jae-myung's supporters is concentrated in the 0-4 range, meaning 55% of responses, typically classified as progressive, were concentrated there, while responses in the 6-10 (conservative tendency) range accounted for only 14.5%. This ideological bias is reproduced in the opposite direction, and even more intensely, among Kim Moon-soo's supporters. The proportion of responses in the 0-4 range among Kim Moon-soo's supporters was only 5.4%, while the 6-10 range accounted for 72.8% of the total, showing an extremely biased distribution toward conservatism. Lee Jun-seok's supporters, from the Reform Party derived from the same party as Kim Moon-soo, show a significantly moderated bias compared to Kim Moon-soo's supporters. Among Lee Jun-seok's supporters, the 0-4 range accounted for 14.6% and the 6-10 range accounted for 44.9%. Kwon Young-guk's supporters, the only candidate advocating progressivism, showed response patterns more biased to the left than Lee Jae-myung's supporters, with 55.6% in the 0-4 range and 5.6% in the 6-10 range.[2]

These differences are clearly evident in the median values of each candidate's support base shown in <Figure 1(e)>. The median values are 4 for Lee Jae-myung, 7 for Kim Moon-soo, 5 for Lee Jun-seok, and 3.5 for Kwon Young-guk, confirming that voters are clearly aligned by ideology in their candidate choices. In particular, the fact that the support bases of the candidates from the two major parties, which have occupied most of the seats in the National Assembly and alternated in the presidency, are so starkly divided ideologically can be seen as evidence supporting the 'ideological polarization phenomenon' frequently raised in academic circles recently.

Notably, the ideological distribution of Lee Jun-seok's supporters is located between those of Lee Jae-myung and Kim Moon-soo. Among Lee Jun-seok's supporters, the proportion who chose 5, the exact center of the 11-point scale, was 40.4%, higher than among supporters of other candidates. These individuals consider themselves centrist or moderate conservatives. Lee Jun-seok's supporters are concentrated in the center, and despite the relatively small sample size, the narrow range of the distribution indicates a strong internal ideological homogeneity.

<Figures 1(a)-(e)> Distribution of Subjective Self-Ideology by Candidate Support Base: A Unidimensional Spectrum from Progressive (0) to Conservative (10)

Meanwhile, as shown in <Figure 1(e)>, in contrast to Lee Jun-seok's supporters, Kwon Young-guk's supporters exhibit noticeable internal heterogeneity in terms of subjective self-ideology. This can be sufficiently understood considering that the sample size of Kwon Young-guk's supporters is extremely small relative to the total sample (18 people). However, on the other hand, it is also possible that the ideological characteristics of Kwon Young-guk's supporters have a complex nature that cannot be fully captured by the unidimensional structure assumed by the subjective self-ideology scale.

2. Lee Jun-seok the Centrist? The Position of Lee Jun-seok's Supporters Between Lee Jae-myung and Kim Moon-soo

Should we then consider Lee Jun-seok's supporters, who define themselves as closest to the center and exhibit high internal ideological homogeneity, as ideologically centrist? Do Lee Jun-seok and his supporters truly represent the center and possess the potential to resolve the harms of ideological polarization?

To examine the characteristics of Lee Jun-seok's supporters in more detail, we closely analyzed the ideological distribution of Lee Jun-seok's supporters, divided by the entire age group and by young people aged 39 and under. Through <Figure 2(a)>, we confirmed that while the median value for the entire age group was 5, the median value for young people aged 39 and under was 6, indicating that the younger demographic has a relatively more conservative tendency. Furthermore, although the proportion of Lee Jun-seok's supporters who identified themselves as extremely conservative (8-10 range) was not very high at 14.6%, this proportion increased to 21.8% when limited to young people aged 39 and under (<Figure 2(b)>). It is also noteworthy that the majority of respondents within Lee Jun-seok's support base who exhibit extreme conservative tendencies are young people aged 39 and under.[3]In other words, it is important to note that the centrism of Lee Jun-seok's supporters varies by age, and particularly within the youth demographic, a significant portion exhibits strong conservative tendencies.

<Figures 2 (a-b)> Distribution of Subjective Self-Ideology within Lee Jun-seok's Support Base: All Ages vs. Under 39

Of course, it would be an overstatement to define Lee Jun-seok's supporters as conservative or even far-right based solely on this analysis. However, election results by age group show that Lee Jun-seok received a particularly high vote share among young people, and it is well-known that men in their 30s and younger are his core support base. Considering all these factors, it is necessary to approach the potential and limitations of Lee Jun-seok and his supporters functioning as a centrist alternative in Korean politics with greater caution and complexity.

3. Ideology of Candidate Support Bases Based on Policy Attitude Items

Next, let's compare the policy attitudes of each candidate's support base using six policy attitude items covering socio-cultural and economic-security domains. Unlike timely and short-term policy issues such as the state of emergency, pension reform, and soaring real estate prices, these items reflect long-term and sustained value orientations, making them interpretable as progressive-conservative ideology. Socio-cultural policy attitudes were measured using items related to the guarantee of rights for social minorities (homosexuals, foreign workers, disabled people) (inclusion and social justice), response to climate crisis (environment), and resolution of gender discrimination (gender equality). The economic-security domain consisted of items related to strengthening military preparedness against North Korea (North Korea security), prioritizing market and corporate competitiveness (marketism), and accepting inequality (meritocracy). For the socio-cultural items, the proportion of respondents who answered 'disagree' was calculated to determine the percentage of voters with conservative attitudes for each item.

