The End of the '86ers' Generation and Generational Effects: An Analysis of Presidential Elections from 1992-2022

1. Introduction

The purpose of this paper is to identify the political characteristics of the voting behavior and ideological orientation of the so-called '86ers' generation in past presidential elections, including the 2022 presidential election. The '86ers' have been recognized as synonymous with Korean democratization. Born in the 1960s, they acquired their political identity during their youth in the 1980s, a period of democratization. Their 'progressiveness' observed around the 2002 presidential election, when they were in their thirties, left a deep impression on the public (Kang Won-taek 2003; Lee Nae-young 2002). This has not always been the case. After the 2007 presidential election, discussions about the 'disappearance of the '86ers'' emerged (Kang Won-taek 2009; Park Myung-ho 2009; Noh Hwan-hee et al. 2013). Debates have continued regarding whether they are a homogeneous group and whether they are ideologically or politically 'progressive.' The question of whether they were 'progressive' and then became 'conservative,' or were never 'progressive' in the first place, remains contentious.

In election research, the aging effect and cohort effect remain competing theories regarding generational voting. It is a very common phenomenon for younger voters to exhibit progressive tendencies and older voters to exhibit conservative tendencies, collectively referred to as the aging effect (Feather, 1977; Barnes 1989; Alwin and Krosnick, 1991; Van Hiel and Brebels, 2011). It is generally accepted that older generations support conservative parties, as an attachment to authority and order aligns with conservative values. Those who rejected established authority are generally no different. It is argued that as income increases and social status stabilizes, it becomes difficult to avoid becoming conservative.

In contrast, the cohort effect refers to phenomena that do not conform to this aging effect. When a particular generation deviates from the aging effect, its unique political characteristics are referred to as the cohort effect. This is because the political orientation formed by the generation's specific socialization experiences manifests differently from the general aging effect (Mannheim 1952; Abramson 1975). Cases where individuals are conservative during their youth or remain progressive in old age are representative examples of the cohort effect. The '86ers' generation has been expected to be an exception to the aging effect. There was an expectation that the tendency for conservatism with age might not apply to this generation, and that the identity acquired through their experience of Korea's democratization period in their youth would be maintained as they aged, unlike the aging effect. This is the so-called 'cohort' effect, similar to the New Deal generation in the United States.

The 1960s generation, collectively referred to as the '86ers,' and the 2022 presidential election are special in two respects. First, the 1960s generation is the only cohort for which the aging effect and cohort effect in Korean electoral politics can be empirically verified (Bae Jin-seok 2017). This is because there are no longitudinal survey data comparing the individual voting preferences of the 1940s and 1950s generations during their youth and middle age. Presidential election surveys were first conducted during the 1992 presidential election. At the time of the 1992 election, the 1940s generation was already in their mid-40s to early 50s, and the 1950s generation was in their mid-30s to early 40s. In other words, there is no empirical way to ascertain their voting preferences in their 20s and 30s. Second, the 2022 presidential election is the first election in which all members of the 1960s generation have fully entered their 50s. As is well known, the symbolic significance of one's 50s in the life cycle of generational voting is considerable (Truett, 1993; Heo Seok-jae 2014). Voters who were progressive in their youth are highly likely to become conservative by their 50s, making the 50s a turning point for conservatism in generational studies. Generations after the 1960s have not yet reached their 50s as of the 2022 election. For example, if the 1970s generation shows progressive voting preferences, it cannot be called a cohort effect. If we assumed that the '86ers' exhibited progressive voting preferences before their 50s and approached this from a cohort effect perspective, it would be a premature diagnosis. Discussing the cohort effect for the 1960s generation before they reached their 50s was indeed premature.

This paper examines the generational factors observed in the 2022 presidential election results and empirically analyzes the scope and speed of changes in generational voting choices and ideological orientations observed in past presidential elections since democratization. From a cohort effect perspective, it focuses on the voting choices and ideological shifts of the 1960s generation, comparing them with preceding and succeeding generations, while also paying attention to intra-generational differences, which are no less significant than inter-generational differences. Chapter 2 reviews the achievements and limitations of related prior research, and Chapter 3 introduces the analysis methods. Using aggregate and individual-level survey data, Chapter 4 analyzes the 2022 presidential election, and Chapter 5 explores answers to the research questions using pooled data from the 1992-2022 presidential elections. Chapter 6 discusses the conclusions and theoretical implications.

