The dynamics of gender gap across age groups has not yet been fully examined. Most of the studies covering gender gaps in political participation were conducted at the aggregate level. The finding that men have a higher turnout rate than women at the aggregate level is regarded as supporting the traditional gender gap. If such gender differences disappear or women have a higher turnout rate at the aggregate level, then it is regarded as evidence for the modern gender gap.

However, this does not mean that we see the same size of gender gap in all age groups. To construct theoretical patterns of gender gap in voter turnout, we should first consider the aging effect in voter turnout. The older the voters are, the more likely they are to adhere to voting as a social norm. This is a phenomenon that is observed in most democracies (Goerres, 2007).

Turnout does not increase linearly with age in most countries, however. As they reach a certain old age which varies across societies, voters tend to disengage themselves from passive forms of political participation including voting (Jennings & Markus 1988). This creates the curvilinear form of voter turnout rates over ages. Many empirical studies show lower rates of participation in elections among widows and widowers than among married people (Wilensky, 1961; Wolfinger & Wolfinger, 2008). However, the effect of the transition to widowhood has not been statistically significant in individual-level analyses (Hobbs, Christakis, & Fowler, 2013; Stoker & Jennings, 1995).

Women are greatly influenced by the widowhood effect, which significantly reduces their motivation for voting. In general, married couples are more likely to participate in voting. Spouses influence each other on many topics and spend some time discussing political issues. This may motivate political interest and participation (Beck, 1991). The primary mobilizer (or persuader) disappears when the spouse dies. Considering that the overall life expectancy of men is shorter than that of women, women are more likely to leave their spouses than men of the same age group. The expectation that the turnout rate of women is lower in old age groups is derived from Beck’s study. In addition, socioeconomic status also deteriorates in widowhood, which has a negative impact on voting participation; it has been empirically verified that voter participation is positively associated with socioeconomic status (Bartels, 2008). In sum, the two theoretical forms of a gender gap in voter turnout can be expressed as in Figure 1.

Figure 1. 
Two theoretical gender gaps in voter turnout by age group.

To robustly check the gender gap in Korea, this paper selected three different kinds of elections: the 2007 presidential election, the 2010 local elections, and the 2012 legislative elections. At the aggregate level, we see the traditional gender gap in voter turnout rates: Men voted more than women in these three elections.

However, we see a different gender gap pattern when I disaggregate voters into specific age groups. The Republic of Korea National Election Commission (NEC) provides valuable information about turnout among young voters. The commission estimated voter turnout rates with a systematic random sampling of about 10% of the total voters, which is projected to be a close reflection of the general population.¹ Indeed, the estimated voter turnout rates are very close to the actual turnout rates. I found the traditional gender gap in the youngest age groups (19 and 20–24) as shown in Table 3. The gender gap reverses in the late twenties group, and it continues at least into the late thirties group, which supports the modern gender gap seen in the U.S. and Canada. Then the gender gap reverses again in older age groups, which supports the traditional gender gap.

Recent empirical findings point out the existence of gender disparity in voter turnout even in established democracies. According to Kostelka et al. (2018), while the gender gap in voter turnout tends to disappear or reverse in more important (first-order) elections, it is still observed in less important (second-order) elections (p. 5). However, the gender gap dynamic pattern in voter turnout across age groups has been consistently observed in recent Korean elections regardless of the election type. This implies that the unique gender gap dynamics may be generated by a structural effect in Korea rather than by temporal effects.

The traditional gender gap in voter turnout was observed in the latest legislative elections (male 58.8% vs. female 57.4%) at the aggregate level. To diagnose fundamental changes in attitudes toward politics by gender and age group, this paper examined two measures of psychological involvement in politics (interest in politics and the elections, and political knowledge) as well as political predisposition (political ideology) (Blais, 2000; Brady, Verba, & Schlozman, 1995; Burns, 2007; Kostelka et al., 2018; Larcinese, 2007). These factors were expected to be associated with the reverse gender gap in voter turnout. For this, I checked the recent survey data of the 2016 legislative election.²

Interest in politics and the elections. The “traditional gender gap” is evidently observed at the aggregate level. However, when disaggregated we find the gap differs across age groups. Figure 2 shows the percentage of respondents who answered that they were interested in politics and the elections. The interest in a specific election was higher than the interest in general politics. Except for those in the early 30s, the percentage of male respondents who answered that they were interested in the legislative elections was higher in all age groups than that of female respondents. The differing interest in elections by gender is evident among those in age groups beyond their late thirties. In particular, this difference by gender reaches a striking point (about 20 percentage points) among those in their forties and above. In other words, the proportion of voters who are interested in politics steadily increases for men, while for women the growth is sluggish.