The results in <Table 1> reveal interesting differences in the policy orientations among the candidate support bases. First, Kim Moon-soo and Lee Jun-seok's supporters shared strong conservative tendencies on security and economic meritocracy issues. For strengthening military response against North Korea and accepting inequality based on merit, Kim Moon-soo's supporters showed 81.5% and 60.7% respectively, while Lee Jun-seok's supporters showed 70.8% and 60.7% respectively. However, regarding the prioritization of market and corporate competitiveness over the improvement of labor rights and working conditions, Kim Moon-soo's supporters showed a very high rate of 60.3%, while Lee Jun-seok's supporters showed 40.4%, not significantly different from Lee Jae-myung's supporters (38.2%), indicating a relatively weaker traditional economic conservative tendency on this item.

<Table 1> Proportion of Conservative Responses by 6 Policy Attitude Items: Comparison of Candidate Support Bases

However, the two candidates' support bases showed subtle but noticeable differences on socio-cultural issues. The proportion of opposition to policies protecting minority rights exceeded half among Kim Moon-soo's supporters at 52.4%, while it was slightly lower at 43.8% among Lee Jun-seok's supporters. On the other hand, the proportion opposing government intervention in policies to resolve gender discrimination was more than twice as high among Lee Jun-seok's supporters (40.4%) as among Kim Moon-soo's supporters (18.8%). Regarding response to climate crisis, the proportion of conservatives was 1.8 times higher among Lee Jun-seok's supporters (25.8%) compared to Kim Moon-soo's supporters (14.3%).

In contrast, Lee Jae-myung and Kwon Young-guk's supporters shared clear progressive values on socio-cultural issues. For Lee Jae-myung's supporters, the proportion showing conservative stances on minority rights (21.7%), gender equality (8.4%), and climate crisis response (4.4%) was very low, indicating a generally strong progressive tendency. For Kwon Young-guk's supporters, 44.4% showed a conservative stance on the item accepting meritocracy based on individual performance, slightly higher than Lee Jae-myung's supporters (35.5%), but overall they were favorable to progressive values. Notably, regarding the North Korea security item, 39.9% of Lee Jae-myung's supporters were conservative, while only 16.7% of Kwon Young-guk's supporters were, showing a clear difference in stance between the two candidates' support bases on the North Korea security issue, which has traditionally been considered a strong ideological component.

Overall, while Kim Moon-soo and Lee Jun-seok's supporters generally shared conservative values in the areas of security and economic meritocracy, they showed distinct differences in the degree and direction of conservatism on socio-cultural issues. Lee Jae-myung and Kwon Young-guk's supporters, while generally supporting progressive values, each exhibited distinct individual characteristics on issues such as marketism, North Korea security, and climate crisis.

4. A Two-Dimensional Ideological Structure of Socio-Cultural and Economic-Security

To understand the structure of ideological orientations, we conducted an exploratory multidimensional item response theory (MIRT) analysis of the six policy attitude items without pre-specifying the number of dimensions. MIRT estimates how sensitively each item responds to multiple latent factors (θ), providing richer information about the ideological structure than unidimensional IRT (Hassan and Miller 2022).

The analysis revealed that the items differentiated along two main dimensions: 'Social, Environmental, and Gender' and 'Security, Market, and Meritocracy' (<Table 2>). Applying the Graded Response Model (GRM) for two dimensions, the correlation coefficient between the two factors was low at 0.112, confirming that each factor operates independently. The communality of each item was around 0.50 on average, indicating that the two latent factors explain more than half of the item variance. The SS loadings for both factors were also at the level of 1.5-1.6, explaining the total variance equally. The reliability coefficients were somewhat low at F1 (0.358) and F2 (0.363), but within the acceptable range for exploratory analysis. Meanwhile, the correlation between the policy ideology factors and subjective self-ideology was r = 0.25 (p < 0.01) for socio-cultural ideology (F1) and r = 0.36 (p < 0.01) for economic-security ideology (F2), which is consistent with previous literature suggesting that economic and security dimensions are more closely related to subjective self-ideology perception.

<Table 2> Factor Loadings and Communalities for 6 Policy Attitude Items in a Two-Dimensional GRM

Using MIRT, we estimated the position (θ value) of each voter on the socio-cultural and economic-security dimensions. The distribution of the estimated θ values theoretically approximates a normal distribution with a mean of 0 and a standard deviation of 1, but it can vary slightly depending on the characteristics of the actual sample and the item configuration. <Figures 3(a)-(b)> visualize the candidates' support bases in the two-dimensional ideological space, color-coded based on this analysis. The socio-cultural dimension (F1, Y-axis) represents more conservative tendencies as it moves upward, and the economic-security dimension (F2, X-axis) represents stronger conservative tendencies as it moves to the right. The areas where the respondents of each group are concentrated are marked with convex hulls using only the high-density points corresponding to the top 25% of kernel density estimation. This allows for effective visualization of the core distribution of the data while reducing the influence of a few extreme data points (outliers).

