Characteristics and Voting Choices of Undecided and Mobile Voters

1. Introduction

The 20th presidential election concluded with the victory of candidate Yoon Suk-yeol of the People Power Party. It was the closest victory in terms of the margin of votes and the percentage difference in votes in the history of presidential elections. With the defeat of candidate Lee Jae-myung, the Democratic Party of Korea failed to retain power, and the Moon Jae-in administration became the first to lose power after a single term since democratization in 1987.

Furthermore, the 20th presidential election, much like the 19th, was characterized by stark differences and confrontations in the preferences and attitudes of supporters of the two major parties. While the 19th presidential election was centered around attitudes towards the impeachment of President Park Geun-hye and the candlelight vigils, the 20th presidential election was centered around attitudes towards the five-year Moon Jae-in administration (or an evaluation of regime change and the major policies of the Moon Jae-in government).

The reason for Moon Jae-in's easy victory in the 19th presidential election was not only due to voters' attitudes towards the overarching issue of President Park Geun-hye's impeachment, but also because the progressive camp had a (de facto) single candidate in Moon Jae-in, while the conservative camp was divided between Hong Joon-pyo and Yoo Seung-min, and the centrist candidate Ahn Cheol-soo further complicated the multi-candidate landscape. The 20th presidential election was a two-way race between Lee Jae-myung of the progressive camp and Yoon Suk-yeol of the conservative camp, but the outcome was difficult to predict due to the uncertain impact of Ahn Cheol-soo's vote share on the two main candidates.

About a month before the election, with neither candidate Lee Jae-myung nor Yoon Suk-yeol having a guaranteed victory, candidate unification emerged as the biggest issue. On February 13, 2022, candidate Ahn Cheol-soo officially proposed a unification with Yoon Suk-yeol based on public opinion polls. However, on February 28, 2022, it was reported that the negotiations had publicly failed (<JoongAng Ilbo> 2022/02/28). Yet, just a week later, in the early morning of March 3, 2022, the two candidates reached a surprise agreement on unification (<Hankyoreh> 2022/03/03). There were various predictions about the effect of the candidate unification, which occurred after much deliberation. Generally, speculations included that Ahn Cheol-soo's supporters would give up voting due to disappointment and backlash from the 'humiliating' and 'opaque' unification process (<JoongAng Ilbo> 2022/03/18), that they would shift their support to Yoon Suk-yeol, or that the Democratic Party's supporters, feeling a sense of crisis from the unification, would rally and increase their voter turnout.

Therefore, this study aims to: first, analyze whether Yoon Suk-yeol's victory was indeed due to the influence of candidate unification. While Yoon Suk-yeol's narrow victory was the result of a complex interplay of various variables, this study attempts to analyze the independent and direct effect of candidate unification. Second, this study further analyzes voters who changed their preferred candidate during the election campaign period (intra-election party changers or intra-election swing voters) using panel data analysis. This election was closely contested, and the results reflected this. It is hypothesized that the number of voters who changed their preferred candidate due to high voting fluidity was greater than in any previous election, and the results are expected to confirm this. Moreover, the unification of candidates Yoon Suk-yeol and Ahn Cheol-soo made it inevitable for Ahn Cheol-soo's supporters to 'forcefully' change their support. Thus, this study analyzes whether the voting choices of these 'mobile voters' were also a major variable that determined the election outcome. Third, this study analyzes undecided voters (floating voters or late deciders) who decided their preferred candidate 'during the election campaign period.' Undecided voters are always the primary target of election campaigns. Theoretically, it is assumed that undecided voters are not 'camp voters' based on party or political ideology, but 'issue publics' who make voting choices based on the central issues of the election, and their voting choices are analyzed.

To this end, this study utilizes panel data constructed from pre- and post-election surveys. The first (pre-election) survey was conducted from January 12-15, 2022, targeting 1,515 individuals selected through stratified sampling based on region, gender, and age, using random digit dialing (RDD) for telephone interviews with men and women aged 18 and above residing nationwide. The margin of error, assuming random sampling, is ±2.5 percentage points at a 95% confidence level. The second (post-election) survey was conducted from March 10-15, 2022, targeting 1,104 individuals from the first survey. The margin of error, assuming random sampling, is ±2.5 percentage points at a 95% confidence level.

