Youth Defection? A Comparison of Ideological and Policy Orientations and Voting Behavior in the 2017 and 2022 Presidential Elections

1. Introduction

In the 20th presidential election held on March 9, 2022, there was significant public interest in the voting behavior of young people, referred to as '20-something males' and '20-something females.' This heightened attention stems from the increasing importance of youth voting patterns in recent Korean elections. Academically, interest in the voting behavior of young people in Korean elections since democratization likely began with the research of Jeong Jin-min (1992), who raised the importance of generational factors. Young people also served as a comparative group to the '386 generation' that began to be actively discussed around the 2002 presidential election (Kang Won-taek 2009; Hwang Ah-ran 2009; Noh Hwan-hee et al. 2013). Particularly, as the proportion of young people participating in elections increased from the 2012 presidential election, interest grew regarding the social changes that new generations entering the electoral process might bring. As the influence of regionalism in voting behavior, which once dominated Korean elections, diminished, ideology and generational conflicts emerged as key factors determining election outcomes (Choi Jun-young and Cho Jin-man 2005). However, the voting behavior of young people in the recent presidential election garnered even more attention due to doubts about whether 'their political preferences have changed from the past.' This is because while their participation numbers were small, they previously showed a strong tendency to support progressive candidates, but their participation has increased while showing a tendency to support conservative candidates. This political preference materialized particularly in the 2021 Seoul mayoral by-election, raising the question of whether young people in South Korean society have become more conservative than in the past. This observation is a fascinating phenomenon that contradicts the theoretical prediction that young people, compared to the elderly, tend to choose progressive candidates.

This study aims to examine whether the preferences and behaviors exhibited by young people in the 20th presidential election were more conservative than in the past, in relation to these recent changes in South Korean society. Journalistic commentary on the conservatism of young people has actually appeared since the mid-2000s. However, among existing literature on Korean electoral processes, claims that young people in South Korean society are becoming more conservative are still difficult to find. Perhaps Jeong Jin-min's (2012) research, which suggests that young people in South Korean society have a strong tendency towards de-ideologization, is an exceptional study indicating that at least Korean youth may not be solely progressive. The scarcity of research focusing on young people can be attributed to several reasons. First, since the 2000s, interest in the impact of age and generational differences on election outcomes has been concentrated on the '386 generation.' The specific characteristics of the '386 generation,' assumed to have undergone political socialization within the context of South Korea's democratization process, were considered a structural factor that could exert a continuous and stable influence on Korean elections. Second, discussions on the progressive or conservative shifts of voters require panel data that surveys voters' ideological orientations and political behavior over extended periods and at multiple points in time. The lack of such data is another reason. Despite being a problem pointed out for a long time, the issue of data scarcity in political science has not yet been resolved. This study, which seeks to examine the issue of youth conservatism raised during the 2022 presidential election, also faces the limitation of data scarcity. However, its significance lies not only in its focus on youth but also in its achievement of tracing changes in the average perceptions of young people at a collective level by focusing on common questions included in survey data from different samples in 2017 and 2022. Furthermore, it attempts to infer the possibility of youth conservatism by examining whether voting behavior for conservative candidates at each point in time is related to age or generational factors.

This study will first address several conceptual and theoretical discussions to verify the conservatism of young people. In particular, it will review theoretical discussions from the perspectives of age, period, and cohort (APC) to define 'youth' conceptually and examine their voting behavior. Next, it will examine changes in the political perceptions and behaviors of young people at a collective level using post-election survey data from 2017 and 2022. Subsequently, it will analyze the determinants of voting behavior for conservative candidates at the individual level and observe whether the independent influence of young people is evident. Finally, the study will present a summary of the findings and their implications in the conclusion.

2. Theoretical Review

① Conceptual Definition and Conservatization of Youth

In the context of Korean elections, interest in 'young people' or 'youth' is often linked to generational conflict or differences. While generations are sometimes defined with socio-cultural labels such as the Millennial Generation or Generation Z, these sociological distinctions have not been academically validated. Rather, they are terms often used from a journalistic perspective to emphasize social and cultural backgrounds like the digital environment. Consequently, these terms carry the risk of defining individuals nearing 40 as 'youth'.

When viewing 'young people' and 'youth' with the possibility of generational conflict in mind, defining who they are and their quantitative composition is crucial for predicting the political outcomes of generational conflicts. This is because studying the political perceptions and behaviors of these groups without a clear definition risks selection bias, where some individuals are included or excluded from these groups. However, as the confusion in terminology suggests, there is a hurdle in that the definition of these groups is not socially consistent. Looking at dictionary definitions, 'young people' is not even a term listed in the Standard Korean Language Dictionary of the National Institute of Korean Language. The 'Uri Mal Saem,' an alternative Korean dictionary, defines it as 'relatively young people belonging to their 20s and 30s among members of society.' On the other hand, both the Standard Korean Language Dictionary and Uri Mal Saem define 'youth' as 'collectively referring to people in their youth among members of society.' According to this definition, 'young people' is closer to a term used in contrast to 'the elderly.' Many are likely to question the definition in Uri Mal Saem, which limits it to the 30s. In contrast, 'youth' is a term based on theoretical discussions that categorize people's life cycles into youth, middle age, old age, and elderly age, as a division within the categories of 'middle-aged,' 'elderly,' and 'old-aged.'