First, looking at the entire age range, in <Figure 3(a)>, the convex hulls of the voter bases for candidates Lee Jae-myung and Kim Moon-soo are clearly distinguished, with the former widely distributed to the left and the latter to the right in terms of economic security. However, in the intermediate region, the two groups overlap considerably. If we mark the average points (centroids) of the x and y values for each voter base with bold X's, the centroids of the two voter bases are positioned almost symmetrically in the first and third quadrants with respect to the origin (0, 0). This indicates that while the voter bases for the two candidates show clearer ideological differences in terms of economic security than in socio-cultural dimensions, a certain distinction also exists in socio-cultural dimensions. Meanwhile, the voter base for Kwon Young-guk, shown in <Figure 3(b)>, had centroids with negative values in both dimensions, indicating a consistently progressive tendency. However, due to the very small sample size, the variance is large, making it difficult to discern a clear pattern.

The voter base for Lee Jun-seok is noteworthy. In <Figure 3(b)>, the voter base for Lee Jun-seok is narrowly distributed in the economic security dimension and concentrated in the center. Therefore, it is difficult to classify them as clearly leaning towards either conservative or progressive on the traditional economic security ideological axis. Indeed, during the election campaign, Lee Jun-seok consistently emphasized a centrist position, drawing parallels to French center-right politician Macron, former President Kim Dae-jung who pursued neoliberal reforms, and former President Roh Moo-hyun, who was sometimes criticized for "turning right with the left blinker on" while advocating for a grand coalition encompassing both left and right. The data also shows that Lee Jun-seok's voter base falls in the centrist or center-right category on the economic security axis.

<Figure 3(a)-(b)> 2D Policy Attitude Ideology Distribution: Comparison by Candidate Voter Base (All Ages)

In contrast, their socio-cultural ideological distribution is much wider vertically, and notably skewed towards the conservative end. Lee Jun-seok can be seen as having successfully mobilized a new political cleavage by actively attracting conservative voters on socio-cultural dimensions—issues surrounding values of inclusivity, openness, and post-materialism—in a situation where differentiation based on the traditional economic security-centric ideological framework is difficult. This can be interpreted as a strategy for Lee Jun-seok to secure a distinct voter base, while the People Power Party and the Democratic Party share broad support bases on the ideological axis and simultaneously occupy a significant portion of the middle ground, converging on policy. In fact, positioning further to the right than the People Power Party carries a high risk of immediate criticism as far-right. In other words, Lee Jun-seok appears to have actively utilized the new political landscape by appealing to centrist voters on the economic security dimension while consolidating conservative voters on the socio-cultural dimension.

5. Divided Youth in the 2D Ideological Space

One of the key characteristics of the 2025 presidential election results was the particularly prominent proportion of voters who cast their ballots for the third candidate, Lee Jun-seok, among young people aged 20-30. Claims of the "conservatization" and even "far-rightification" of so-called 'Lee Dae-nam' (men in their 20s) are frequently raised. How, then, does the ideological distribution of young voters compare to that of the overall electorate when analyzed by candidate voter base? <Figure 4(a)-(b)> reconstructs the same 2D ideological space as above, focusing solely on voters aged 39 and under.

Comparing the young voter bases for Lee Jae-myung and Kim Moon-soo, the degree of overlap between the two groups is significantly reduced compared to the overall age range. In the case of Lee Jun-seok and Kwon Young-guk's voter bases, the top 25% density regions do not overlap at all. The distance between the centroids of each group is also larger among young people, clearly revealing ideological division. This division is particularly pronounced along the socio-cultural dimension, i.e., the Y-axis, rather than the left-right (economic security) axis. Furthermore, Lee Jun-seok's voter base shows completely separate density regions from Lee Jae-myung's voter base, while showing a broad overlap in the centrist region with Kim Moon-soo's voter base. At the same time, conservatism in the socio-cultural dimension is most prominent in Lee Jun-seok's voter base.

Meanwhile, the Justice Party has faced internal criticism for a weakened identity centered on labor. Just before the presidential election, the party changed its name to the 'Democratic Labor Party' to emphasize 'labor' while actively promoting socio-culturally progressive issues such as the climate crisis, feminism, and anti-discrimination laws. Indeed, Kwon Young-guk's voter base among those aged 39 and under (even considering the extremely small sample size) shows progressive tendencies in both socio-cultural and economic security dimensions, suggesting that the Democratic Labor Party has actively responded to new political cleavages. In summary, these results suggest that the socio-cultural dimension has become important in the ideological differentiation and partisan alignment of young people based on ideology.