2. Literature Review

Korean election studies related to age or generation have primarily focused on the cohort effect. Following pioneering studies by Jung Jin-min (1992, 1994), the cohort effect became a subject of significant academic interest due to the political role of the '386 generation' that gained attention in the 2002 presidential election (Kang Won-taek 2003; Lee Nae-young 2002). As a generation that experienced democratization in their twenties, their progressive orientation was noteworthy. In the 2007 presidential election, when Lee Myung-bak of the Grand National Party received broad support from both middle-aged and young voters, questions were raised about the generational effect of the '386 generation.' However, the significant differences in generational support between Park Geun-hye and Moon Jae-in in the 2012 presidential election rekindled academic interest in generational voting (Noh Hwan-hee et al. 2013; Heo Seok-jae 2014; Oh Se-jae/Lee Hyun-woo 2014; Oh Se-jae 2015; Moon Woo-jin 2016).

The political orientation and voting behavior of the 1960s generation have been largely divided into two perspectives. The perspective emphasizing the aging effect focused on their conservatism, while the perspective advocating the cohort effect focused on the progressiveness observed in the 1960s generation. The 17th presidential election in 2007, won overwhelmingly by President Lee Myung-bak, led to numerous studies suggesting the demise of the '386 generation effect' (Seo Hyun-jin 2008; Kang Won-taek 2009; Park Myung-ho 2009; Park Won-ho 2012; Heo Seok-jae 2014). This was because a significant portion of the middle-aged and young adult demographic, including the 1960s generation, supported Lee Myung-bak, a conservative candidate. The '386 generation' also garnered attention in the 18th presidential election in 2012 (Noh Hwan-hee et al. 2013). This time, consistent progressive characteristics were found within the '386 generation,' evidenced by their voting choices for progressive candidates and their self-reported ideological evaluations. Conditional '386 generation effects' also received attention (Hwang Ah-ran 2009; Oh Se-je/Lee Hyun-woo 2014; Oh Se-jae 2015).

This study questions the appropriateness of previous research approaching the progressive political orientation of the 1960s generation from a cohort effect perspective. From a life-cycle perspective, it was premature to interpret the progressive political orientation of the 1960s generation, who had not yet reached their 50s, as a cohort effect (Bae Jin-seok 2017). It is only with the 2017 presidential election, when a considerable number of the 1960s generation entered their 50s, and the 2022 presidential election, when they fully entered their 50s, that it becomes possible to discuss the cohort effect of the 1960s generation. This election, where they have all fully entered their 50s, a potential turning point for conservatism, provides a valuable opportunity to verify whether the 1960s generation has embarked on a path of conservatism like previous generations (aging effect) or maintained their unique generational political orientation (cohort effect).

Another point of contention is the definition of a generation. The cohort effect is based on the premise that the specific political circumstances experienced by a generation during their socialization period influence their political orientation. The issue is how to define this 'generation.' The criteria for generational division in Korean election studies remain controversial. Early studies began with divisions such as the pre-democratization generation (~1949), the democratization generation (1950-1961), and the new generation (1962-1970) (Jung Jin-min 1992, 1994; Jung Jin-min & Hwang Ah-ran 1999). Subsequent studies developed more refined classifications. The industrialization generation (~1959), the '386 generation' (1960 or 1962-1969 or 1970), the Generation X/IMF generation/digital generation, etc., have been primarily discussed (Kim Hyung-jun 2006; Cho and Eom 2012; Noh Hwan-hee et al. 2013; Lee Woo-jin 2013; Heo Seok-jae 2014; Oh Se-je 2015). Cohort effects cannot be simply defined by biological age; the political circumstances influencing their socialization period are crucial (Oh Se-je 2015). Nevertheless, the 1960s generation is generally classified as an independent generation. This classification is particularly persuasive given that the 1960s generation is the only cohort whose political orientation in their 20s can be observed as it changes through their 50s, a turning point for conservatism in Korean election history.

3. Analysis Methods

This paper utilizes aggregate data and individual-level survey data. Chapter 4 analyzes the results of the 2022 presidential election, and Chapter 5 analyzes pooled survey data from presidential elections from 1992 to 2022. The data used in this study are post-election survey data collected every five years from the 14th presidential election in 1992 to the 20th presidential election in 2022, covering a 30-year period. The data for the 14th-16th presidential elections are from interview surveys conducted by the Korean Social Science Data Center (KSDC), and the data for the 17th-20th presidential elections are from the East Asia Institute (EAI) Presidential Election Panel Survey.

① Aggregate Data

Aggregate data analyses in previous studies lacked sufficient control for period effects in elections. There is a significant difference between the 16th presidential election in 2002, which showed a pronounced progressive tendency, and the 17th presidential election in 2007, which showed a pronounced conservative tendency. Therefore, a simple comparison of support rates for conservative or progressive candidates can be problematic. This implies that period effects, which can arise from the ideological landscape of a particular election and campaign mobilization strategies, were not adequately reflected in the analysis. The approach of Noh Hwan-hee et al. (2013), which compared the relative positions of generations in terms of voting preferences, is noteworthy.