Second, even when aiming for terminological consistency and discussing the voting behavior of 'youth,' it remains unclear how many years of biological age should be considered youth. Defining youth based on biological age can also face criticism. Sociologically, youth is focused on certain events or social relationships experienced during the life course, making biological age less significant. Instead, youth is defined as a period of transition to adulthood, characterized by the transition from school to the labor market, from the family home to independent housing, and from the original family to a new family through marriage (Levinson et al. 1976). However, considering that these transition periods can vary by country and society, as well as by individual within a country or society, analyzing youth based on such definitions would require extensive individual-level information. Furthermore, even with such information, defining the transition period for each individual is likely to be challenging.

Even when using biological age, various ages are legally used to define youth. First, there are diverse legal definitions within South Korean society. For example, Article 3 of the Youth Basic Act, enacted in 2020, defines youth as individuals aged 19 to 34. Article 2 of the Enforcement Decree of the Youth Employment Promotion Special Act defines youth as those aged 15 to 29. The Small and Medium Enterprise Human Resources Development Special Act defines youth as those aged 15 to 34. The recently amended Public Official Election Act stipulates 18 as the voting age but has no separate definition for youth. Therefore, regarding electoral behavior, it seems reasonable to start defining youth from age 18. However, a decision must be made on how old youth should be considered. For instance, if the age group for whom candidates' youth-focused pledges apply is defined as youth in the electoral process, defining youth up to age 34 seems reasonable. However, this could be a very different definition from the tradition of defining youth primarily around the 20s. Alternatively, the definitions used by various international organizations such as the United Nations, the Organization for Economic Cooperation and Development (OECD), and the International Labour Organization (ILO), including those of several developed countries in Europe that define youth up to age 24, could be followed. In this case, considering the five-year presidential election cycle, young people in South Korean society might be overrepresented by those experiencing their first presidential election at each election period.

Based on this understanding, this study will categorize generations by five-year age intervals based on the 2022 presidential election and broadly define youth as up to age 39 as of 2022, then examine whether homogeneity or heterogeneity exists within the youth group. Specifically, by comparing their behavior in the 2017 presidential election with their behavior in 2022, we will examine the changes observed in the age group up to 34 in 2017 by 2022, and compare and review the characteristics of those who newly participated in the presidential election in 2022. More concretely, this study targets a total of 10 generations, divided into five-year intervals, starting with those born between 1993 and 1997, and those born between 1988 and 1992, based on individuals born in 1998, who are 24 years old or younger as of 2022.

Discussions on the conservatism of young people require not only operationalization of how to define the 'youth generation' but also an operational definition of the concept of 'conservatism.' Given that conservatism is related to a certain trend of change in ideological orientation, assessing conservatism requires data that surveys individual epistemological changes over a long period. However, as mentioned above, in the absence of panel data that surveys individuals' political perceptions over extended periods,[1]

it is difficult to verify conservatism at the individual level. Therefore, discussions on the conservatism of young people must be limited to examining average changes at a collective level among the generation defined as youth. Recognizing these limitations, this study will operationalize conservatism from two epistemological aspects and one behavioral aspect, and explore how these conservative tendencies change among young people between 2017 and 2022. For epistemological analysis, we will utilize self-placement ideological orientations and policy preferences expressed by respondents in the surveys. Specifically, for policy preferences, we will use questions regarding 'policy toward North Korea' and 'growth versus welfare policies.' Behaviorally, we will operationalize conservatism based on whether conservative candidates were supported. In particular, for the 2017 election, respondents who voted for candidates Hong Joon-pyo and Yoo Seong-min will both be operationalized as having supported conservative candidates.

② Age, Period, and Cohort (APC) Research

The recent trend of youth conservatism in Korean elections can be theoretically explained by three different factors. First, it is an explanation centered on the age effect according to the traditional life-cycle theory. The life-cycle theory posits that individuals experience psychological changes in ideological values and preferences as they age (Niemi and Hepburn 1995). Scholars who support this theory point out that as individuals age, they generally become more conservative and may develop authoritarian tendencies (Cornelis et al. 2009). They suggest that as people go through the aging process, they fear change and prioritize stability (Alwin and Krosnick 1991). Therefore, according to the life-cycle theory, the elderly are more likely to support conservative parties or candidates than young people, and conversely, young people are more likely to support progressive parties or candidates.

The conservatism of young people discussed in recent South Korean society does not necessarily overturn these theoretical predictions. In other words, it does not predict a South Korean society where young people are more conservative than the elderly. Rather, it implies that the theoretical prediction that conservative tendencies will increase linearly or non-linearly following the human life cycle may not be supported, due to the conservative tendencies of the youngest age group.