<Figure 4(a)-(b)> 2D Policy Attitude Ideology Distribution: Comparison by Candidate Voter Base (Ages 39 and Under Only)

6. Why Vote for Lee Jun-seok, Not Lee Jae-myung or Kim Moon-soo?

Binary logistic regression was used to analyze the factors determining candidate choice among voters. In the case of multinomial logistic regression, the number of voters for the three candidates in the analysis—Lee Jae-myung (705), Kim Moon-soo (496), and Lee Jun-seok (89)—was significantly imbalanced, with the proportion of Lee Jun-seok's voters (7%) being particularly low. This sample imbalance can increase the risk of unstable regression coefficient estimation, distorted p-values, and overfitting for the minority category (Lee Jun-seok) in a multinomial logit model. Therefore, to enhance the stability and clarity of interpretation, this paper adopted separate binary logistic models for Lee Jun-seok versus Kim Moon-soo and Lee Jun-seok versus Lee Jae-myung. This approach aims to analyze the factors influencing choice between candidates more stably and clearly.

First, let us examine only the voter bases of Kim Moon-soo and Lee Jun-seok, who were found to overlap considerably in ideological terms. In the analysis of Lee Jun-seok (=1) vs. Kim Moon-soo (=0) in <Table 3>, the variable that best explained the choice between candidates was the perception of emergency martial law (1=martial law declaration is not justifiable, 0=neutral or justifiable). Voters who assessed martial law as unjustifiable had approximately 4-6 times higher odds of voting for Lee Jun-seok compared to those who did not. Furthermore, the less conservative voters were in their subjective self-placement of ideology and their economic security ideological dimension, the higher their likelihood of choosing Lee Jun-seok. Similarly, the less conservative voters were in their subjective self-ideology or economic security policy attitude ideology, the higher their probability of choosing Lee Jun-seok over Kim Moon-soo increased.

Conversely, however, a statistically highly significant tendency emerged where the more conservative voters were on the socio-cultural dimension, the higher their probability of voting for Lee Jun-seok (Model 3). When other factors were held constant, a shift in socio-cultural ideological position from 0 to 1 (towards the conservative end) increased the odds of voting for Lee Jun-seok by approximately 42%. Additionally, voters aged 39 and under, and males, were more likely to choose Lee Jun-seok.

The analysis of Lee Jun-seok vs. Lee Jae-myung in <Table 4> also shows that the perception of emergency martial law significantly influenced candidate choice. The odds of choosing Lee Jae-myung over Lee Jun-seok were 3-4 times higher for voters who assessed martial law as unjustifiable. Furthermore, stronger conservative tendencies not only in subjective self-ideology but also in both socio-cultural and economic security dimensions significantly increased the probability of choosing Lee Jun-seok over Lee Jae-myung. Among these three ideological dimensions, the socio-cultural dimension showed the strongest explanatory power; a shift from 0 to 1 (more conservative) in this dimension increased the odds of voting for Lee Jun-seok by 2.5-2.7 times. On the other hand, voters aged 40 and above, or females, were more likely to vote for Lee Jae-myung than Lee Jun-seok.

There are several other interesting findings. First, voters who perceived themselves as belonging to the upper or upper-middle class were significantly more likely to choose Lee Jun-seok over Lee Jae-myung. This variable was not significant in the Lee Jun-seok vs. Kim Moon-soo comparison, suggesting that, all else being equal, Lee Jun-seok's voter base can be considered economically closer to the upper class than Lee Jae-myung's voter base.

<Table 3> Binary Logit Regression Results for Lee Jun-seok vs. Kim Moon-soo Vote Choice

<Table 4> Binary Logit Regression Results for Lee Jun-seok vs. Lee Jae-myung Vote Choice

Second, higher external political efficacy was associated with a tendency to choose Lee Jae-myung rather than Lee Jun-seok. This suggests that Lee Jun-seok's supporters may be closer to being 'observers' rather than active 'actors.' Interestingly, the media and some commentators have assessed that the political efficacy of Lee Jun-seok's supporters—particularly the young male voter group—has surged in recent years. For example, analyses suggest that young people gained political confidence following the 2021 Seoul mayoral by-election and the election of Lee Jun-seok as the People Power Party leader, and subsequently actively participated in politics during the 20th presidential election by voluntarily producing content such as memes and shorts (Ko Hye-ji, 2022). Furthermore, during the impeachment proceedings in the first half of 2025, it was assessed that "the political efficacy and confidence gained by young conservatives have increased incomparably compared to eight years ago" (Lee Dong-soo, 2025). However, contrary to these assessments, the analysis results of this study indicate that respondents with high external efficacy were more likely to choose Lee Jae-myung over Lee Jun-seok.

Interesting results also emerged from the regional context. To identify the strongest regional cleavage in Korean elections, the Yeongnam-Honam axis, the residential region was controlled by dummy coding Honam as the reference category and Yeongnam and Others (Seoul Metropolitan Area, Chungcheong, Gangwon, Jeju) as comparison categories. The analysis revealed that, while controlling for all other factors, residents of Honam showed a slight tendency to choose Lee Jun-seok over Kim Moon-soo, and residents of Yeongnam showed a slight tendency to choose Lee Jun-seok over Lee Jae-myung. This suggests that, in Korean elections where regionalist voting behavior remains strong—albeit at a very weak level compared to the overwhelming first-choice candidates in each region—Lee Jun-seok may have been considered an alternative choice for some voters.