Figure 2. 
Interest in politics and the election by gender and age group.

Political ideology. Female respondents are more conservative than male respondents at the aggregate level, which supports past studies (Lee, 1985, 1995). Table 4 which shows the distribution of self-placed ideologies by gender illustrates that females score an average of 5.4 while males score 5.1 on the liberal-conservative continuum ranging from 0 (most liberal) to 10 (most conservative). The proportion of female respondents who evaluate themselves as liberal (or progressive) is 26.5%, which is significantly less than that of male respondents, 35.4%. The ratio of female respondents who position themselves as conservative is also slightly higher than that of male counterparts (36.7% vs. 34.9%). In sum, I infer that Korean female voters tend to be more conservative than male counterparts at the aggregate level.

However, the difference in gender gap varies across age groups. Figure 3 shows the average scores of self-placed ideology by gender and age group.³ In the late thirties, the gender gap in self-evaluated ideology may be reversed. Based on this data, among voters aged younger than 30, the proportion of women who evaluate themselves as liberals is higher than that of men. In contrast, the gender gap is reversed among those aged in their forties and older. At the aggregate level, female voters appear to be more conservative than male voters, but when examined by age, dynamics in the gender gap are evident. In particular, it is notable that the female group (30–34) showing greater interest in the elections than their male counterparts is the most liberal across all age groups, which supports the modern gender gap.

Figure 3. 
Self-placed ideology by gender and age group.

Political knowledge. I used five questions (open- and closed-ended questions) to measure political knowledge. In all the questions, I observed that the male respondents are more politically sophisticated than the female counterparts. However, gender differences are not observed in easy questions that address issues such as the term of legislators, the second-largest party in the current legislature, and elected candidates in their legislative districts. In contrast, in harder questions that address topics such as the total number of legislative seats and the name of the current National Assembly Speaker, there is a clear difference in accuracy according to gender (see Table 5).

The differing gender gap could be seen across age groups. Figure 4 shows correctly answered choices for hard questions according to gender. It is difficult to find a consistent pattern in gender gaps. However, we observe small gender gaps in some age groups. In particular, the age group of 19 shows a reverse gap by gender on the question about the name of the KNA Speaker. This shows that the modern gender gap can be recognized in the youngest group.

Figure 4. 
Political knowledge by gender and age group.

I suggest two feasible explanations about the dynamic gender gap across age groups (Koo et al., 2015). The first explanation is based on political socialization and the social and economic burdens mainly imposed on Korean male voters. The second explanation is based on the modern gender gap introduced by new generations.

Explanation 1: Political socialization lag. Korean women, not obliged to serve in the military, tend to experience political socialization earlier than men in the same age group. Women are employed earlier than men of the same age, but they experience sexual discrimination in promotion. In particular, women who are concerned about their career disruption after childbirth feel the necessity for welfare policies such as parental leave policies, and thus recognize the importance of politics earlier than their male counterparts. They are also more politically sophisticated because they have greater opportunities to discuss politics in the workplace. This earlier political socialization of women motivates them to engage in politics.

The explanation becomes different when the focus is shifted to male voters. The reason the turnout rate of men in their late 20s to 30s is lower than that of women is essentially due to the fact that the lives of men in this age group are not stable compared to women. In a society where the democratic system is settled to a certain extent, stable life is a prerequisite for active political participation. Voters will be less inclined to actively participate in politics when of a pessimistic frame of mind, believing they cannot escape the difficulties of everyday life or when their environment shows no sign of improvement. In sum, earlier political socialization of women along with the burdens imposed on men delay the political socialization of men in Korea. This induces the reverse gender gap in some age groups.

In terms of political participation, it is important to consider whether young voters have stable jobs and also think that they are likely to raise a family. Men in their late twenties to thirties feel more pressured to get a job and a place to live than do their female counterparts, which may not induce a higher male voter turnout rate.