To overcome these limitations, this study analyzes aggregate data using the conservative advantage rate, mean centering, and a life-cycle comparison strategy. First, the conservative advantage rate is calculated as the difference between the conservative support rate and the progressive support rate. This approach considers the difficulty in accurately grasping the power relationship between the two groups when simply comparing conservative or progressive support rates. In particular, simple conservative or progressive support rates can have limitations as they can vary significantly depending on whether an election is a two-candidate or multi-candidate race.[1]Using the conservative advantage rate can resolve this issue. Second, the mean centering technique is used to control for period effects in each election. Considering that the ideological landscape of voters changes significantly with each election, this method aims to identify differences in generational voting behavior by subtracting the generational average from the average value for each election period. Third, generational ideological and voting preferences are compared along the life-cycle axis. Instead of comparing elections, the political characteristics of each generation across their life cycles are compared. The scope and speed of changes in generational political ideology will be analyzed by using 10-year intervals as the horizontal axis.

② Individual-Level Survey Data

This study agrees with the prior research claim (Heo Seok-jae 2014) that the cohort effect emerges 'despite' controlling for the age variable. However, to measure this, including both the age variable and the generation variable simultaneously in a regression equation inevitably leads to multicollinearity issues.[2]

To address this problem, this study employs a strategy that distinguishes between aging effects and cohort effects within the regression equation. Age and generation variables are included simultaneously, but not all generation dummy variables are included at once; rather, separate regression equations are constructed by including them individually. This method has the significant advantage of resolving the multicollinearity issue, but requires caution in interpreting the generation dummy variables.[3]

The primary dependent variable for voting choice in the 2022 presidential election is support for the conservative candidate. Support for the conservative candidate is coded as 1, and opposition as 0. In the 2022 presidential election, the progressive candidates included Lee Jae-myung and Sim Sang-jung, while the conservative candidate was Yoon Suk-yeol. The main independent variables of interest are age and generation. Age is based on the respondent's chronological age or, in some cases, divided into 10-year brackets such as 20s, 30s, etc. Control variables included in the analysis are ideology (ordinal variable), education level (ordinal variable), gender (dummy variable), region of origin (Hohnam and Yeongnam dummy variables), presidential job performance evaluation (continuous variable), and approval of strengthening inter-Korean exchange and cooperation and expanding welfare policies (dummy variables).

In the repeated cross-sectional analysis to verify the cohort effect of the 1960s generation, the dependent variable is also support for the conservative candidate. The independent variables of interest are generation dummy variables, categorized in 10-year intervals as the 1960s generation, preceding generations, and succeeding generations. This model includes control variables such as age, subjective ideological self-placement, education level, gender, and region of origin. Additionally, to control for varying period effects across elections, dummy variables for the seven presidential elections were included in the regression equation. Since support for the conservative candidate is the dependent variable in the above analyses, logistic regression was used.

In the analysis to identify generational ideological characteristics, the dependent variable is the respondent's self-rated ideology. An 11-point scale was used as the base, and for some KSDC data that used a 5-point scale, it was adjusted to an 11-point scale. In this analysis, the main independent variables of interest are age and generation variables. Since the dependent variable was measured on an ordinal scale, linear regression analysis was used for this analysis.

4. The 2022 Election

The 2022 presidential election shows no distinct differences based on age or generation compared to the 2017 and 2012 elections. As confirmed in [Figure 1], the differences in voting tendencies between generations have also significantly diminished in the 2022 election. This is markedly different from previous elections where young and middle-aged adults supported progressive candidates, and middle-aged and older adults supported conservative candidates. This trend is even more pronounced when compared to the 2012 presidential election, which was a two-candidate race. It also differs from the generational voting differences observed in the 2017 presidential election. Generally, the voting choices of the 1960s generation in the 2022 presidential election closely align with those of the overall electorate ([See Figure 1]).

[Figure 1] Generational Voting Tendencies in Recent Presidential Elections

The variables that significantly influenced voters' candidate choices in this election were subjective ideological perception, the evaluation of President Moon Jae-in's job performance, and preferences regarding North Korea policy. Socio-demographic characteristics known to influence voting choices, such as age, generation, gender, region, education level, and income, had little impact. Of course, political issues that emerged during the campaign, such as the moral issues surrounding the candidates and their families from both camps, had a significant influence. When all these variables were controlled for, the influential variables were ideology, presidential job performance evaluation, and North Korea policy preferences.