This time, we will visually examine the influence of socio-cultural ideological dimensions on the choice of Lee Jun-seok. By fixing the variables included in Model 3 of <Tables 3-4> as follows and simulating, we calculated the predicted probability of choosing Lee Jun-seok. Age was set to 39 and under, gender to male, residential region to areas other than Honam and Yeongnam, and the remaining variables such as economic and security ideology, subjective self-ideology, internal and external political efficacy, and subjective class perception, and whether or not a 4-year university degree was held, were applied at the average values of the entire sample. Additionally, the perception of former President Yoon Suk-yeol's declaration of emergency martial law was divided into 4-5 points (perceived negatively) and 1-3 points (neutral or justifiable measure), and we compared how the probability of choosing Lee Jun-seok differed between the two groups. By controlling all conditions equally in this manner, we can visually represent how the probability of choosing Lee Jun-seok changes when socio-cultural ideology is purely conservative or progressive.

<Figure 5(a)> shows the change in the probability of voting for Lee Jun-seok as the socio-cultural ideological tendency moves from -2.5 (very progressive) to +2.5 (very conservative) in the competition between Lee Jun-seok and Kim Moon-soo. For both groups—those who perceive emergency martial law negatively and those who do not—the probability of choosing Lee Jun-seok increases by approximately 39% and 18%, respectively, when the socio-cultural ideological dimension is very conservative compared to when it is very progressive. In particular, for young males aged 39 and under (outside the Honam-Yeongnam regions) who are very conservative socio-culturally (+2.5) and critically view martial law, the probability of choosing Lee Jun-seok reaches about 60%. Even if they do not view emergency martial law critically, if they have conservative socio-cultural ideology, the probability of voting for Lee Jun-seok increases up to 23%.

<Figure 5(b)> reveals a trend where the probability of supporting Lee Jun-seok increases more sharply as socio-cultural ideology becomes more conservative in the competition between Lee Jun-seok and Lee Jae-myung. This suggests that socio-cultural ideological orientation is critically important when voters decide between Lee Jae-myung and Lee Jun-seok. Voters who are very socio-culturally progressive (-2.5) choose Lee Jae-myung with a probability of over 95%, regardless of their perception of emergency martial law. However, if their ideology in this dimension is very conservative (+2.5), the probability of choosing Lee Jun-seok soars to 84% (when viewing martial law as neutral or justifiable).

This group shows a probability of voting for Lee Jun-seok of approximately 61%, even when they perceive emergency martial law negatively, compared to Lee Jae-myung. However, in contrast to <Figure 5(a)>, the group that holds a neutral or justifiable view of the emergency martial law declaration (Others) shows a higher probability of voting for Lee Jun-seok across all ideological ranges compared to the group that perceives it negatively. For individuals with very conservative socio-cultural ideological tendencies, there is a difference of about 23% in the probability of choosing Lee Jun-seok between those who perceive martial law negatively and those who do not.

<Figure 5(a)-(b)> Simulation of Lee Jun-seok Vote Probability by Change in Socio-Cultural Ideology Score (Comparison by Perception of Emergency Martial Law)

Notably, very socio-culturally progressive voters turn away from Lee Jun-seok and choose Lee Jae-myung, regardless of their perception of emergency martial law. This suggests that socio-cultural ideologies such as post-material values, inclusivity, and openness, which have received relatively little attention in studies of Korean political behavior, have emerged as key factors in voters' electoral decisions. Furthermore, it is even more significant that for some voters, these ideological orientations have such a powerful influence that they override evaluations of specific political events that have severely divided Korean society in recent months.

For voters who are distinctly socio-culturally progressive, it appears that Lee Jun-seok's political message or policy stance fundamentally conflicts with their belief system, leading them to reject him almost automatically, irrespective of their views on sensitive political events like emergency martial law. This pattern suggests that the new ideological cleavage that has emerged in the recent Korean political landscape is not merely about attitudes towards individual policies or specific political events, but is forming a fundamental and robust political dividing line rooted in deeper values and worldviews.

V. Conclusion

This study confirmed that 'ideology' was a major factor influencing vote choice in the 2025 by-election. Voters exhibited a pattern where the voter bases for the established two-party candidates were clearly divided into progressive and conservative extremes on the subjective ideology scale. This supports the notion that Korean voters' choices are made within a framework of clearly divided ideological camp competition. Furthermore, this study demonstrated that, instead of the subjective ideology on a one-dimensional scale commonly used to measure ideology, the ideological structure of voters in the 2025 presidential election manifested along at least two dimensions, utilizing multiple policy attitude items.

The first dimension is the economic security dimension, which includes North Korean security and market-oriented and meritocratic principles that have become core ideological components of Korean politics since democratization. The second dimension is the socio-cultural dimension, which includes attitudes towards minority rights, the environment, and gender equality. The voter bases for Lee Jae-myung and Kim Moon-soo showed relatively clear ideological distances from opposing camps on the economic security dimension. However, at the same time, their voter bases overlapped considerably around the midpoint of this axis. In other words, while the voter bases of the established two parties are ideologically opposed, they are not completely separate extremes; there are also a considerable number of individuals around the center who share policy attitudes and value orientations with voters of the opposing party. Despite concerns about ideological polarization or extreme partisan alignment within and outside academia, the analysis results of this study further support the possibility that the voices of a small extreme minority are being overrepresented and echoing throughout our society.