2. Undecided Voters

Undecided voters are defined as those estimated to be hesitating and deliberating without deciding on a preferred candidate during the election campaign period. They are also referred to as late deciders or campaign deciders, distinguishing them from pre-campaign deciders who had already chosen their candidate before the campaign began.

However, research on undecided voters, a crucial voting bloc that determines election outcomes, is very limited. According to a study by Cho Sung-dae (2013), undecided voters are generally women, young people, and highly educated voters, as well as those who are unaffiliated with any party or have weak preferences for their chosen party. Other research indicates that undecided voters are generally young, unaffiliated voters, and they do not differ from non-undecided voters in terms of gender, education, income level, or political knowledge (Ryu Jae-seong 2012). Nevertheless, further research is needed on the size and demographic characteristics of undecided voters, as well as their political orientations and the variables that determine their choices.

Among the hypotheses regarding undecided voters, two appear valid. The first concerns party factors: the hypothesis that young voters without a party affiliation experience difficulty in choosing a candidate. However, these young, unaffiliated voters are not a group characterized by ignorance, indifference, or nonattitudes towards politics, unlike some unaffiliated voters in the United States; rather, they are inferred to be a group with negative attitudes towards parties (Ryu Jae-seong 2013). The second hypothesis, as argued by V.O. Key, Jr. (1966), is that undecided voters are 'issue publics' who decide their preferred candidate based on information about the candidates' personal backgrounds or policy proposals. Research indicates that undecided voters consider policy proposals or candidates' careers and capabilities as more important factors in their voting choices than the candidates' party affiliation (Ryu Jae-seong 2014). In this context, unaffiliated voters are hypothesized to be 'information seekers' who actively strive to obtain information about policies, proposals, or candidates during the election campaign period to decide on their preferred candidate.

Although various discussions about undecided voters occur in the media, they are often based on hypotheses rather than theories that have passed rigorous empirical verification. Therefore, this study aims to provide a more scientific approach to the discussion of undecided voters by analyzing empirical data on their size, characteristics, and voting choices.

① Size of Undecided Voters

In the 18th National Assembly election in 2012, 34.1% of respondents decided on their preferred candidate after the campaign began. In the 2012 presidential election, 19.5% of voters decided on their preferred candidate after the official election campaign started. In the 2014 local elections, undecided voters accounted for 54.6% (Ryu Jae-seong 2014). The proportion of undecided voters was highest in local elections, followed by general elections and then presidential elections.

The author defined respondents who decided their vote 'one week before their own vote' as undecided voters. This is a conservative estimate of the size of the undecided group. According to this criterion, 33.36% of respondents indicated that they decided on their preferred candidate after the official election campaign began in the 2022 presidential election. This size is larger than in the 2012 presidential election and similar to the results of the 2012 general election.

[Table 1] Timing of Voting Decision

f2e34dd1a320d063

f2e34dd1a320d063

f2e34dd1a320d063

Although there is no established theory regarding the size of the undecided voter group, several inferences can be made. First, the size of the undecided group will be determined by the effectiveness of the 'party-in-electorate,' meaning the degree and strength of a party's influence over voters. If party competition is stable and a party's influence over voters is sufficiently strong, voters will make voting choices based on their 'party affiliation,' resulting in a smaller undecided group that does not significantly impact election outcomes. Conversely, in the opposite scenario, the size of the undecided group will inevitably increase, and the electoral volatility due to their decisions will also rise. The size of the undecided group is proportional to the size of the unaffiliated group.

Second, the electoral competition structure. If an election is highly competitive and the winners cannot be easily predicted, voters will deliberate longer to choose their candidate. In this context, compared to presidential elections with competition between two strong major parties, local elections requiring a choice among multiple candidates demand more information. Furthermore, due to lower interest in the election, the number of undecided voters who postpone their decision will increase (Ryu Jae-seong 2014).

The significant size of the undecided voter group in the 20th presidential election indicates that the effectiveness of parties among voters was low and that the election was a close contest where predicting the winner was difficult. In this survey, 29.69% of respondents answered that they 'do not support any party,' which is comparable to the 30.35% for the Democratic Party and 30.63% for the People Power Party. Overall, one-third of voters responded that they 'do not support any party.' The existence of a substantial number of unaffiliated voters was confirmed. Of these unaffiliated respondents, 55.87% were undecided voters, while 19.88% of Democratic Party supporters and 20.0% of People Power Party supporters were undecided voters. As the 20th presidential election unfolded as a close race between candidates Yoon Suk-yeol and Lee Jae-myung, the number of voters who agonized over their candidate choice until the final stages of the election appears to have been higher than in any previous election.

