6.2 Local Elections: An Upset in the Seoul Metropolitan Area, Was It Hidden Votes or Changed Votes?

EAI OPINION Review No. 201006-02

Jointly by EAI, SBS, JoongAng Ilbo, and Hankook Research

Based on the Analysis of Panel Survey Data (Waves 1-3) for the 5th Local Elections in Seoul and Gyeonggi Province

The Choices of 'Pro-MB/Government Judgment Proponents' and 'Anti-Opposition/Anti-MB Government Judgment Proponents'

The 6.2 Local Elections: Failure in Prediction, Was It Hidden Votes or Changed Votes?

Throughout the 6.2 local elections, the political circles and media consistently focused on whether the elections would serve as a judgment on the incumbent government, as seen in past local elections, or as a platform to bolster the government's post-election administrative operations through a victory for the ruling party. The ruling party, citing its nearly 50% approval rating and leading party support, confidently predicted a victory for the government and ruling party that would revitalize the economy and national security, deviating from the historical pattern of local elections serving as mid-term judgments. The opposition, on their part, anticipated a repeat of the government judgment scenario. The outcome saw the ruling party win only in the Seoul Metropolitan Area (two regions) and Daegu/Gyeongbuk and Busan, areas where they had strong bases, while the opposition secured victories in all other regions, creating an unexpected outcome (?). The vote count revealed that opposition candidates won easily in Gangwon and Incheon, which were initially predicted to be close races, and the races in Seoul and Gyeonggi were narrow, with a 0.2%P difference in Seoul and a 4.4%P difference in Gyeonggi, where landslide victories were expected.

With the actual vote count showing significant divergence from the poll forecasts conducted a week before the election, arguments attempting to explain the shocking election results have been disseminated through political and media channels, invoking unfamiliar hypotheses such as the 'Bradley effect,' the '10% hidden vote theory,' the 'Minerva effect,' and the 'spiral of silence,' none of which have been empirically verified in Korean elections. While the endeavor to explain unexpected phenomena through new hypotheses and theories is natural and necessary, negating existing hypotheses and entire survey methodologies, or disseminating unverified hypotheses as if they were proven results, can lead to further confusion.

These hypotheses fundamentally attribute the failure of election predictions to the existence of 'hidden votes'—a significant portion of voters who chose opposition candidates but were unable to express their preferences or deliberately concealed them. The difference lies in the emphasis: some focus on the limitations of survey methods that failed to capture these hidden votes (theory of survey limitations), while others emphasize the political and social factors that prevented voters from accurately expressing or being able to express their true intentions (Minerva effect—freedom of expression, Bradley effect—social norms, spiral of silence theory—suppression of minority opinions).

However, this paper argues that the upset in the recent local elections was not due to a 'hidden vote effect' that failed to identify or concealed existing opposition candidate supporters, but rather a result of 'vote preference changes' where undecided voters, who were conflicted about their choice, decided to support opposition candidates in the week leading up to the election. [Figure 1] illustrates the trend of changing support rates based on panel surveys in the Seoul Metropolitan Area, which garnered the most attention for the electoral upset, revealing several significant characteristics. Between the first survey (May 10-13) and the second survey (May 24-26), unlike the general telephone surveys conducted concurrently which showed a 15-20%P gap in support rates for these candidates, the panel survey indicated a much smaller gap in support rates in Seoul and Gyeonggi. In the first survey, the gap between the first and second place candidates was 10.6%P in Seoul and 8.9%P in Gyeonggi. The second survey showed a gap of 11.0%P in Seoul and 7.8%P in Gyeonggi. These support gaps were within a range that could be narrowed in the week before the election, during which no opinion polls were conducted. Furthermore, the results of the third survey, conducted immediately after the election (June 3-5), confirm a surge in support for candidates Han Myeong-sook and Yoo Si-min, indicating a close contest between the ruling and opposition candidates in the final stages of the election.

General one-off surveys, which recruit different respondents for each survey, cannot verify whether observed changes in results between two time points are due to actual shifts in public opinion or variations in the characteristics of the respondents recruited each time. However, panel surveys, which conduct repeated surveys on a representative sample of respondents, provide empirical evidence that differences in survey results at each time point reflect actual changes in respondents' attitudes.

[Figure 1] Changes in Support for Candidates and Voting Intentions in the 6.2 Local Elections (Seoul/Gyeonggi Province)

f2e34dd1a320d063

f2e34dd1a320d063

f2e34dd1a320d063

This raises the question: who, and why, caused the shift in voter sentiment in Seoul and Gyeonggi, regions expected to be stable with high national approval ratings for the government, excluding the TK region, and with opposition candidates trailing in terms of candidate competitiveness? This paper aims to empirically explain the process and causes of the vote preference changes that led to the upset in the Seoul Metropolitan Area, utilizing panel survey data from 801 individuals who participated in both the second and third survey waves out of the 1,200 individuals recruited in Seoul and Gyeonggi from May 10 to June 5 by EAI, SBS, JoongAng Ilbo, and Hankook Research.