The group highlighted in the two-dimensional ideological space was the voter base of Lee Jun-seok, the reformist candidate who emerged as a third political force in this election. While generally conservative in the economic security dimension, they were concentrated near the center, overlapping considerably with the voter bases of the two major parties. In terms of the economic security dimension, which is generally perceived as constituting the left-right ideology in Korea, Lee Jun-seok's voter base is the most centrist group among the three parties.

However, on the other ideological axis, the socio-cultural dimension, this group exhibited a significantly strong conservative tendency. Specifically, they showed a pattern of opposing values such as openness, inclusivity, and post-materialism while supporting conservative policies. In other words, a voter group that wore the mask of centrism in traditional conservative ideologies like economy and security, but was firmly conservative on the newly emerging socio-cultural ideological dimension, which has become a major axis of conflict, appeared in the 2025 by-election. Although the analysis results of this study alone cannot fully explain when and through what process this group first appeared in Korean politics, it demonstrates that the ideological landscape of voters is evolving into a more complex ideological structure by adding new ideological dimensions, which are divided according to individual values and experiences, beyond traditional cleavages based on the acceptance or rejection of anti-communist ideology or the opposition between free markets and government intervention.

Notably, the voter group most sensitive to this new ideological cleavage and experiencing distinct differentiation is none other than the youth. While the traditional progressive-conservative cleavage was primarily formed at a macro-discourse and abstract level, the new cleavage is a micro-level division surrounding identity, individual values, and lifestyles. Therefore, young people can experience the conflicts in Korean society as concrete and emotional issues such as discrimination, hatred, and unfairness encountered in daily life. Today's youth are highly likely to continue to differentiate and politicize along the socio-cultural ideological dimension, and their current experiences will shape their political attitudes even when they become middle-aged. Therefore, before this ideological cleavage escalates into more acute conflict and social tension, serious consideration is needed on how to socially educate and establish the fundamental norms of democracy—inclusivity and pluralism. This requires a more fundamental and transformative discussion on the future direction of Korean democracy, beyond the issue of institutional representation for specific ideological groups.

This study has the following limitations, which suggest several avenues for future research. First, as it is a cross-sectional analysis using data from a single point in time—the 2025 by-election—follow-up research is essential to confirm how persistent and stable the new cleavage is. Second, the limited number of policy attitude items used in this study may not fully capture the complex ideological structure of voters. Therefore, future research needs to develop additional items based on theoretical review. Applying a multidimensional item response theory model based on expanded items is expected to enhance the information content of each factor, thereby reducing the uncertainty of the respondent's estimated position (θ value) and increasing the reliability of the limits.

Third, since this study uses cross-sectional survey data collected immediately after the election count, the possibility that the values of key political attitude variables (subjective self-ideology, perception of emergency martial law, policy attitude ideology) are influenced by post-hoc rationalization or motivated reasoning after the vote choice cannot be ruled out. Therefore, it is difficult to conclude that a causal relationship has been established between these variables and vote choice. For more rigorous causal inference, alternative designs such as pre- and post-election panel data or randomized experiments are necessary. Fourth, although this study elucidated the relationship between socio-cultural conservatism among some young people and their vote choice, the experiences and conditions that shape these attitudes need to be investigated in future research to clarify the micro-level mechanisms.

Finally, it is necessary to examine from a comparative politics perspective whether the results of this study are limited to the Korean context or are part of a global trend where similar socio-cultural cleavages are emerging in other countries. For example, reports indicating that Generation Z males in Australia adhere more strongly to traditional gender role beliefs than older males or females of the same age (Clarke, 2025), and analyses showing that the performance of far-right parties in 27 European countries heavily relies on the support of young men, with the gender gap in support for far-right parties among young people widening since 2020 (Milosav et al., 2025), suggest that the strengthening of conservative tendencies among young men may not be an exception limited to specific countries. Future comparative research using data from the World Values Survey, along with comparative case studies of East Asian countries, is needed to determine the universality and specificity of the Korean case.■

VII. References

Gasal, Jun, and Sung-jin Yoo. 2023. “A Study on the Ideologization of Public Conflict: Conflict Perception by Partisanship.” *Journal of Social Science Research* 31(2), 8-36.

Kang, Won-taek. 1998. “Voters' Ideological Tendencies and Voting Behavior.” In *Elections in Korea II: Focusing on the 15th Presidential Election*, edited by Nam-young Lee. Seoul: Purun-gil.

______. 2003. *Korean Electoral Politics: Ideology, Region, Generation, and Media*. Seoul: Purun-gil.

______. 2005. “Ideological Conflict in Korea and the Boundary Between Progressive and Conservative.” *Korean Party Studies Review* 4(2), 193-217.