② Characteristics of Undecided Voters

Undecided voters are generally women, young people, and highly educated voters, as well as those who are unaffiliated with any party or have weak preferences for their chosen party (Cho Sung-dae 2013). Other research identifies young people, the highly educated, unaffiliated voters, centrists, and those with low interest in elections as characteristics of undecided voters (Ryu Jae-seong 2014). The undecided voters in the 20th presidential election are shown in [Table 2]. Young people, centrists, unaffiliated voters, low political knowledge, low election interest, and weak partisan preference are identified as distinguishing characteristics of undecided voters. The results of a logistic regression analysis with these variables as independent variables and undecided voters as the dependent variable are shown in [Table 3].

According to the logistic regression analysis, young age, being unaffiliated with a party, low election interest, and weak partisan preference were statistically significant characteristics of undecided voters. Gender, socioeconomic status, political ideology, and knowledge were not statistically significant variables in determining undecidedness.

[Table 2] Proportion of Undecided Voters by Group

f2e34dd1a320d063

f2e34dd1a320d063

f2e34dd1a320d063

[Table 3] Logistic Regression Analysis Results for the Undecided

f2e34dd1a320d063

f2e34dd1a320d063

f2e34dd1a320d063

③ Voting Choices of the Floating Stratum

The floating stratum was analyzed to have voted for Lee Jae-myung at 54.55% and Yoon Suk-yeol at 45.45%. This means the floating stratum preferred candidate Lee Jae-myung by 9.1 percentage points more than candidate Yoon Suk-yeol. In contrast, the decided voters chose candidate Yoon Suk-yeol (54.41%) more than candidate Lee Jae-myung (45.59%), and this voting choice is statistically significant (F=7.21, p=0.0074).

Candidate Lee Jae-myung is presumed to have been relatively more successful in 'persuading' floating voters during the election campaign. On the other hand, candidate Yoon Suk-yeol appears to have been relatively more successful in 'mobilizing' his supporters before the election campaign began. Lee Jae-myung's 'persuasion of floating voters' and Yoon Suk-yeol's 'mobilization of supporters' seem to conflict with the candidates' respective election strategies at first glance. This will be discussed in detail later.

[Table 4] Voting Choices of the Floating Stratum

f2e34dd1a320d063

f2e34dd1a320d063

f2e34dd1a320d063

3. Mobile Voters

There exist voters who change their preferred candidate during the campaign period and move (intra-election party/candidate changers or intra-election swing voters). Due to a lack of empirical data, research on 'mobile voters' has been scarce. Research on voters who change their party/candidate preference between elections (inter-election party/candidate changers) or swing voters also remains a gap in election studies. In the 2022 presidential election, there were two categories of voters who changed or moved their preferred candidate during the campaign period. The first category includes voters who had a preferred candidate but changed it during the campaign period. The second category includes voters who moved from supporting Ahn Cheol-soo to supporting Yoon Suk-yeol or Lee Jae-myung (or other candidates).

① Distribution of Mobile Voters

In the 20th presidential election, voters who changed or moved their preferred candidate during the campaign period were found to be 28.75% of all voters. Voters who did not change their preferred candidate during the same period accounted for 71.25%. Among voters for Lee Jae-myung, 75.16% were persistent voters and 24.84% were mobile voters, while among voters for Yoon Suk-yeol, 69.77% were persistent voters and 30.23% were mobile voters.

[Table 5] Persistence and Mobility of Preferred Candidate Support

f2e34dd1a320d063

f2e34dd1a320d063

f2e34dd1a320d063

Of the 127 Ahn Cheol-soo supporters in the preliminary survey, 42 (33.1%) chose Lee Jae-myung and 77 (60.6%) chose Yoon Suk-yeol, respectively. As a result of the asymmetric division of Ahn Cheol-soo supporters due to candidate consolidation, it was analyzed that 8.48% (42/495) of Lee Jae-myung voters and 14.64% (77/526) of Yoon Suk-yeol voters were Ahn Cheol-soo supporters. Ultimately, the effect of the Yoon Suk-yeol-Ahn Cheol-soo candidate consolidation appears to have worked more favorably for candidate Yoon Suk-yeol. This will be discussed later.