Second, the conservatism of young people can be examined through the generational effect (cohort effect) centered on the youth generation. The generational effect can be defined as individuals who have grown up under similar social backgrounds sharing certain political experiences and perceptions, leading them to make similar political decisions (Jennings and Niemi 1981). The pioneering research on the concept of generations can be found in Mannheim (1928). He divided the concept of 'generation' into three dimensions: generation location, generation as actuality, and generation unit. Generation location refers to a group of people born around the same time who are assigned similar positions in the course of social development. Generation as actuality refers to people in the same generation location developing a sense of solidarity and shared identity through participation as a community of fate. Finally, generation unit refers to units within the generation as actuality that have processed their participatory experiences in different ways. According to this conceptual framework, generations are formed against the backdrop of significant social events (generation location), require shared political participation within them (generation as actuality), and include more finely differentiated units within the shared political experiences (generation unit). Following this definition, a certain generation exhibits political behaviors distinct from other generations through the specific political and social perceptions they experienced in historical processes, and these political behaviors tend to be sustained over a long period.

Finally, the conservatism of young people may be a temporary observation due to the influence of a specific period, known as the period effect or temporal effect. The specific circumstances of the 2016 candlelight vigils and presidential impeachment might have led to a dominant progressive tendency throughout society, or conversely, a dominant conservative tendency might emerge under certain social environments.

The fact that the political orientations and behaviors of young people can be influenced by these three factors is further complicated by the fact that these three factors are related in the following linear equation:

Age = Period - Birth Year (Generation)

Therefore, if any two factors are fixed, the other factor is automatically determined. For example, at the time of the 2022 presidential election, the generation born in 1970 is subject to the influence of age 52. Consequently, methodologically, multicollinearity among these three factors leads to an identification problem in the process of estimating the influence of each factor, making it impossible to estimate their independent influences. Existing literature on Age, Period, and Cohort (APC) research has excluded or overlooked one of these three factors in the modeling process to address this problem. While recent proposals have suggested ways to overcome these limitations using hierarchical models (Yang and Land 2013), these methods also face numerous methodological criticisms (Bell and Jones 2014). Furthermore, the use of hierarchical models requires data that can encompass various time points, but survey data with identical survey questions for analysis at different time points is scarce. This study, using data from the 2017 and 2022 presidential elections, also does not fully overcome these limitations. The data is insufficient to control for period effects, and the same survey questions were not maintained across the two periods. Nevertheless, this study will discuss the conservatism of young people at an exploratory level using common survey questions that can explain the voting behavior of young people, through a multilevel logistic model.

3. Empirical Analysis

3. Empirical Analysis

① Data and Variables

This study utilizes panel survey data collected by the East Asia Institute during the 2017 and 2022 presidential elections. The advantage of using survey data collected by the same research institution lies in the similarity of survey questions asked during the two presidential election periods. It is particularly useful because it includes several important variables necessary for examining the conservative orientation or aspects of conservatism among young people.

First, the dependent variables of this study are as follows, allowing for an examination of conservative orientation or aspects of conservatism from epistemological and behavioral perspectives. First, to examine conservative orientation from an epistemological perspective, we used self-placement ideological orientation and policy preferences on two policy issues. Self-placement ideological orientation refers to respondents individually defining their ideological position on a given scale from 0 to 10. Here, 0 indicates a left-leaning ideological orientation, and 10 indicates a right-leaning ideological orientation. Of the two policy preferences, one is a question asking, 'What do you think about the current North Korean policy?' with options to choose between 'It is important to strengthen exchanges and cooperation between North and South Korea' and 'It is important to maintain a hardline policy toward North Korea.' The other is a question asking, 'Which is more important, welfare or growth in our society?' with options to choose between 'Welfare is more important' and 'Growth is more important.' Existing literature largely agrees that choosing 'exchanges and cooperation between North and South Korea' and 'welfare' represents a left-leaning or progressive orientation, while choosing 'hardline policy' and 'growth' represents a right-leaning or conservative orientation. Second, to examine conservative orientation from a behavioral perspective, we measured whether respondents voted for a conservative party candidate in the 2017 and 2022 elections as a binary variable. In the case of the 2017 presidential election, we assumed that candidates Hong Joon-pyo and Yoo Seong-min represented the conservative camp. Therefore, respondents who voted for either of these two candidates were coded as 1, and those who did not were coded as 0. Respondents who did not vote were treated as missing. Similarly, in the 2022 presidential election, the measurement was based on whether respondents voted for candidate Yoon Suk-yeol. To ensure consistency between the epistemological and behavioral analyses, the subjects for the epistemological analysis were also limited to respondents who voted in the 2017 and 2022 elections.

The independent variables of this study primarily include variables measuring age, period, and generation. Age was measured as the chronological age obtained by subtracting the birth year from the survey year. Period refers to 2017 and 2022, when each survey was conducted. Generation was measured in five-year intervals, starting with those born in 1998 (aged 24 in 2022), and extending to generations born before 1988, with the oldest generation being those born before 1957. Thus, generation was measured across 10 cohorts. To estimate the distribution and changes in average ideological orientation at the collective level, we examined the relationship between these generational variables and ideological orientation represented by self-placement ideological orientation and policy preferences. To verify the factors influencing voting behavior for conservative candidates, among these generational variables, generation was measured using three dummy variables representing young people. Considering the ambiguity in defining youth, this study subdivides generations into those aged 24 and under, those aged 25 to 29, and those aged 30 to 34, and measures each as a dummy variable.