______. 2011. “Historical Origins of Political Cleavage Structures in Korea: Applying the Lipset-Rokkan Model.” *Korea and International Politics* 27(3), 99-129.

______. 2013. “'Class-Defecting Votes' and Social Class in Korean Elections.” *Korean Party Studies Review* 12(3), 5-28.

Ko, Hye-ji. 2022. “The 2022 Presidential Election Was a 'Meme' Election for Youth Feeling Political Efficacy.” *Seoul Shinmun*. 2022-02-01.

Gu, Bon-sang. 2024. “Modern Gender Gap in the Voting Behavior of Voters in their 20s: Analysis of the 22nd National Assembly Election.” *Journal of Political Information* 27(2), 33-63.

Gil, Jeong-ah, Jun-seok Kim, and Sung-yeon Kim. 2013. “Do Korean Voters Vote According to Policy? — An Analysis of the Relationship Between Policy Attitudes and Voting Behavior.” *Korean Political Science Review* 47(1), 167–183.

Kim, Sung-yeon. 2022. “Proximity Model, Directional Model, and Ideological Voting of Korean Voters: Analysis Results of Presidential Elections Since the 2010s.” *Journal of Future Politics* 12(2), 5-28.

______. 2023. “Ideological Polarization and Partisan Alignment of Korean Voters: Analysis Results of Five Presidential Elections Since the 21st Century.” *Korean Political Research* 32(3), 123-144.

Kim, Jeong. 2022. “Political Polarization and Vote Choice of Korean Voters: A Comparison of the 2012 and 2022 Presidential Elections.” *Korea and International Politics* 38(2), 169-198.

Ryu, Jae-seong. 2013. “Direction, Intensity, and Level of Political Ideology.” *Korean Party Studies Review* 12(1), 61-86.

Gasal, Jun, and Sung-jin Yoo. 2023. “A Study on the Ideologization of Public Conflict: Conflict Perception by Partisanship.” *Journal of Social Science Research* 31(2), 8-36.

Park, Young-deuk, and Han-na Kim. 2022. “Ambivalent Sexism Tendencies and Vote Choice of Young Korean Men: Analysis of the 2021 Seoul Mayoral By-election.” *Korean Political Science Review* 56(2), 5–35.

Yoon, Sung-yi, and Min-kyu Lee. 2014. “Intergenerational Characteristics of Ideological Conflict in Korean Society: Focusing on Determinants of Subjective Ideology and Ideological Representation.” *Journal of 21st Century Political Science* 24(3), 271-292.

Lee, Nae-young. 2009. “Changes in Korean Voters' Ideological Tendencies and Ideological Voting.” *Peace Studies* 17(2), 42-72.

Lee, Kap-yoon, and Hyun-woo Lee. 2008. “Analysis of the Influence of Ideological Voting: Composition, Measurement, and Meaning of Ideology.” *Journal of Contemporary Political Studies* 1(1), 137-166.

Lee, Dong-soo. 2025. “Men in their 20s and 30s Reject Lee Jae-myung's Presidency More Than Yoon's Impeachment.” *Hankook Ilbo*. 2025-02-03.

Lee, Dong-han. 2025. “Party Conflict and Ideological Conflict Have Become the Core of Conflict in Korean Society.” *Korea Research Public Opinion within Public Opinion*. 2025-05-20.

Lee, Hye-in. 2025. “Social Conflict Level Highest in the Last 6 Years... Progressive-Conservative Conflict Most Severe.” *Kyunghyang Shinmun*. 2025-03-03.

Lim, Jae-hyung. 2024. “A Study on the Status of Ideological Conflict in Korean Society and the Ideologization of Conflict.” *Dispute Resolution Studies* 22(3), 83-125.

Chung, Jin-min. 2003. “Ideological Tendencies of Korean Society and the Realignment of the Party System.” *Korean Party Studies Review* 2(1), 95–118.

Choi, Hyo-no. 2022. “A Study on the Conflict between Voters’ Subjective Ideological Tendency, Symbolic Ideology, and Attitudes toward Government Role and Size, Operational Ideology.” 『Journal of Parliamentary Studies』 17(1), 121–156.

Benoit, K. and Laver, M. 2007. “Estimating party policy positions: Comparing expert surveys and hand coding.” Electoral Studies 25(1), 182–191.

Bornschier, S. 2010. “The new cultural divide and the two-dimensional space in Western Europe.” West European Politics 33(3), 419–444.

Clarke, E. 2025. “Gen Z men are emerging as an outlier on gender norms.” plus61 Newsletter (e61 Institute).https://e61.in/new-data-shows-gen-z-men-are-more-likely-to-hold-traditional-gender-beliefs-than-older-men-and-far-more-so-than-their-female-peers/

Dalton, Russell J. 2000. “The Decline of Party Identifications.” Russell J. Dalton and Martin P. Wattenberg, eds. Parties Without Partisans: Political Change in Advanced Industrial Democracies,Oxford: Oxford University Press, 19–36.

______, Scott C. Flanagan, and Paul Allen Beck, eds. 1984.“ Electoral Change in Advanced Industrial Democracies: Realignment or Dealignment? ” Princeton, N.J.: Princeton University Press.