② Characteristics of Mobile Voters

[Figure 1] shows the characteristics of mobile voters compared to sustained voters. While 32.63% of women were mobile voters, 22.47% of men were mobile voters, and this difference is statistically significant (F=13.32, p=0.0003). The proportion of mobile voters by age group was also statistically significant (F=73.61, p=0.0000); the proportion of mobile voters aged 18-29 was 54.79%, while the proportion of mobile voters aged 60 and over was only 16.77%. Subjective class perception was not a statistically significant variable distinguishing mobile voters from sustained voters (F=0.23, p=0.6337). While 34.26% of centrist voters were mobile voters, 22.01% of progressive voters and 23.92% of conservative voters were mobile voters, and this difference is statistically significant (F=15.25, p=0.0001). 50.68% of undecided voters were mobile voters, while 11.67% of Democratic Party supporters and 15.21% of People Power Party supporters were mobile voters, and this difference is statistically significant (F=126.90, p=0.0000).

[Figure 1] Characteristics of Mobile Voters (1)

According to [Figure 2], 33.8% of voters with very low political knowledge and 31.58% of voters with very high political knowledge were mobile voters, while 26.61% and 21.77% of voters with intermediate political knowledge were mobile voters, respectively (F=6.21, p=0.0129). Voters with low interest in the election showed a higher proportion of mobile voters compared to those with high interest (F=87.73, p=0.0000). Party favorability, defined as the difference in positive-negative emotional attitudes toward the Democratic Party and the People Power Party, showed that the smaller the difference in emotional attitudes between the two parties, the higher the proportion of mobile voters (F=128.0, p=0.0000).

[Figure 2] Characteristics of Mobile Voters (2)

In summary, mobile voters, compared to sustained voters, are characterized by being female, young, centrist, undecided, having low political knowledge, low election interest, and low party favorability difference. According to logistic regression analysis with these variables as independent variables and mobile voters-sustained voters as the dependent variable, female, young, undecided, low political knowledge, and low party favorability difference were statistically significant variables determining mobile voters.

[Table 6] Logistic Regression Analysis Results for Mobile Voters

f2e34dd1a320d063

f2e34dd1a320d063

f2e34dd1a320d063

③ Voting Choices of Mobile Voters

[Figure 3] shows the voting choices of mobile voters/persistent voters. Among voters aged 18-29, 32.19% are voters who previously supported other candidates but shifted their support to candidate Lee Jae-myung during the election campaign period. Furthermore, among voters aged 18-29, 22.6% are voters who previously supported other candidates but shifted their support to candidate Yoon Suk-yeol during the election campaign period.

Ultimately, 54.79% of voters aged 18-29 are voters who changed their candidate support during the election campaign period. The group where voters who shifted their support to Lee Jae-myung outnumbered those who shifted to Yoon Suk-yeol was solely the 18-29 age group. The groups where respondents who shifted their support to Yoon Suk-yeol outnumbered those who shifted to Lee Jae-myung were all groups except the 18-29 age group, including those in their 40s, 50s, and 60s, as well as male, female, unaffiliated, and moderate voters. It was found that candidate Yoon Suk-yeol received substantial support from voters who changed their candidate support during the election campaign period.

[Figure 4] illustrates the voting choices of mobile voters based on major election issues. Among respondents who agreed that 'The comprehensive real estate holding tax is excessive,' 67.29% continued to support candidate Yoon Suk-yeol, and 20.56% shifted their support to Yoon Suk-yeol after previously supporting other candidates. It was found that 87.85% of respondents who agreed that 'The comprehensive real estate holding tax is excessive' voted for candidate Yoon Suk-yeol. In contrast, 82.67% of respondents who disagreed that 'The comprehensive real estate holding tax is excessive' supported candidate Lee Jae-myung. In a similar vein, among respondents who agreed that 'This presidential election is a judgment on the Moon Jae-in administration,' 59.37% continued to support candidate Yoon Suk-yeol, and 18.2% were respondents who shifted their support to Yoon Suk-yeol. Among respondents who disagreed that the government's COVID-19 response was successful, 61.15% continued to support candidate Yoon Suk-yeol, and 20.65% were respondents who changed their support to Yoon Suk-yeol.