Additionally, we utilized independent variables that are well-known as key factors in candidate selection in Korean elections: the candidate's personal image and policy aspects. To measure these two variables, we used a question asking the most important reason for voting for a specific candidate in the 2017 and 2022 elections. Respondents who chose 'candidate's personal ability' or 'candidate's morality' as the most important reason were considered to have chosen the candidate based on their personal image, and were coded as 1, while those who chose other reasons were coded as 0. Similarly, respondents who chose 'candidate's policy' or 'candidate's pledges' were measured as prioritizing the candidate's policy aspects. In the 2017 survey, candidate policy and pledges were presented as a single option, whereas the 2022 survey distinguished between them. Therefore, in the 2022 survey, respondents who chose either candidate policy or pledges were all measured as prioritizing candidate policy.

To measure voters' political orientations, another key variable influencing candidate selection in Korean elections, we utilized respondents' ideological orientation and party identification. Respondents' ideological orientation was measured by self-placement ideological orientation. Party identification was measured as a dummy variable indicating whether respondents identified with the Democratic Party or the People Power Party (or the United Future Party in 2017), respectively, for Democratic Party identification and conservative party identification. Finally, respondents' socio-demographic characteristics were controlled for. This includes respondents' gender, income, and region of residence. Gender was coded as a dummy variable with 1 for male and 0 for female. Income was measured using a variable with 11 intervals, ranging from less than 1 million won to over 10 million won per month for household income. Region of residence was measured using dummy variables for the five major metropolitan areas, with Seoul and Gyeonggi as the comparison units.

② Youth Conservatism: Epistemological Change at the Collective Level

Using survey data collected through identical questions in 2017 and 2022, we examined changes by age, generation, and period, focusing on self-placement ideological orientation and ideological orientation measured by policy preferences. [Figure 1] first shows the kernel density distribution of self-placement ideological orientation at the two time points. According to this, at both time points in 2017 and 2022, the ideological orientation of Korean voters remained predominantly centrist, with a relatively low frequency of voters expressing extreme ideological positions. Despite growing discussions about social polarization (Kim Seong-yeon 2015; Jang Seung-jin and Han Jeong-hoon 2021), a typical polarization with a low frequency of centrists and a high frequency of left- or right-leaning individuals does not appear to have occurred.

[Figure 1] Distribution of Self-Placement Ideological Orientation by Time Point

However, a change is observed that suggests a conservative shift has occurred in society as a whole during the five-year period from 2017 to 2022. In [Figure 1], compared to 2017, the frequency of respondents who marked their ideological orientation between 6 and 8, indicating a conservative orientation, increased in 2022, while the frequency of respondents who marked their ideological orientation between 2 and 4, indicating a progressive orientation, decreased. Furthermore, the frequency of respondents who defined their ideological orientation as 5 (centrist) also decreased in 2022 compared to 2017. Therefore, the conservatism of young people in 2022 should not overlook the possibility that it reflects the overall ideological distribution of society between these two time points.

[Figure 2] Comparison of Self-Placement Ideological Orientation Distribution by Age Cohort (2017 vs. 2022)

Second, we compared the self-placement ideological orientation of survey respondents by generation across the two time points, 2017 and 2022. [Figure 2] presents box plots of self-placement ideological orientation by generation at the two time points. In the figure, the box shows the distribution from the 25th percentile to the 75th percentile, and the whiskers represent the value calculated by multiplying 1.5 times the difference between the 75th percentile and the 25th percentile. The upper whisker extends to the value of the 75th percentile plus this calculated value, and the lower whisker extends to the value of the 25th percentile minus this calculated value. Values exceeding the 0 to 10 spectrum of ideological orientation are displayed at the extreme ends of the spectrum. Dots indicate the distribution of respondents exceeding these upper and lower whisker values. First, comparing 2017 and 2022, it can be assessed that conservative tendencies have strengthened across all generations. Based on the lower whisker of the box plot, excluding the two generations born between 1958 and 1962, and between 1963 and 1967, the lower whisker values in 2022 have either increased or remained the same compared to 2017 for all other generations. Considering that higher values on the ideological spectrum indicate a more conservative orientation, conservatism has progressed for these generations. Furthermore, the upper whisker values for the two generations whose lower whisker values decreased have also increased. Therefore, it can be argued that these two generations, born between 1958 and 1962, and between 1963 and 1967, have experienced increased ideological heterogeneity, if not conservatism. Considering that these generations largely constitute the '386 generation,' the increased ideological heterogeneity within these groups can be interpreted as a departure from their past progressive generational characteristics. Second, and more interestingly, conservatism has strengthened from the generation born between 1993 and 1997 (aged 20 to 24 in 2017) to the generation born between 1978 and 1982 (aged 35 to 39 in 2017). Specifically, the generation born between 1988 and 1992, who were aged 25 to 29 in 2017, shows that between the two time points, the ideological orientation of respondents from the 25th to the 75th percentile has become entirely distinct, indicating a strengthening of conservative orientation. Ultimately, [Figure 2] suggests the possibility of a generational effect where young people have strengthened their conservative orientation compared to middle-aged and elderly people when comparing generational ideological orientations between 2017 and 2022.