[Table 1] Determinants of Yoon Suk-yeol's Vote Choice: 2022 Presidential Election

In summary: Voters chose Yoon Suk-yeol if they were more conservative, evaluated President Moon Jae-in's job performance negatively, and favored a hardline stance on North Korea. Conversely, voters chose Lee Jae-myung if they were more progressive, evaluated President Moon Jae-in's job performance positively, and favored a conciliatory approach toward North Korea. One's generational affiliation had little impact. The '86ers' were no exception. Generational identity did not influence voting choices. The '86ers' behaved similarly to other voters, choosing candidates based on ideology, presidential job performance evaluation, and stance on North Korea policy ([See Table 1]). The only statistically significant finding related to cohort effects was that the probability of supporting Yoon Suk-yeol decreased for those born in the 1970s.

Among the 1960s generation, 42% identified themselves as ideologically centrist, the highest proportion ([See Figure 2]). Conservatives accounted for about 32%, and progressives for about 25%. This is nearly identical to the average for all voters. When scoring from 0 (most progressive) to 10 (most conservative), the average subjective ideological perception for the 1960s generation was 5.15, slightly leaning towards conservative from the center. The average ideological perception for all voters was 5.28, indicating they were nearly similar or slightly more progressive. They fall between the 1970s generation, who are the most progressive with an average of 4.79, and the 1940s-1950s generations, who are considerably more conservative. They were slightly more progressive than the 1990s generation, who are reportedly becoming more conservative.

[Figure 2] Subjective Ideological Perception by Generation

Interestingly, the generational subjective ideological perceptions observed in the 2022 presidential election exhibit a 'U-shaped' pattern. In previous presidential elections, younger generations generally perceived themselves as progressive, and as age increased, they tended to perceive themselves as conservative. This suggests a change in the linear relationship between age and ideological perception in the 2022 presidential election. As age increases, ideological perception becomes negative, and then, after a certain point, becomes positive again. Therefore, it is necessary to examine the coefficients in the regression model equation to confirm the U-shaped curve in a quadratic regression analysis.

y = b0 + b1*x1 +b2*x1^2

To confirm the U-shaped curve in the equation above, b2 must be positive, and b1 must be negative. In [Table 2] below, Model (1) shows the linear relationship between age and ideological perception, while Model (2) is the regression equation to confirm the non-linear U-shaped curve. In Model (1), age is not statistically significant, implying that the relationship may not be simply linear. In Model (2), as expected, the coefficient for the age variable (b1) is negative, and the coefficient for the squared age variable (b2) is positive. This means that as age increases, people perceive themselves as more progressive, but after a certain age, they begin to perceive themselves as more conservative. The inflection point is confirmed at x = -b1/2b2, which is 42.62 years old. This implies that from age 18 onwards, as age increases, individuals perceive themselves as more progressive, reaching a peak at 42.62 years old, after which, as age increases, they perceive themselves as more conservative.

The 1960s generation gave President Moon Jae-in's job performance a score of 4.8 out of 10 ([See Figure 3]), which is exactly the same as the average for all voters. While this was more negative than the evaluation by the 1970s and 1980s generations (in the mid-to-upper 5s), it was more positive than that of the preceding generations (in the low 4s).

61% of the 1960s generation preferred reconciliation and cooperation policies toward North Korea over hardline policies ([See Figure 4]). This is the same proportion as the 1970s generation, who perceive themselves as slightly more progressive than the 1960s generation. This proportion is about 10 percentage points higher than the average for all voters. The so-called inverted U-shaped pattern observed for over a decade has reappeared, where young and old voters prefer hardline policies, and middle-aged voters prefer reconciliation and cooperation policies. This contradicts the previous notion that the young prefer reconciliation and cooperation, while the middle-aged prefer hardline measures. Regarding the priority between growth and welfare, 39% of the 1960s generation prioritized welfare, which is lower than the overall voter average of 46%. A notable characteristic is that the difference between preferring reconciliation and cooperation policies toward North Korea and prioritizing welfare was largest among the 1960s generation. While the difference between these two variables was negligible or small for most generations, it was pronounced for the 1990s generation and the 1960s generation.