[Figure 3] Voting Choices of Mobile Voters/Persistent Voters (1)

[Figure 4] Voting Choices of Mobile Voters/Persistent Voters (2)

This confirms that the comprehensive real estate holding tax, judgment on the Moon Jae-in administration, and COVID-19 response were the primary variables determining candidate support, and it also confirms that support shifts to either candidate Yoon Suk-yeol or Lee Jae-myung occurred based on stances on these issues.

[Table 7] shows the issues that determined the candidate support of mobile voters. The reasons for shifting support to candidate Yoon Suk-yeol or the issues that determined support for Yoon Suk-yeol were, respectively, 'Failure of the Moon Jae-in administration's real estate policy' (50.94%) → 'Lee Jae-myung's moral controversy' (18.24%) → 'Unification of candidates Yoon Suk-yeol and Ahn Cheol-soo' (13.21%), and 'Unification of candidates Yoon Suk-yeol and Ahn Cheol-soo' (32.73%) → 'Controversy over the abolition of the Ministry of Gender Equality and Family' (16.36%) → 'Failure of the Moon Jae-in administration's real estate policy' (14.55%) and 'Controversy over Yoon Suk-yeol's wife's false resume and stock manipulation' (14.55%).

[Table 7] Issues Determining Candidate Support for Mobile Voters

f2e34dd1a320d063

f2e34dd1a320d063

f2e34dd1a320d063

Units: %, ( ) indicates n

4. Vote Choices of Undecided and Mobile Voters

[Table 8] shows the proportion of undecided and mobile voters among Yoon Suk-yeol and Lee Jae-myung voters. Among the 526 Yoon Suk-yeol voters, 30.23% were mobile voters, and among the 495 Lee Jae-myung voters, 24.85% were mobile voters. Among these, the proportion of mobile voters who were originally Ahn Cheol-soo supporters accounted for 14.64% of all Yoon Suk-yeol voters and 48.28% of mobile voters. The proportion of mobile voters who were Ahn Cheol-soo supporters accounted for 8.48% of all Lee Jae-myung voters and 30.77% of mobile voters. Among Yoon Suk-yeol voters, the proportion of mobile voters and mobile voters from Ahn Cheol-soo supporters was higher than that among Lee Jae-myung voters. Overall, candidate Yoon Suk-yeol received more choices from mobile voters, and the proportion of mobile voters from Ahn Cheol-soo supporters was also high. In contrast, among Lee Jae-myung voters, 36.36% were undecided voters, and among Yoon Suk-yeol voters, 28.52% were undecided voters, indicating a higher proportion of undecided voters among Lee Jae-myung voters.

[Table 8] Vote Choices of Undecided and Mobile Voters

f2e34dd1a320d063

f2e34dd1a320d063

f2e34dd1a320d063

[Table 9] presents the results of a logistic regression analysis with vote choice as the dependent variable (1=Yoon Suk-yeol, 0=Lee Jae-myung). In all models, age, undecided voters, mobile voters, and the two main issues (judging the Moon Jae-in administration and COVID-19 response) were statistically significant. In summary, the 20th presidential election was determined by age, major issues, undecided voters, and mobile voters (largely due to candidate consolidation effects). Generational division was confirmed as a statistically significant variable characterizing the 20th presidential election. The influence of issues was also confirmed.

Notably, the choice of undecided and mobile voters, rather than the centrist or unaffiliated, was a statistically significant variable in the Yoon Suk-yeol-Lee Jae-myung vote choice. Undecided voters statistically significantly chose candidate Lee Jae-myung, while voters who changed their preferred candidate, i.e., mobile voters, were more likely to vote for candidate Yoon Suk-yeol. This indicates that candidate Lee Jae-myung succeeded in persuading undecided voters during the campaign, while candidate Yoon Suk-yeol secured more choices from voters who switched their preferred candidate. This implies a significant influence from Ahn Cheol-soo supporters who shifted their vote to Yoon Suk-yeol after candidate consolidation.

[Table 10] shows the change in the dependent variable in response to changes in the independent variables of Model 3. Women were 4.54 percentage points more likely to vote for candidate Lee Jae-myung than men, and individuals aged 18-19 were 18.52 percentage points more likely to vote for candidate Lee Jae-myung compared to voters aged 60 and above. Voters who 'completely' agreed that the election was a referendum on the Moon Jae-in administration were a significant 70.01 percentage points more likely to vote for candidate Yoon Suk-yeol compared to those who 'completely' disagreed. Agreement on the success of the COVID-19 response also resulted in a 59.01 percentage point difference in voting probability between the two candidates.