[Figure 3] Comparison of Self-Placement Ideological Orientation Distribution by Age (2017 vs. 2022)

Third, we examined the self-placement ideological orientation by age at the two time points, 2017 and 2022, using locally weighted scatterplot smoothing (LOESS). LOESS is a method used to visualize the relationship between two variables in a scatterplot when it is difficult to grasp intuitively, by representing it as a smoothed, non-linear relationship. The 2017 figure includes the LOESS curve within the 95% confidence interval of ideological orientation for each age, represented by a line. The 2022 figure includes the LOESS curve within the 95% confidence interval of ideological orientation, represented by an area. Examining the ideological distribution by age at the two time points, a typical age effect is observed after the two curves intersect around age 48. In other words, as people age beyond 48, their conservative orientation gradually increases, which is observed consistently at both time points in 2017 and 2022. In contrast, excluding the overlapping sections of the confidence intervals for ideological orientation by age at the two time points, a clear conservative shift is observed for ages approximately 36 and below. In 2017, the age effect of increasing conservatism with age is clearly evident from age 20 onwards. However, in 2022, from age 20 to 48, there are slight fluctuations, but conservatism tends to increase, and after age 48, conservatism increases.

In summary, examining the changes in ideological orientation at the collective level between 2017 and 2022 from the perspectives of period, generation, and age, we can conclude the following: First, there is a general tendency for conservatism to have strengthened in society as a whole in 2022 compared to 2017. Second, there is a tendency for conservatism to have relatively strengthened among generations born after 1978 in 2022 compared to 2017. The conservative orientation of the generation born between 1988 and 1992 is particularly noteworthy. Third, the possibility of conservatism is observed among those aged 48 and under in 2022 compared to 2017. In particular, due to the conservatism of those aged 35 and under, the tendency for conservatism to increase with age, known as the age effect, is not clearly evident. Based on these findings, it can be concluded that the conservative orientation of young people in South Korean society has strengthened at the collective level as of 2022. However, it is necessary to consider the limitation that 2017, the comparative period, might have been an exceptional time due to the unprecedented events of the 2016 candlelight vigils and presidential impeachment. Considering only 2022, those aged 48 and under are generally more progressive than older individuals, and it is possible that the age group between 40 and 50 might even represent a relative progressive generation reflecting the specific characteristics of South Korean society. Furthermore, the fact that some within the youth group exhibit even stronger conservative tendencies suggests that a blanket assessment of youth conservatism as a whole requires more rigorous verification. In particular, it is not clear whether these changes translate into differences in individual political choices. Before discussing this issue further below, we will examine whether the possibility of youth conservatism revealed in self-placement ideological orientation is also evident through policy preferences.

[Figure 4] Distribution of Preference for Growth-Oriented vs. Welfare-Oriented Policies by Generation and Time Point

Note: In [Figure], 0 represents respondents who prefer growth, and 1 represents respondents who prefer welfare.

First, let us examine preferences for growth-oriented and welfare-oriented policies, which are known to converge with left-right ideological spectrums according to existing literature. [Figure 4] shows the distribution of policy preferences by generation at the two time points, 2017 and 2022. First, two distinct trends are observed by generation. One is that in 2017, a clear difference is observed between the generation born before 1957 (aged 60 or older in 2017) and the generation born after 1978 (aged 39 or younger in 2017). Those aged 39 or younger in 2017 overwhelmingly preferred welfare over growth, while those aged 60 or older, corresponding to the elderly generation, preferred growth over welfare. The second is that a drastic change has occurred in the youth generation five years later, in 2022. The generation born between 1978 and 1987 still prioritizes welfare over growth, albeit by a narrow margin, while the generation born after 1988 has reversed this trend, prioritizing growth over welfare. Therefore, in terms of ideological orientation centered on economic policy, a conservative trend is observed among generations younger than 34 as of 2022.

[Figure 5] Distribution of Preference for Hardline North Korea Policy vs. Cooperative Policy by Generation and Time Point

Note: In [Figure], 0 represents respondents who prefer hardline policy, and 1 represents respondents who prefer cooperative policy.

[Figure 5] examines whether similar changes in youth preferences are observed regarding policy preferences related to North Korea, as in economic policy. Interestingly, regarding policy preferences related to North Korea, it is evident that the generation born between 1993 and 1997 (aged 20 to 24 at the time) preferred a hardline policy over a cooperative policy even in 2017. Within the youth group, preferences have already diverged regarding North Korean policy. This divergence has further expanded in 2022. The generation born between 1993 and 1997 still prefers a hardline policy at an even higher rate, and the generation born between 1988 and 1992 has also shifted from preferring a cooperative policy to preferring a hardline policy. These preferences of young people regarding North Korean policy are similar to those of the generation born before 1958 (aged 65 or older in 2022), and in fact, show a higher proportion of preference for hardline policies. This phenomenon seems to reflect the fact that during the 2022 presidential election, proposals such as 'preemptive strike against North Korea' from conservative candidates garnered significant support from young people.