Downs, A. 1957. An Economic Theory of Democracy. New York: Harper & Row.

Embretson, S. E. and Reise, S. P. 2000. Item Response Theory for Psychologists. Mahwah, NJ: Lawrence Erlbaum Associates.

Evans, G. Ed.. 1999. “ The End of Class Politics? Class Voting in Comparative Context. Oxford: Oxford University Press.

Hambleton, R. K., Swaminathan, H., & Rogers, H. J. 1991. “ Fundamentals of Item Response Theory. Oxford: Oxford University Press.

Hassan, M. U. and Miller, F. 2022. “Discrimination with unidimensional and multidimensional item response theory models for educational data.” Communications in Statistics - Simulation and Computation 51(6), 2992–3012.

Heywood, A. 2021. Political Ideologies: An Introduction 7th ed.. London: Red Globe Press.

Hinich, M. J. and Munger, M. C. 1994. Ideology and the Theory of Political Choice. Ann Arbor: University of Michigan Press.

Hooghe, L. and Marks, G. 2018. “Cleavage theory meets Europe's crises: Lipset, Rokkan, and the transnational cleavage.” Journal of European Public Policy 25(1), 109–135.

Inglehart, R. 1977. The Silent Revolution: Changing Values and Political Styles among Western Publics. Princeton, NJ: Princeton University Press.

______. 1990. Culture Shift in Advanced Industrial Society. Princeton, NJ: Princeton University Press.

Jou, Willy and Koo, Sejin. 2019. “Redistribution Preferences and Ideological Orientations Across Countries. ” Chong-Min Park and Eric M. Uslaner, eds. Inequality and Democratic Polities in East Asia, Singapore: Palgrave Macmillan, 141–157.

Kang, W. T. 2008. “How ideology divides generations: The 2002 and 2004 South Korean elections.” Canadian Journal of Political Science 41(2), 461–480.

Kim, H. M., Choi, J. Y., & Cho, J. 2008.“Changing cleavage structure in new democracies: An empirical analysis of political cleavages in Korea.” Electoral Studies 27(1), 136–150.

Knutsen, O. 1995.“Value orientations, political conflicts and left-right identification: A comparative study.” European Journal of Political Research 28(1), 63–93.

______and Scarborough, Elinor. 1995.“Cleavage Politics.” Jan W. van Deth and Elinor Scarborough, eds. The Impact of Values, Oxford University Press, 493–523.

Kriesi, H. 2010.“Restructuration of partisan politics and the emergence of a new cleavage based on values.” West European Politics 33(3), 673–685.

Laponce, J. A. 1981. Left and Right: The Topography of Political Perceptions. Toronto: University of Toronto Press.

Lipset, S. M. and Rokkan, S. 1967. Party Systerns and Voter Alignments: Cross-National Perspectives. New York: Free Press.

Mair, Peter. 2007.“Left–Right Orientations.” Russell J. Dalton and Hans-Dieter Klingemann, eds. The Oxford Handbook of Political Behavior, Oxford: Oxford University Press, 206–222.

Milosav, Ð., Dickson, Z., Hobolt, S. B., Klüver, H., Kuhn, T., & Rodon, T. 2025. “The youth gender gap in support for the far right.” Journal of European Public Policy, 1–25.https://doi.org/10.1080/13501763.2025.2481181

Mieriņa, I., & Koroļeva, I. 2015.“Support for far right ideology and anti‐migrant attitudes among youth in Europe: A comparative analysis.” The Sociological Review 63(S2), 183–205.

Pedersen, M. N. 1979.“The dynamics of European party systerns: Changing patterns of electoral volatility.” European Journal of Political Research 7(1), 1–26.

Sartori, G. 1969.“From the sociology of politics to political sociology.” Government and Opposition 4(2), 195–214.

_________. 1976. Parties and Party Systerns: A Framework for Analysis. Cambridge: Cambridge University Press.

______. 1990.“The Sociology of Parties: A Critical Review.” Peter Mair, ed. The West European Party Systern, Oxford: Oxford University Press, 150–182.

Treier, S. and Hillygus, D. S. 2009.“The nature of political ideology in the contemporary electorate.” Public Opinion Quarterly 73(4), 679–703.

[1] The actual vote counts were Lee Jae-myung 49.2%, Kim Moon-soo 41.15%, Lee Jun-seok 8.34%, and Kwon Young-guk 0.98%, showing a slight difference from the survey results.

[2] However, as the sample size for Kwon Young-guk voters is very small, the reliability of the statistical estimation is low; therefore, caution is needed in interpreting these results.

[3] Among Lee Jun-seok voters, those who are in the subjective ideology range of 8-10 across all age groups total 14 individuals. Of these, 13 are under 39 years old. Of course, careful interpretation is required considering the insufficient sample size (89 Lee Jun-seok voters in the sample).

■ Author: Koo, Se-jin _Professor, Department of Political Science and International Relations, Inha University.

■ Contact and Editing: Lim, Jae-hyun_EAI Research Fellow

    Inquiries: 02 2277 1683 (ext. 209) | jhim@eai.or.kr