[Table 9] Logistic Regression Results for Vote Choice

f2e34dd1a320d063

f2e34dd1a320d063

f2e34dd1a320d063

*** p<0.001, ** p<0.01, * p<0.05

[Table 10] Changes in Voting Choice based on Logistic Regression Analysis

f2e34dd1a320d063

f2e34dd1a320d063

f2e34dd1a320d063

Compared to decided voters, undecided voters were 19.74%p more likely to vote for candidate Lee Jae-myung than for candidate Yoon Suk-yeol, and mobile voters were 21.0%p more likely to vote for candidate Yoon Suk-yeol than for candidate Lee Jae-myung. While persuading undecided voters was successful for candidate Lee Jae-myung, it can be inferred that candidate Yoon Suk-yeol benefited from absorbing Ahn Cheol-soo's supporters or from the candidate consolidation effect.

Table 11 shows the reasons for voting choice among voters for Ahn Cheol-soo who voted for Yoon Suk-yeol or Lee Jae-myung. For Yoon Suk-yeol voters, the reasons for their voting choice were 'Other' (27.27%), 'Candidate's Party Affiliation' (22.08%), and 'Candidate's Likelihood of Winning' (14.29%). For Lee Jae-myung voters, the reasons were 'Candidate's Competence and Experience' (35.71%), 'Candidate's Pledges' (21.43%), and 'Other' (21.43%). Lee Jae-myung voters centered their decision on the 'candidate,' while Yoon Suk-yeol voters centered their decision on 'party' and 'election victory.'

According to Table 12, the political favoritism ratings for Yoon Suk-yeol voters were in the order of Ahn Cheol-soo (7.04) > Yoon Suk-yeol (6.24) > Moon Jae-in (3.22) > Lee Jae-myung (2.49). For Lee Jae-myung voters, the order was Lee Jae-myung (5.38) > Moon Jae-in (5.26) > Ahn Cheol-soo (4.52) > Yoon Suk-yeol (2.21). While the most favored politician for Yoon Suk-yeol voters was 'Ahn Cheol-soo,' it is natural that they voted for 'Yoon Suk-yeol,' who had the next highest preference, due to the candidate consolidation. For Lee Jae-myung voters, since 'Lee Jae-myung' had the highest preference, voting for Lee Jae-myung is also a natural transfer of votes (however, it is difficult to explain why Ahn Cheol-soo's preference among Ahn Cheol-soo supporters was ranked third, lower than Lee Jae-myung or Moon Jae-in).

Table 11 (Post-election Survey) Reasons for Voting Choice of Lee Jae-myung and Yoon Suk-yeol Voters (%)

f2e34dd1a320d063

f2e34dd1a320d063

f2e34dd1a320d063

[Table 12] (Post-election Survey) Lee Jae-myung-Yoon Suk-yeol Voters' Favorability of Politicians (Mean Values)

f2e34dd1a320d063

f2e34dd1a320d063

f2e34dd1a320d063

5. Summary and Conclusion

This study analyzed the size, characteristics, and voting choices of undecided and mobile voters. Undecided and mobile voters shared common characteristics such as being younger, unaffiliated with any party, having low favorability differences towards political parties, and low interest in elections. However, their voting choices differed. Specifically, undecided voters were more likely to vote for candidate Lee Jae-myung, while mobile voters were more likely to vote for candidate Yoon Suk-yeol. The voters who switched their support to Yoon Suk-yeol primarily came from those who had previously supported Ahn Cheol-soo, suggesting that Yoon Suk-yeol's victory in the presidential election would have been impossible without candidate consolidation.

Due to the forced and artificial nature of the Yoon Suk-yeol-Ahn Cheol-soo candidate consolidation, which led to changes and shifts in support among Ahn Cheol-soo's voters, the analysis of mobile voters in this report is conflated with mobile voters resulting from other candidate consolidations and other 'natural,' 'voluntary' mobile voters. Furthermore, the scale of mobile voters was artificially inflated, meaning the number of mobile voters in this study is likely overestimated. Nevertheless, it is clear that a significant number of mobile voters existed due to candidate consolidation in this election, and their greater shift towards candidate Yoon Suk-yeol played a decisive role in his victory. This study explained why mobile voters, particularly those among Ahn Cheol-soo's supporters, chose Yoon Suk-yeol more frequently, through an analysis of reasons for candidate choice and differences in politician favorability.