In conclusion, examining the conservative trend based on preferences in economic and North Korean policies suggests that young people in South Korea are likely to have become more conservative in 2022 compared to 2017. Particularly in terms of economic policy, it can be said that conservative tendencies have strengthened among young people aged 34 and under as of 2022. Furthermore, regarding North Korean policy, the generation born between 1988 and 1992, who maintained progressive preferences in 2017, has shifted to conservative preferences in 2022. These results, when compared with the findings on self-placement ideological orientation, allow us to conclude that significant epistemological conservatism has progressed among young people aged 34 and under in South Korean society in 2022 compared to 2017.

Did this epistemological conservatism translate into actual individual choices of candidates? Below, we will examine through regression analysis whether the conservatism of young people has led to changes in political behavior.

③ Youth Conservatism: Behavioral Change at the Individual Level

This section aims to verify whether age or generational effects are observed in the support for conservative party candidates, while controlling for each time point of 2017 and 2022. Given the ambiguity in defining youth in South Korean society, we attempt to verify this by subdividing youth into generations aged 24 and under, 25 to 29, and 30 to 34 at each time point. As discussed in the theoretical review, due to the linear relationship between age, period, and generation, simultaneously verifying the influence of all three variables is methodologically very challenging. While recent verification methods using hierarchical models have been proposed (Yang and Land 2013), methodological debates persist, and there is a lack of data in Korean elections that allows for simultaneous consideration of multiple time points. Therefore, this study controls for time by independently verifying each time point, following the approach of many previous studies that attempted to verify age and generational effects.

The dependent variable used for the regression analysis is voting for a conservative party candidate. Specifically, considering that candidates Hong Joon-pyo and Yoo Seong-min competed in the conservative camp in the 2017 election, we consider voting for either of these two candidates as voting for a conservative party candidate. Therefore, to analyze the binary variable of voting for a conservative party candidate, we employed a logistic regression model. [Table 1] presents the results of the analysis of factors influencing the choice of a conservative party candidate in the 2017 presidential election. First, the basic model examines the influence of socio-demographic factors, political factors, and candidate-specific and policy factors, without considering generational influence. The analysis results indicate that factors significantly influencing the choice of a conservative party candidate in the 2017 presidential election were region of residence, party identification, the respondent's individual ideological orientation, and candidate policy factors. Regarding region of residence, respondents residing in the Honam region were less likely to choose a conservative party candidate compared to those in Seoul and Gyeonggi, while residents in the Gyeongbuk or Gyeongnam regions were more likely to choose a conservative party candidate. The influence of party identification, observed in previous literature, is also evident in the 2017 presidential election. Voters identifying with the Democratic Party showed a strong tendency not to vote for conservative candidates, while voters identifying with the conservative party showed a strong tendency to vote for conservative candidates. The respondent's individual ideological orientation is also consistent with the results of previous literature; voters with a conservative ideological orientation were more likely to choose a conservative party candidate. Interestingly, in the 2017 presidential election, voters chose conservative party candidates based on their policies or pledges rather than their personal abilities or morality.

[Table 1] Analysis of Factors Influencing Conservative Candidate Choice in the 2017 Presidential Election

f2e34dd1a320d063

f2e34dd1a320d063

f2e34dd1a320d063

** p<0.05, * p<0.1

The support for conservative party candidates among the youth, which is of interest in this study, can be inferred through the age variable in the basic model. The theoretical prediction regarding the age effect is that younger individuals will choose progressive candidates and older individuals will choose conservative candidates. Therefore, the positive regression coefficient of the age variable aligns with this theoretical prediction. However, the fact that the regression coefficient for the age variable is not statistically significant implies that younger voters may have shown stronger support for conservative candidates, or older voters may have shown stronger support for progressive candidates. Given that the 2017 presidential election took place immediately after the 2016 candlelight vigils and the impeachment of the president, it is difficult to rule out the possibility of the latter scenario. Therefore, it appears difficult to predict the likelihood of young people choosing conservative candidates based solely on the results of the basic model.

To improve upon the basic model, which focused on the age effect, three additional models were analyzed. The first model included the generation aged 24 and under at the time of the 2017 election and considered generational effects. The second model included the generation aged 24 and under, as well as the generation aged 25 to 29. The final model included three generations in five-year increments: those aged 24 and under, those aged 25 to 29, and those aged 30 to 34. The analysis results of the models show that the estimation results did not significantly improve in the first and second models. Compared to the basic model, the log-likelihood values of these two models showed little improvement. Furthermore, neither the age nor the generational effects showed statistically significant results for supporting conservative candidates. In contrast, the final model revealed that the generation aged 30 to 34 (born between 1983 and 1987) at the time of the 2017 election had a significant influence on supporting conservative candidates. Compared to the older youth and adult generations aged 35 and over, who were not included in the model, this generation showed a stronger tendency to choose conservative candidates in 2017. Interestingly, the age effect was also observed simultaneously in the final model. This implies that when the influence of the 30-34 age group supporting conservative candidates is independently verified as a generational variable, a tendency for older individuals to support conservative candidates compared to younger individuals is observed. Considering these results, it can be inferred that although the 30-34 age group in 2017 was not epistemologically particularly more conservative, they exhibited a strong tendency to support conservative candidates politically. Consequently, it might be inferred that the remaining younger age groups did not support the Democratic Party candidates as much compared to older voters.