[Table 2] Determinants of Ideological Perception: 2022 Presidential Election

Interestingly, this pattern is the exact opposite of that of the 1990s generation. While a majority of the 1960s generation, the parental generation, are positive towards reconciliation and cooperation policies with North Korea, they are negative towards welfare-priority policies. Conversely, a majority of the 1990s generation, the children's generation, are negative towards reconciliation and cooperation policies with North Korea but somewhat positive towards welfare-priority policies. This finding is consistent with prior research stating that 'factors influencing the subjective perceptions of progressivism-conservatism differ between generations.' Middle-aged and older adults primarily define their ideology based on North Korea policy, while young adults define their ideology based on preferences for welfare versus growth. The 1970s and 1980s generations were influenced by both variables towards a progressive orientation, while for the 1960s and 1990s generations, the influence of these two variables may have canceled each other out. In this election, North Korea policy preferences significantly influenced voting choices, but policy preferences regarding growth versus welfare did not exert statistically significant influence.

[Figure 3] Presidential Job Performance Evaluation by Generation

[Figure 4] Preferences for North Korea Policy and Growth-Welfare Policy by Generation

In conclusion, the 1960s generation showed patterns nearly identical to the overall voter average in terms of subjective ideological perception and presidential job performance evaluation, which were the most prominent determinants of voting choices in this election. Regarding North Korea policy, a majority of the 1960s generation were slightly more favorable towards reconciliation and cooperation policies than the voter average, which was a notable characteristic. Ultimately, the unique generational identity of the 1960s generation did not influence their voting choices. This was true for other generations as well. For the 1970s generation, generational identity appears to have had a negative impact on their vote for Yoon Suk-yeol.

5. Pooled Data Analysis from 1992-2022

In the 2022 presidential election, the '86ers' chose candidates in a manner very similar to the average voter. The same applied to subjective ideological perception and presidential job performance evaluation, which were the most influential factors in voting choices. Their preference for North Korea policy was slightly more inclined towards reconciliation and cooperation than the voter average, but the difference was not significant. So, did the '86ers' become 'conservative' with age after being 'progressive' in the past? Or did they follow a different path?

Post-democratization elections have seen ideological fluctuations. Voting choices alone, especially in two-candidate versus multi-candidate races, differ. Therefore, it is difficult to definitively label voting as 'progressive' or 'conservative' based solely on the proportion of votes for a particular party. These fluctuations have been adjusted using two methods. First, the conservative advantage rate measures the extent to which the vote share of conservative party candidates was dominant. Second, the overall mean centering technique subtracts the average value for a specific generation from the average value for the election period. The graph applying this method is [Figure 5]. An additional method was incorporated: instead of comparing generational voting choices election by election, the voting choices of each generation across their life cycles were examined. The aim was to observe the voting choices of each generation at specific age points.

The red, bold line represents the voting pattern of the 1960s generation. Interpreting the graph with 0 as the center, points below indicate progressive voting, and points above indicate conservative voting. The most distinctive feature is the very gentle slope of the 1960s generation's voting choices. This slope is markedly gentler compared to the preceding generations, the 1950s and 1940s. This pattern differs from the aging effect, which suggests conservatism with age. In the 2017 presidential election, when the median member of the 1960s generation, born in 1965, was 53 years old, they voted slightly more conservatively than the average, but in other elections, they generally voted slightly more progressively than the average. Another distinctive point is that the voting behavior of the 1960s generation during their young and middle adult years was not particularly 'progressive.' While direct comparison is impossible due to the lack of empirical data for preceding generations during their young and middle adult years, comparison with subsequent generations is possible. When compared at the same age, the 'progressive' voting of the 1970s and 1980s generations is clearly more pronounced than that of the 1960s generation.

In summary, the 1960s generation exhibits distinct voting choice characteristics compared to previous generations. This is because the voting choices observed during their young and middle adult years have been maintained as they aged. Signs of rapid conservatism in their 40s and 50s have not yet been observed. In this regard, there is a clear difference from previous generations. At the same time, the 1960s generation did not exhibit 'progressive' voting behavior in their young and middle adult years as commonly believed. In fact, in only three of the eight presidential elections held since democratization did the 1960s generation support progressive candidates more than conservative ones: the 1997, 2002, and 2017 elections. In the remaining five elections, including the current one, they supported conservative candidates more. The three elections where this generation supported progressive candidates more were elections where the entire electorate shifted towards progressivism. Conversely, the five elections where this generation supported conservative candidates more were elections where the entire electorate shifted towards conservatism. These were elections where the so-called 'period effect' was at play. After correcting for this apparent effect, it was confirmed that the 1960s generation, even in their 20s and 30s, voted with a slight progressive tendency, close to the overall average for all generations. This trend did not change significantly thereafter.