College education has been expanded in Korea, compared to other advanced countries. In 2012, 65% of 25- to 34-year-olds in Korea had been the recipients of a college education, compared to an OECD average of only 38%. In fact, Korea ranked first among all 37 OECD countries in this respect (2012). This higher college enrollment rate leads to an overpopulation of the college graduate workforce, resulting in an increase in the unemployment rate of college graduates. Compared to other OECD countries in 2012, the employment rate of B type tertiary education graduates was 74.9%, lower than the OECD average of 80.5% and the lowest level in the OECD (except for Japan at 73.6%). The employment rate of type A (tertiary education institution) graduates, which is equivalent to four-year colleges and graduate school graduates, is 78.1%, the lowest of all OECD countries (average 84.2%).

The relatively low employment rates of these highly educated young people generally leads to over-education in labor markets. Getting employed in places not related to their majors and academic backgrounds leads them to consider changing jobs for the better. In particular, since the mid-2000s, the increase in the flexibility of the labor market and the reduction in the numbers of new jobs with large corporations have resulted in a steady decline in so-called “good jobs,” and generated increasing numbers of temporary workers. Even if they are employed, they are in a state of constant instability in that they continue to leave their workplaces or lose their jobs. The overall wage level has continued to deteriorate. In 1997, the wage of those in their late 20s was 86% of the average age wage, but it dropped to 78% by 2010 (Korean Employment Information Service [KEIS], 2012, p. 15). The deterioration of the employment environment is a major cause of psychological instability in most young eligible voters.

The current work employment for the young is in itself a very pessimistic phenomenon regardless of gender.⁴ However, young men may feel disadvantaged, which is evidently revealed in several issues related to the mandatory military service (e.g., advantage points on entrance exams for government jobs to male applicants who have fulfilled their mandatory military duty, which was ruled unconstitutional in 1999). When a man has to complete his military service, then his preparation for employment will be delayed. In the case of college graduates, they are in a situation where they should secure a job at least before they reach 30. According to the data on the unemployed in their twenties and thirties, we find a high ratio of unemployed male graduates preparing for a job.

On the other hand, for highly educated, unmarried women, unemployment trends are much lower compared to men. This can be explained by the fact that in the case of men, they need to consider the long-term responsibility they have to take for their family, and thus the search process itself becomes longer in order for them to obtain better-paid, or better-promising jobs under the patriarchal culture (KEIS 2012, p. 115). In addition, they tend to increase their wage levels by switching jobs rather than staying at one job (KEIS 2012, p. 53). On the other hand, women shoulder a relatively low burden in trying to obtain a stable job, and are more likely to leave the labor market if they are married regardless of their employment. Men in their 20s and 30s who are struggling to find a stable job are less likely to be interested in politics than their female counterparts.

Korean young males also bear a heavy burden of marriage-related problems. Marriage basically starts with the arrangement of the residence where the newlyweds are to live. Most of them prefer places close to work. However, most of the workplaces favored by young married people tend to be concentrated in the Seoul metropolitan area. They therefore hope to settle in Seoul itself or Seoul commuter areas. Besides the high housing prices, marriage itself is perceived as a serious burden for men as they are under the patriarchal legacy that housing is the man’s responsibility in Korea. This reality is shown in a recent survey on marriage: 46.3% of male respondents in their twenties and thirties would be willing to forego marriage, which is much lower in comparison to their female counterparts (33.7%). This overall situation may be a contributing reason for the relatively low interest or participation in politics shared among men in the younger age groups. In sum, the pressure of securing a well-paying job and also the responsibility that comes with marriage under the patriarchal Korean culture are delaying the political socialization of young men. This is considered as the key cause of lower turnout among young male voters.

Explanation 2: Generation, temporal factors, and individual-level factors. Besides the economic conditions unfavorable to young male participation in voting on which Explanation 1 mainly focuses, I consider other potential effects: new generation, temporal factors, and individual-level factors.

The factors affecting turnout can be divided into structural factors, temporal factors, and individual-level factors. The structural factors refer to the overall political system, including the electoral system that may affect the voters’ incentive to participate in voting (Karp & Banducci, 1999). Temporal factors include the degree of polarization of voters (Hetherington, 2001) and the significance of elections such as an election resulting in partisan realignment (Glaser, 1962). On the other hand, comparative literature generally regards age as the most powerful variable explaining turnout at the individual level, along with education (Blais, 2000); many empirical studies have shown that the level of education is positively associated with voter turnout (Lazarsfeld, Berelson, & Gaudet, 1944; Lewis-Beck, Jacoby, Norpoth, & Weisberg, 2008; Verba, Schlozman, & Brady, 1995). As the level of education increases, the stability and income of the profession also increases and the level of political knowledge becomes relatively higher. Therefore, the more educated the voters are more likely to turn out. Here I consider the three potential effects: generation effect, aging effect, and year-specific effect.