To evaluate whether the conservative tendency of the youth has strengthened during the 2022 presidential election process and whether this conservative tendency translated into voting behavior, the following outcomes would need to be observed, based on the results in [Table 1]. First, a tendency to support conservative candidates should emerge in the generations aged 24 and under, and 25 to 29. When a tendency to support conservative candidates is revealed in youth generations that did not show such a tendency in 2017, it can be assessed that the youth have become more conservative in terms of political behavior. Second, the age effect should not be significant when controlling for youth generations up to age 34. Since the age effect became significant due to the generational effect of the 30-34 age group during the 2017 presidential election, if the age effect is not significant in the 2022 presidential election while controlling for these generations, it is possible to infer the overall conservatism of the youth.

Examining the results of the basic model in [Table 2], the influence of ideology and party identification is similar to that of the 2017 presidential election. In contrast, regarding residential area, only voters in the Honam region showed a tendency not to vote for conservative candidates compared to voters in the Seoul and Gyeonggi regions; other regions showed no significant difference compared to Seoul and Gyeonggi. Furthermore, in the 2022 presidential election, it was observed that the less candidates' personal abilities, morality, or policies were considered, the less likely voters were to vote for conservative candidates. Regarding the age effect, which is the focus of this study, it was found that the older the voters, the more likely they were to vote for conservative candidates, and this effect was statistically significant. Notably, this age influence was consistently observed in the remaining three models after controlling for generational effects. Even when the youth generations were segmented into three groups and their influence was tested, the influence of each group was not significant.

[Table 2] Analysis of Factors Influencing Conservative Candidate Choice in the 2022 Presidential Election

f2e34dd1a320d063

f2e34dd1a320d063

f2e34dd1a320d063

** p<0.05, * p<0.1

Consequently, it cannot be inferred that the conservative shift among youth in the 2022 presidential election led to greater support for the conservative party candidate among youth, or that specific age cohorts within youth supported the conservative candidate more. This does not necessarily mean that youth did not become more conservative. It could mean that even if youth became more conservative, this conservatism did not manifest as a greater tendency to support the conservative candidate compared to other age groups. Alternatively, the conservative shift among youth might have led them to support the conservative candidate similarly to other age groups. However, considering the statistical significance of the age effect, it is unlikely that youth supported the conservative candidate in a manner similar to other age groups. Therefore, an interpretation that, despite the potential for a conservative shift in terms of perception, this shift was not fully realized in actual candidate choice appears to be the most reasonable.

4. Conclusion

This paper sought to answer two questions based on the social concern regarding the conservative shift among youth in the 2022 presidential election. First, did youth in South Korea become more conservative in the 2022 presidential election compared to the past? Second, if a conservative tendency among youth is observed, did this tendency translate into voting behavior that supported the conservative party candidate? Answering both questions simultaneously would not only enhance our understanding of South Korean youth but also allow for predictions about the future direction of South Korean society as the youth politically replace older generations.

To this end, the study began by defining the concept of youth. Given that youth in South Korean society are defined in various social and legal ways, the understanding of youth can differ depending on how they are defined. Therefore, this study first posited youth as a group under 24 years of age according to general Western standards, and then examined which age groups could be considered as a homogeneous cohort of youth by subdividing them into three groups in five-year increments: ages 25-29 and 30-34. Furthermore, by establishing age cohorts based on their birth years, the study aimed to verify which generational cohorts could be encompassed by the assessment that youth in South Korean society are tending towards conservatism.

However, due to data limitations required for verifying the conservatism of youth at the individual level, this study compared data from two points in time, 2017 and 2022, at the aggregate level. The analysis results indicated that in 2022, South Korean society as a whole had become more conservative compared to 2017, not only in terms of self-defined ideological orientation but also within the age groups under 35 and the three subdivided age cohorts. Notably, the ideological conservative orientation was prominent in the age cohort born between 1988 and 1992, corresponding to those aged 30-34 in 2022. This aggregate-level ideological shift was further confirmed through preferences for economic and North Korea policies. Thus, the conservative shift among youth in South Korea, which was a concern during the 2022 presidential election, was found to have occurred broadly up to the age of 34 in 2022.