[Figure 5] Generational Voting Across the Life Cycle[4]

Based on the above, let us analyze the pooled data from seven presidential election surveys conducted from 1992 to 2022. As confirmed in [Table 3], age, ideology, and region clearly act as factors supporting conservative candidates in South Korean presidential elections since democratization. The older the age, the more conservative the self-perception, and the more likely one is to be from the Yeongnam region, the higher the probability of supporting a conservative candidate. In [Table 3], our variables of interest are the generational dummy variables. The 1970s generation is less likely to support conservative candidates compared to other generations, while the 1950s and 1990s generations are more likely to support conservative candidates. The 1960s generation, the primary variable of interest in this study, was found to be statistically insignificant.

How should the statistically insignificant dummy variable coefficient for the 1960s birth cohort be interpreted? The value of 0 for the 1960s birth cohort dummy variable includes both earlier and later generations. Generally, those born before the 1960s cohort, i.e., the 1950s cohort, are likely to be more conservative, and those born after, i.e., the 1970s cohort, are likely to be more progressive. This analysis indeed confirmed this expectation. Ultimately, the remaining group corresponding to the value of 0 for the 1960s birth cohort dummy variable is not homogeneous and is likely to have its influences offset by those from both sides. Therefore, the 1960s cohort likely voted more progressively than the 1950s cohort and more conservatively than the 1970s cohort. This means that the 1960s cohort did not engage in the most progressive voting behavior as an icon of progressivism, as is often discussed.

[Table 3] Presidential Elections and Generational Effects

[Table 4] Presidential Elections and Generational Effects by Life Cycle Stage

[Table 4] compares the political characteristics of voting choices for each generation at the same life cycle stage, categorized by age group, using survey data from five-year intervals from 1992 to 2022. Models (1) through (6) address whether the dummy variables for each generation are significant for those under 40, i.e., in their 20s and 30s. Because the analysis grouped individuals of similar age, the aging effect was almost non-existent within these groups. Ideology, region, and period effects for each election were confirmed as expected.

The focus of this study is the effect of generational dummy variables. In the 20s-30s age group, no generation was statistically significant. Although the sign of the coefficients indicated that the 1950s and 1960s generations were more conservative, and the 1970s and 1980s generations were more progressive, these findings were not statistically significant. In other words, within the under-40 age group, no particular generation is significantly more progressive or conservative. In the 50s age group, there were also no statistically significant dummy variables. This can also be interpreted to mean that no specific generation is inherently progressive or conservative.

A dummy variable that was statistically significant only for those in their 40s was identified among those born in the 1970s. This can be interpreted to mean that individuals in their 40s born in the 1970s did not support conservative candidates more than other generations. However, this issue also requires careful interpretation. It could be interpreted that those born in the 1970s, rather than those born in the 1960s, are more progressive than other generations in their 40s, but this cannot be considered a 'generational effect.' If individuals born in the 1970s continue to make more progressive voting choices than other generations even after reaching their 50s, which is the turning point for conservatism, then there might be room for interpretation as a 'generational effect.' In other words, it is premature to approach and interpret the voting choices of those born in the 1970s, who have not yet fully entered their 50s, as a generational effect. This is in the same vein as pointing out the limitations of previous studies that approached the voting choices of those born in the 1960s as a generational effect about a decade ago.

An impressive finding in [Table 4] is that the age variable was statistically significant for all age groups in their 50s. Up to their 40s, it could not be concluded that voting choices became more conservative with increasing age within that group, but in their 50s, it can be interpreted that an age effect operates within the group. In other words, the conventional wisdom that people rapidly become more conservative upon entering their 50s is confirmed in this analysis as well.

Examining the changes in ideological perception reveals another interesting point. [Figure 6], like the voting choices, is a graph tracking generational ideological perceptions using the life-cycle effect and the centering technique. In the left graph, individuals born in the 1960s perceived themselves as progressive at a similar level to the preceding generation, those born in the 1970s, when they were in their 20s and 30s. However, as seen in [Figure 5], their actual voting choices differ significantly from those born in the 1970s. This can be interpreted to mean that although individuals born in the 1960s perceived themselves as progressive, their voting choices were not as progressive as their ideological perceptions.

[Figure 6] Subjective Ideological Perception by Life Cycle Stage

[Table 5] Past Presidential Elections and Ideological Perception

Indeed, the discrepancy between [Figure 5] and [Figure 6] is confirmed when examining the effect of generation dummy variables on ideological perception in [Table 5]. Individuals born in the 1960s, like those born in the 1970s, perceive themselves as more progressive than other generations. However, as confirmed in [Table 3] and [Table 4], the voting choices of those born in the 1960s were not more progressive than other generations. In other words, their voting choices were not as progressive as their self-perceived progressivism.