Why Was a President with a 50% Approval Rating Judged?

Who Were the Main Actors in the Seoul Metropolitan Area Voting Upset? 'Pro-MB/Government Judgment Proponents' and 'Anti-Opposition/Anti-MB Government Judgment Proponents'

Why did the situation, where President MB's approval rating was rising and Grand National Party candidates were leading in the Seoul Metropolitan Area until a week before the election, not continue until election day? The author's core argument is that this outcome resulted from a public opinion surge of protest and balancing among ambivalent voters—those holding both favorable and unfavorable views toward the government and opposition parties—in response to the government and ruling party's sharp shift towards an aggressive campaign strategy.

Voters with such contradictory tendencies require more conflict and time in candidate selection compared to consistent voters, classifying them as undecided or fluid. In typical elections, these voters often abstain due to unresolved internal conflicts or their effects tend to cancel each other out, thus not significantly influencing election outcomes. However, in this election, the government and ruling party's campaign strategy significantly stimulated latent sentiments of restraint among these voters, leading to a concentration of support for opposition candidates in the final stages of the election, which in turn resulted in the upset in the Seoul Metropolitan Area.

[Table 2] Distribution of Voter Types by MB Approval and Agreement with Government Judgment Theory: Seoul/Gyeonggi, 794 Respondents

*Note: Based on the second survey data from 794 respondents in Seoul/Gyeonggi who participated in waves 1-3.

One in Five Voters in the Seoul Metropolitan Area Simultaneously Holds 'Pro-MB Tendencies' and 'Government Judgment Sentiments'

Who were these ambivalent voters in this election, and in what numbers? Above all, their ambivalent and conflicting attitudes toward the President and the government are evident. Analysis of the 794 panelists in the Seoul/Gyeonggi region in [Table 2] shows that 52.9% evaluated the government's performance positively, while 47.1% responded negatively. In the same survey, 65.1% agreed with the statement 'The government's failures under President MB should be judged in the local elections,' while 34.9% disagreed, indicating a considerable spread of agreement with the government judgment theory.

Cross-tabulating voter attitudes on these two axes allows for a classification of public perception of the MB administration into four broad types: 'Pro-MB Stability Proponents' who support the President's performance and oppose government judgment; 'Pro-MB Judgment Proponents' who support the President's performance but agree with government judgment; 'Anti-MB Judgment Proponents' who are critical of the President's performance and advocate for government judgment; and 'Anti-MB Stability Proponents' who are critical of the President's performance and also oppose government judgment. Based on the 794 panelists in the Seoul Metropolitan Area who participated in all three survey waves, the consistent poles representing extreme views on MB's performance and government judgment are 'Pro-MB Stability Proponents' (250 individuals) and 'Anti-MB Judgment Proponents' (347 individuals), comprising 31.5% and 43.7% of all voters, respectively.

Conversely, among the other two types classified as ambivalent, 'Anti-MB Stability Proponents,' who are critical of the President's performance but also oppose government judgment, numbered only 27 individuals, or 3.4% of all respondents, and thus did not significantly influence the election outcome. However, 'Pro-MB Judgment Proponents,' who agree with government judgment while supporting the President's performance, numbered 170 individuals, accounting for 21.4% of all respondents. In other words, one in five voters in the Seoul Metropolitan Area exhibited ambivalent attitudes, supporting MB while simultaneously acknowledging the need for judgment in the election. Given that the proportion of 'Pro-MB Stability Proponents' and 'Anti-MB Judgment Proponents' differed by only 12.2%P, this suggests that the overall election landscape could have shifted depending on whether these voters leaned towards judgment or stability.

'Anti-Opposition' Anti-MB Judgment Proponents Are Also Ambivalent Voters, Constituting 17.5% of All Voters

Among the Anti-MB Judgment Proponents, a considerable number of voters are critical of the President's performance but are withholding their support for the current opposition parties due to disappointment. Poll results from January to April indicate that while 55-65% of those who approved of President MB's performance consistently supported the Grand National Party, the approval rating for the Democratic Party among voters critical of MB's performance remained at only 25-39%. Even when combining the support for the Democratic Labor Party, Creative Korea Party, and the Progressive New Party, the total barely reached 50%. This means that half of the voters with anti-MB tendencies were reluctant to support the existing opposition camp, demonstrating the presence of a significant number of voters with ambivalent attitudes toward opposition parties.

Specifically, to analyze the size and attitudes of 'Anti-Opposition Anti-MB Judgment Proponents,' we examined the party support of 374 'Anti-MB Judgment Proponents.' Respondents were categorized into those supporting parties participating in the anti-Grand National Party coalition (Democratic Party, Democratic Labor Party, Creative Korea Party, Participation Party) and those who were unaffiliated (no party support) or supported parties not part of this coalition (Grand National Party, Liberty Forward Party, Progressive New Party, etc.). Among the 374 'Anti-MB Judgment Proponents,' 208 individuals (59.9%) supported parties in the anti-Grand National Party coalition, while the remaining 139 individuals (40.1%) were either unaffiliated or supported parties with different orientations, such as the Grand National Party or Liberty Forward Party. These individuals constitute 17.5% of the total voter base (794 individuals). Combined with the 21.4% of Pro-MB Judgment Proponents, this group of 38.9% represented voters with ambivalent attitudes toward both the government and the opposition, a significant enough number to act as a variable in the final stages of the election.