Conversely, undecided voters largely chose candidate Lee Jae-myung, but a satisfactory explanation for this could not be found. The higher proportion of undecided voters among Lee Jae-myung's voters compared to Yoon Suk-yeol's voters suggests that Lee Jae-myung's campaign was highly effective in the final stages of the election. It can be inferred that undecided voters were mobilized to support Lee Jae-myung during the campaign period and were successfully brought to the polls. While this appears to be the result of Lee Jae-myung conducting a more effective campaign targeting undecided voters, more precise empirical analysis is needed.

The significance of this study lies in its 'discovery' of the 'fact' that the voting choices of undecided and mobile voters differ from those of decided and persistent voters. Specifically, undecided voters were more likely to vote for candidate Lee Jae-myung than decided voters, and mobile voters were more likely to vote for candidate Yoon Suk-yeol than persistent voters. However, there is no existing theory that can interpret or explain this discovered fact. This represents a serious gap in the research. Although undecided and mobile voters share many common characteristics, their differing voting choices also require interpretation and explanation, yet theories to enable this are absent. A hypothetical approach is to understand the voting choices of undecided and mobile voters as a result of campaign effects. However, existing theories on campaign effects explain the 'reinforcement' of existing attitudes and preferences, but not the 'persuasion' or 'change' of attitudes and preferences. New theorization appears necessary to explain the voting choices of undecided and mobile voters. ■

References

Kim Mi-na. 2022. “Yoon Suk-yeol-Ahn Cheol-soo Candidate Consolidation Agreement Reached... Press Conference at 8 AM”. Hankyoreh.https://www.hani.co.kr/arti/politics/politics_general/1033290.html (Accessed: 2022. 05. 10.).

Ryu, Jae-seong. 2012. “Who Are the Undecided? A Study on the Factors Influencing the Decision to Support a Candidate.” In Park, Chan-wook & Kang, Won-taek, eds., Analysis of the 2012 General Election. Paju, Gyeonggi Province: Nanam.

Ryu, Jae-seong. 2013. “Characteristics of Centrist and Unaffiliated Voters: Non-attitudes or Negativity?” Korean Political Science Review 20(1): 101-127.

Ryu, Jae-seong. 2014. “Who Are the Undecided? Comparative Analysis of the 2012 General Election and Presidential Election, and the 2014 Local Elections.” Journal of Peace Studies 22(2): 113-144.

Park, Tae-in & Sung, Ji-won. 2022. “Yoon Suk-yeol and Ahn Cheol-soo’s Failed Unification Process.” JoongAng Ilbo. https://www.joongang.co.kr/article/25051562#home (Accessed: 2022.05.10.).Cho, Sung-dae. 2013. “A Study on Undecided Voters.” Korean Political Science Review 47(3): 109-129.

Choi, Jun-won. 2022. “Humiliating Opposition Unification Process: Ahn’s Supporters Couldn’t Vote for Yoon.” JoongAng Ilbo. https://www.joongang.co.kr/article/25056390#home (Accessed: 2022.05. 10.).

Converse, Philip E. 1962. “Information Flow and the Stability of Partisan Attitudes.” Public Opinion Quarterly 26(4): 578-599.

Converse, Philip E. 1966 “Information Flow and the Stability of Partisan Attitudes.” In Angus Campbell et al., eds., Elections and the Political Order. New York: John Wiley.

Key, V. O. Jr. 1966. The Responsible Electorate. Cambridge, Mass.: Belknap Press.

■ Author: Ryu, Jae-seong_Current Vice President of the Korean Political Science Association, Former President of the Korean Association of Party Studies, Professor at the Department of International and Area Studies, Keimyung University, Ph.D. in Political Science from the University of Texas at Austin. His research areas include parties and elections, American politics, and political psychology. Recent publications include "An Analytical Essay on Downs' P: 'Prospect of Success' and 'Memory of Failure'" (2021), "Types and Motivations of Split Voting: Analysis of the 20th General Election" (2020), "Voter Types and Characteristics: Party Voters, Cross-Pressured Voters, Intermittent Non-voters, Habitual Non-voters" (2020), and "How Does Framing Activate or Suppress Ideological Propensity?" (2019).

■ Editor: Jeon, Ju-hyun _EAI Researcher

    Inquiries: 02 2277 1683 (ext. 204) | jhjun@eai.or.kr