6. Major Findings and Conclusion

The findings from the analysis above are as follows. First, the tendency of voting choices related to age or generation in the 2022 presidential election has weakened compared to previous elections. Second, the relationship between age and ideological perception showed a U-shaped curve rather than the traditional linear structure. Up to the early 40s, individuals perceive themselves as more progressive with increasing age, and from around age 42 onwards, they perceive themselves as more conservative with increasing age. Third, individuals born in the 1960s have not made, nor are they making, more progressive voting choices than other generations, even during their youth or middle age. Fourth, although individuals born in the 1960s have evaluated themselves as progressive, their voting choices have not been as progressive as their ideological evaluations. Fifth, individuals born in the 1970s are actually making more progressive voting choices in their 40s than those born in the 1960s. However, since those born in the 1970s have not yet fully entered their 50s, the turning point for conservatism, it is difficult to interpret this as a 'generational effect.' Sixth, upon entering their 50s, all generations have become rapidly conservative.

In the 2022 presidential election, the generational identity of the '86 Generation' did not manifest in voting choices. The same applies to other generations. Voters did not vote based on generational characteristics. Ideological and policy preferences heterogeneity influenced voting choices, such as ideological perception, evaluation of the president's job performance, and North Korea policy. Heterogeneity in income, assets, home ownership, and class perception has not received sufficient attention. It is a reasonable interpretation that these variables were not expressed due to the constraints of the available political parties. Ultimately, a generation cannot be a monolithic bloc. This is why discussions in academia in recent years emphasizing the need to focus on intra-generational differences rather than inter-generational differences are gaining attention (Shin Jin-wook 2022).

At the beginning of this text, it was stated that the term '86 Generation' is controversial. Although the method of dividing generations by decade is widely criticized in academia, the political characteristics of the so-called '86 Generation' were tracked using this method. This 'generation' was not significantly different from any other 'generation.' It is now time to consider returning to the term 'those born in the 1960s' instead of '86 Generation.' At the very least, based on the empirical analysis of voting choices and ideological perceptions, the 'generational effect' that has been associated with the '86 Generation' can no longer be found. It is more accurate to say it was never there in the first place. ■

References

Kang, Won-taek. 2003. *Korean Electoral Politics: Ideology, Region, Generation, and Media*. Seoul: Purunghil.

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 Korean Voters 3: The 18th National Assembly Election Through Panel Data*. Seoul: East Asia Institute.

Noh, Hwan-hee, Song, Jeong-min, and Kang, Won-taek. 2013. “Generational Effects in Korean Elections: Focusing on Presidential Elections from 1997 to 2012.” *Korean Political Party Studies Review*, Vol. 12, No. 1: 113-140.

Moon, Woo-jin. 2016. “The Persistence and Change of Ideological Conflict in Korean Electoral Competition.” *Korean Political Party Studies Review*, Vol. 15, No. 3: 37-60.

Park, Myung-ho. 2009. “An Analysis of Generational and Age Effects in the 2008 General Election: Focusing on the 386 Generation.” *Korean Political Party Studies Review*, Vol. 8, No. 1: 65-86.

Park, Won-ho. 2013. “The Transformation of Generational Theory: The 18th Presidential Election and Generation.” In Park, Chan-wook, Kim, Ji-yoon, and Woo, Jeong-yeop (eds.), *The Choice of Korean Voters 2: The 18th Presidential Election* (Seoul: Asan Institute for Policy Studies): 201-247.

Bae, Jin-seok. 2017. “Generational Factors in Voting Choice and Ideological Orientation: An Analysis of Presidential Elections from 1992-2017.” In Kang, Won-taek (ed.), *Changing Korean Voters 6: Candlelight Protests, Impeachment Crisis, and the 19th Presidential Election*. East Asia Institute.

Seo, Hyun-jin. 2008. “A Study on Voter Turnout and Generation in the 17th Presidential Election.” *Journal of Parliamentary Research*, Vol. 14, No. 2: 117-142.

Shin, Jin-wook. 2022. *There Is No Such Generation: Generational and Political Stories in the Age of Inequality*. Gaemagowon.

Oh, Se-je. 2015. “A Study on the Characteristics of Generational Effects of the 386 Generation: Focusing on the Conditional Manifestation of Generational Effects.” *Journal of 21st Century Political Science*, Vol. 25, No. 1.

Oh, Se-je and Lee, Hyun-woo. 2014. “Conditional Generational Effects of the 386 Generation: Focusing on Ideological Orientation and Presidential Election Voting.” *Journal of Parliamentary Research*, Vol. 20, No. 1.

Lee, Nae-young. 2002. “Generation and Political Ideology.” *Quarterly Thought*, Autumn issue: 53-79.

Lee, Woo-jin. 2013. “Gender and Generation in Political Economy.” *Journal of Fiscal Studies*, Vol. 7, No. 4.