Generation effect. A generation effect is “a change which characterizes populations born at a particular point in time, but which is independent of the process of aging” (Blanchard, Bunker, & Wachs, 1977, p. 137). The range of change in the gender gap in voter turnout may result from the generational effect. If a new generation enters the eligible voter pool, the gender gap in voter turnout, which is measured by subtracting female turnout rates from male turnout rates, will decrease. The generation effect may be combined with the year-specific effect. For instance, Korean voters had the first competitive female presidential candidate in the presidential election of 2012, which may have boosted female voters’ participation. Young voters were angry with the president’s influence-peddling scandal and authoritarian leadership, and it may have strongly motivated them to vote in the latest presidential election.

Aging effect. According to Blanchard et al. (1977), an aging effect is “a change in variable values which occurs among all cohorts independently of time period, as each cohort grows older” (p. 137).⁵ As shown in the previous section, past studies on the relationship between age and turnout found that turnout graphs assume a curvilinear form across ages. The turnout rate is relatively low for the younger eligible age groups but increases gradually until it passes the middle age group where it begins to decrease slightly (Milbrath, 1965; Wolfinger & Rosenstone, 1980). Once voters have settled with stable jobs, male voters, as they get older, will become more interested in politics and more actively engaged in elections than women voters. This will make the difference in turnout rates between male and female voters increase.

Election-Specific effect. Voters are more interested in presidential elections than legislative or local elections. According to Kostelka et al. (2018), presidential elections belong to the first-order election type. In general, presidential candidates are widely covered in news media and public opinion polls. Political events such as televised debates can also increase voters’ interest in presidential elections (Lee & Lee, 2017). This can be seen from lower voter turnout rates in legislative or second-order elections such as local elections. If this is the case, in presidential elections the gender gap in voter turnout may be less subject to change by other factors.

To analyze the gender gap in Korea, I selected the Seoul Metropolitan City consisting of 25 administrative districts. I conducted a district level analysis for a number of reasons. First, individual-level analyses of voter turnout based on the post-election survey data are contaminated by respondents’ tendency to over-report driven by their social desirability bias (Anderson & Silver, 1986; Clausen, 1968; Katosh & Traugott, 1981). In particular, in elections when the turnout rate was high, nonvoters tend to respond as if they had voted (Karp & Brockington, 2005). Second, survey data quality on voter turnout often deteriorates due to unrepresentative samples. In particular, most Korean surveys have failed to provide representative samples of young voters in terms of voter turnout as they tend to overweigh the politically opinionated young voters. For instance, a survey conducted by a reputable polling company shows that the turnout rate of male voters in the late twenties is the highest, whereas in reality it is actually the lowest (Koo et al., 2015). The difference in the turnout rates of male voters between the individual-level sample and the aggregate-level sample is greater than fifty percentage points, which makes us no longer trust the representativeness of the former sample. Thus, despite there being a risk of an ecological inference fallacy, instead of using the individual-level data, I used the district-level data which were obtained using a more rigorous sampling method.⁶

Figure 5 shows the gender gaps in voter turnout by age group between 2007 and 2017. The y-axis indicates the gender gap calculated by subtracting the female turnout rate from the male turnout rate in each district. The solid line illustrates the average of the gender gaps across age groups in each election. In comparison to the youngest age group (19-year-old group), it increases in the early twenties group (20–24). It tends to decrease in the late twenties group (25–29) and increase gradually to the sixties group. After then it declines significantly. Although the ranges and locations of changes in gender gap vary across elections, the average of the gender gaps shows a consistent pattern. The reverse gender gap which is located below zero (horizontal dashed line) can at least be observed in the late twenties group. Thus, I infer that this dynamic pattern is caused by structural factors which are different from those in other democracies such as the United States and Canada.

Figure 5. 
Gender gap in voter turnout by age group (Seoul, District Level).

According to Explanation 1, women’s relatively early political socialization along with the economic burdens arising from the marriage responsibilities placed on men delay the political socialization of men—political socialization lag—in Korea. From this, this paper derived a hypothesis that if young male voters were less burdened (e.g., stable jobs and housing), then political socialization would not be delayed. Such voters are more likely to be interested in politics and participate in elections. As a result, the modern gender gap would not be observed.