Another interesting finding of this study is that despite the perceived conservatism of youth in South Korean society at the aggregate level, this conservatism does not appear to be closely linked to the choice of conservative party candidates. Among the various factors influencing the choice of conservative party candidates in the 2017 and 2022 presidential elections, the subdivided youth age cohorts did not have a significant impact. Although the age cohort born between 1988 and 1992 showed a stronger tendency to vote for the conservative candidate in the 2017 election, it is not valid to infer that youth as a whole chose the conservative candidate based on the effect of this particular age cohort. This is because the significant age effect in the 2017 election implies that youth outside the 1988-1992 cohort were relatively less likely to support the conservative candidate. Furthermore, in the 2022 election, none of the three subdivided youth age cohorts had a significant influence on the choice of the conservative candidate. In contrast, the age effect was clearly observed, indicating that youth had a weaker tendency to choose the conservative candidate.

From the findings of this study, it can be inferred that the perceived conservative shift among youth in the 2022 presidential election did not translate into a behavioral conservative shift in supporting the conservative party candidate. Considering that the fundamental reasons for the strengthened conservative orientation of youth in 2022 have not been identified, this conservative shift observed only in perceptual terms may reflect the specific characteristics of the year 2022. To overcome the limitations of this inference, it is necessary to track changes in the ideological orientations of youth over a longer period and to identify the factors underlying such changes. ■

References

Kang, Won-taek. 2009. “Where Did the 386 Generation Go?: Ideology and Generation in the 2007 Presidential Election and 2008 General Election.” In Kim Min-jeon and Lee Nae-young (eds.), *Changing South Korean Voters 3*, EAI, pp. 69-96.

Kim, Sung-yeon. 2015. “Political Attitude and Perception Polarization, Partisan Bias, and Democracy.” *Democracy and Human Rights* 15(3): 459-491.

Noh, Hwan-hee, Song, Jeong-min, and Kang, Won-taek. 2013. “Generational Effects in South Korean Elections—Focusing on Presidential Elections from 1997 to 2012.” *Journal of Korean Political Parties* 12(1): 113-140.

Jang, Seung-jin, and Han, Jeong-hoon. 2021. “Does YouTube Politically Polarize Users? A Survey Analysis of Subscribers to Major Political and Current Affairs YouTube Channels.” *Journal of Modern Political Studies* 14(2): 5-35.

Chung, Jin-min. 1992. “Generational Factors in South Korean Elections.” *Korean Political Science Review* 26(1): 145-167.

Chung, Jin-min. 2012. Voting Behavior and Generation of South Korean Voters: Focusing on the 2010 Local Elections. *Korean Political Research* 21(2): 1-21.

Choi, Jun-young, and Cho, Jin-man. 2005. “An Empirical Study on the Potential for Change in Regional Cleavages: Focusing on the Effects of Ideological and Generational Cleavages in the 17th National Assembly Election.” *Korean Political Science Review* 39(3): 375-394.

Hwang, Ah-ran. 2009. “Political Generations and Ideological Orientation: Focusing on the Democratization Achievement Generation.” *Journal of National Strategy* 15(2): 123-151.

Alwin, D. F., and J. A. Krosnick. 1991. “Aging, Cohorts and the Stability of Sociopolitical Orientations over the Life Span”, American Journal of Sociology, 97(1), 169-195.

Bartels, M. M., and L. L. Carstensen. 1999. “Social-psychological theories and their application ot aging: from indiviudal to collective”, In V. L. Bengtson and K. W. Schaie (eds.) Handbook of Theories of Aging. Springer Publishing Company.

Cornelis, I., A. Van Hiel, A. Roets, and M. Kossowska. 2009. “Age Differences in Conservatism: Evidence on the Mediating Effects of Personality and Cognitive Style”, Journal of Personality, 77(1), 51-88.

Converse, P. E. 1976. Dynamics of Party Support: Cohort Analytical Party Identification. Sage Publication.

Jennings, M. K., & R. G. Niemi. 1981. Generations and politics. Princeton, NJ: Princeton University Press.

Levinson, D. J., M. D. Charlotte, E. B. Klein, M. H. Levinson, B. McKee. 1976. “Periods in the Adult Development of Men: Ages 18 to 45” Counseling Psychologist 6(1):21-25.

Mannheim, K. 1928. The Problem of Generations. London: Routledge.

Niemi, Richard G., and Mary A. Hepburn. 1995. “The Rebirth of Political Socialization”, Perspectives on Political Science, 24(1), 7-16.

[1] Exceptionally, the East Asia Institute (EAI) possesses panel survey data, but this also has the limitation of not being a long-term survey.

■ Author: Han Jeong-hoon_Professor at the Graduate School of International Studies, Seoul National University, and Director of the EU Center. He lectures on Korean politics, parliamentary and party politics, and comparative politics. He received his Ph.D. from the University of Rochester in New York, USA, and taught as a professor in the Department of Political Science at Soongsil University before joining Seoul National University. His main research interests include electoral and parliamentary systems, party politics and parliamentary politics, and the politics of the European Union. He has published articles in numerous major international and domestic journals, including European Union Politics, Journal of European Public Policy, Korea Observer, Contemporary Politics, and Korean Political Science Review.

■ Contact and Editor: Jeon Ju-hyeon_EAI Research Fellow

    Inquiries: 82 2 2277 1683 (ext. 204) | jhjun@eai.or.kr