Jeong, Jin-min. 1992. “Generational Factors in Korean Elections.” *Korean Political Science Review*, Vol. 26, No. 1: 145-167.

Jeong, Jin-min. 1994. “Political Generations and the 14th National Assembly Election.” *Korean Political Science Review*, Vol. 28, No. 1: 257-274.

Jeong, Jin-min and Hwang, Ah-ran. 1999. “Korean Electoral Politics After Democratization: Focusing on Generational Factors.” *Korean Political Science Review*, Vol. 33, No. 2: 115-134.

Heo, Seok-jae. 2014. “Generation, Life Cycle, and Voting Choice: An Analysis of Presidential Elections from 1992-2012.” *Korea and International Politics*, Vol. 30, No. 2: 167-199.

Abramson, Paul. 1975. Generational Change in American Politics. Lexington: Lexington Books.

Alwin, Duane F. and Jon A. Krosnick. 1991. “Aging, Cohorts, and the Stability of Sociopolitical Orientations over the Life Span,” American Journal of Sociology, Vol. 97, No. 1: 169-195.

Barnes, S.H. 1989. “Partisanship and Electoral Behavior.” In M. K. Jennings and J.W. Van Deth, eds. Continuities in Political Action. Berlin: De Gruyter.

Cho, Jinman and Kihong Eom. 2012. “Generation Effects? An Empirical Analysis of the Korean National Assembly and Presidential Elections,” Asian Perspective, Vol. 36, No. 3: 353-386.

Feather, N. T. 1977. “Value Importance, Conservatism, and Age,” European Journal of Social Psychology, Vol. 7, No. 2: 241-245.

Mannheim, Karl. 1952. “The Problem of Generations,” Paul Kecskemeti (ed.), Essays on the Sociology of Knowledge (New York: Oxford University Press): 276-322.

Truett, K. R. 1993. “Age Differences in Conservatism,” Personality and Individual Differences, Vol. 14, No. 3: 405-411.

Van Hiel, Alain and Lieven Brebels. 2011. “Conservatism is Good for You: Cultural Conservatism Protects Self-esteem in Older Adults,” Personality and Individual Differences, Vol. 50, No. 1: 120-123.

[1] “In the 2012 presidential election, which was effectively a two-way race, candidate Moon Jae-in, who lost, received 48.2% of the vote, significantly ahead of candidate Kim Dae-jung, who won with 40.3% in the 1997 election, which had a three-way race. It is self-evident that a simple comparison between the vote share of the losing candidate Moon Jae-in and the winning candidate Kim Dae-jung cannot accurately capture voter behavior.” (Bae Jin-seok 2017, 172)

[2] “When replicating the statistical analysis of studies that simultaneously included age variables and dummy-format generation variables (Heo Seok-jae 2014; Oh Se-je 2015) in a similar manner, the variance inflation factor (VIF) for the generation variables of interest exceeded or approached 10, raising concerns about multicollinearity. Furthermore, when four or five detailed generation variables are included simultaneously in the regression equation, it is difficult to rule out the possibility that they might be interpreted as age effects rather than the intended generational effects.” (Bae Jin-seok 2017, 167-168)

[3] “For example, if the dummy variable for the pre-1960s generation results in supporting conservative candidates more, it can be easily interpreted that this generation exhibited more conservative voting behavior compared to those born in the 1960s or later. The same applies to generations born after the 1960s. The problem lies in the interpretation of the dummy variable for those born in the 1960s. The 0 value for the 1960s generation dummy variable includes both the preceding and succeeding generations. Generally, the preceding generation is likely more conservative than those born in the 1960s, and the succeeding generation is likely more progressive, making the remaining group heterogeneous. Therefore, the effect of the 1960s generation dummy variable may be offset by the influences of both sides. If the influence of the 1960s generation appears ambiguous, it can only be indirectly confirmed based on whether the influences of the preceding and succeeding generations align with the expected direction and are statistically significant.” (Bae Jin-seok 2017, 174)

[4] The 'conservative advantage rate' and 'centering the overall average' techniques were used.

■ Author: Bae Jin-seok_Assistant Professor, Department of Political Science and International Relations, Gyeongsang National University. He received his Ph.D. in Political Science from the University of Texas, Austin. His primary research areas include elections, political parties, and public opinion in the context of democratization and new democracies. As a Senior Research Fellow at the East Asia Institute (EAI), he was involved in the practical work during the founding of the Asian Democracy Network (ADN) and the Asian Democracy Research Network (ADRN) in 2013.

■ Editor: Jeon Ju-hyun_EAI Research Fellow

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