To confirm the association between economic burdens and young male voters’ turnout, this paper selected the two richest districts (Kangnam and Seocho) out of 25 districts and then compared the voter turnout rates in those two districts with the rest. Whether in terms of the Gross Regional Domestic Product (GRDP) estimated by the structure of industry and workers or average income, there is no doubt that Kangnam is the richest district in Seoul and Seocho is the second richest.

Figure 5 shows different patterns for the gender gap in voter turnout in the richest districts. The partial reverse gender gap shown in other districts does not appear in the richest district (thick dotted lines). With the exception of the 2017 presidential election, we cannot find a reverse gender gap among younger groups. The second-richest district (thick dashed lines), also shows a similar pattern. Though there may be slight occurrences of the reverse gender gap, it is a temporary one in Seocho district. In sum, the modern gender gap is not clearly observed in the richest districts.

This result does not fit well with existing knowledge. In many comparative studies of development, economic development is a proxy for modernization. For instance, economic development induces democratization or consolidation of democracy according to the modernization hypothesis (Huntington, 1991; Lipset, 1959). Paradoxically, we observe the strongest traditional gender gap in voter turnout in the most modernized areas of Korean society.

We have clearly observed that young female voter turnout exceeds male voter turnout in the same age group. It is clear that the range of changes in the gender gap has decreased over time for presidential elections (upper panel of Figure 5). While the range was more than twenty percentage points in the presidential election of 2007, it shrank to five percentage points ten years later. The theoretical modern gender gap is observed in all ages before the age of 60 in the latest presidential election (upper right). There is a difference in degree, but I find a similar trend in legislative elections (lower panel of Figure 5), meaning the gender gap in voter turnout has decreased over time in legislative elections.

It is worthwhile to note that due to absentee votes resulting from mandatory military service, male voter turnouts at the early twenties group (20–24) have exceeded female voter turnouts across all elections before 2016. As mentioned above, young men tend to complete their military service in their early twenties, and almost all of them register for absentee voting in the military. This has explained the traditional gender gap in voter turnout for this specific age group. However, this is the first time a reverse gender gap can be seen even in this age group. This implies that a fundamental change has occurred in the new cohort.

This new generation is often called the Sewol Generation. After experiencing the Sewol ferry disaster in 2014, young South Koreans began to show greater interest in politics.⁷ The tragic accident affected young people even more than other age groups, and shaped the identity of young voters, particularly those aged 19 and 20, during the 2017 election. They tend to actively express their opinions. More importantly, young female voters from the generation show highly progressive attitudes toward political and social issues, compared to the older female voters.⁸

Indeed, the two explanations (Explanation 1 and Explanation 2) do not conflict with each other but rather supplement each other. The extended reverse gender gap, which starts from the youngest voter group and continues to at least the late forties group, can be explained by the entry of a new generation as well as delayed political socialization of male voters.

However, it is notable that certain aspects of the reduced gender gap in voter turnout in the latest presidential election can be explained by the election-specific effect. As the overall voter turnout rate increases beyond a certain point, the gender gap stops increasing. In general, voters are more interested in presidential elections than legislative elections, as can be seen from the lower voter turnout rates in legislative elections. For instance, the voter turnout rate for the 2017 presidential election was 77.2% while that for the legislative election held one year before was 58.0%. We should not overlook the outcomes of different kinds of election in that gender gaps in voter turnout can be easily affected by fundamental conditions underlying the legislative or local elections.

There are other possibilities that may affect the gender gap as can be seen in the latest presidential election. First, young and rich voters can be conservative and prefer a small government and a liberalized economy. These young, conservative voters may be reluctant to participate in voting because of a feeling of aversion to reactionaries who lead the main conservative party. If this is the case, we can expect that young voters who are rich and conservative may abstain from voting regardless of gender, which contrasts with the behavior of older voters (over sixty) who are willing to participate in elections. Indeed, this is revealed in the voter turnout outcomes in the richest district.

Second, we should consider the lowered cost of voting in recent years. With the utilization of IT technology, South Korea has attempted to incorporate early voting (or advance voting) in order to increase voter turnout. Voters can conveniently vote in advance at a nearby polling place a week before election day. Since it is so convenient, the cost of voting for voters is low. For instance, 26.06% of the eligible voters participated in early voting in the latest presidential election. In particular, the early voting participation ratios for the actual voters aged 19 and between 20 and 24 were 35.5% and 35.7%, respectively, and the ratio decreased with age. The percentage (21.8%, electorate composition ratio 15.9%) was the highest for those in the early twenties.