Conversely, among 'Pro-MB Stability Proponents,' those not supporting the Grand National Party can be considered ambivalent. However, an analysis of the 250 'Pro-MB Stability Proponents' in the Seoul Metropolitan Area reveals a high degree of cohesion in their support for the Grand National Party. Of these, 200 individuals (80.0%) expressed support for the Grand National Party. Only 50 individuals (20.0%), representing 6.3% of all respondents, supported the Pro-MB Stability stance without supporting the Grand National Party.

Ultimately, among the types of ambivalent voters, the focus of analysis should be on the 'Pro-MB Judgment Proponents' and the 'group of Anti-MB Judgment Proponents critical of existing opposition parties,' as these are the groups that could have influenced the election outcome in the final stages. It is now time to examine their voting choices.

Voting Choices of Ambivalent Voters

D-7: Pro-MB Judgment Proponents, Initially Leaning Towards the Grand National Party, Achieved Balance in the Actual Vote

According to the second panel survey results from [Table 3] and [Table 4], conducted one week before the election, 'Pro-MB Stability Proponents' showed high support for Oh Se-hoon (83.6%) and Kim Moon-soo (81.0%). Conversely, 'Anti-MB Judgment Proponents' showed relatively lower support for Han Myeong-sook (64.2%) and Yoo Si-min (66.7%). Furthermore, at this time, ambivalent 'Pro-MB Judgment Proponents' were leaning towards Oh Se-hoon (46.6%) and Kim Moon-soo (60.6%). Support for Han Myeong-sook was only 19.2%, and for Yoo Si-min, it was 18.2%. Although public sentiment for anti-MB judgment was stronger than for pro-MB stability overall, the Grand National Party candidates were leading in the polls by drawing in 'Pro-MB Stability Proponents' and 'Pro-MB Judgment Proponents' to support the ruling party.

However, the third panel survey, conducted immediately after the election, reveals a drastic change among 'Pro-MB Judgment Proponents' and 'Anti-MB Judgment Proponents.' In Seoul, support for Han Myeong-sook among 'Pro-MB Judgment Proponents' rose from 19.2% to 39.4%. While Oh Se-hoon's support in this group reached 47.9%, surpassing Han Myeong-sook's, the gap had narrowed considerably compared to the second survey. Among 'Anti-MB Judgment Proponents,' support for Han Myeong-sook also increased by 7.4%P to 71.6% compared to the second survey, significantly widening the support gap with Oh Se-hoon within this group.

A similar pattern was observed in the Gyeonggi region. Among 'Pro-MB Stability Proponents,' Kim Moon-soo maintained 81.0% support in the second survey and 82.1% in the third. In contrast, only 66.7% of 'Anti-MB Judgment Proponents' supported Yoo Si-min in the second survey. However, by the third survey, the proportion supporting Yoo Si-min surged to 81.8%, increasing the gap with Kim Moon-soo. Among 'Pro-MB Judgment Proponents,' Kim Moon-soo's support remained stagnant, while Yoo Si-min's support increased by approximately 10%P from 18.2% to 28.6%, narrowing the support gap with Kim Moon-soo. This indicates that the rapid narrowing of the support gap between ruling and opposition candidates in Seoul and Gyeonggi was driven by the consolidation and increase in support for Han Myeong-sook and Yoo Si-min among 'Pro-MB Judgment Proponents' and 'Anti-MB Judgment Proponents.'

[Table 3] Changes in Supported Candidates by Voter Type in Seoul

[Table 4] Changes in Supported Candidates by Voter Type in Gyeonggi

Strategic Choices of 'Anti-Opposition' Anti-MB Judgment Proponents

We compared the voting choices of another group of ambivalent voters, the 'Anti-Opposition Anti-MB Judgment Proponents,' with those of 'Anti-MB Judgment Proponents who supported opposition parties.' As hypothesized, the latter group should have shown a greater tendency to choose opposition candidates in the final stages of the election compared to the former.

We selected 346 'Anti-MB Judgment Proponents' in Seoul and Gyeonggi and grouped them based on their support for opposition parties participating in the anti-Grand National Party coalition. We then examined changes in their supported candidates between the second and third survey waves. As expected, among 'Anti-MB Judgment Proponents' who supported opposition parties in the anti-Grand National Party coalition, the proportion supporting Grand National Party candidates Oh Se-hoon/Kim Moon-soo remained unchanged at 4.5% in the second survey and 4.8% in the third. Support for Han Myeong-sook/Yoo Si-min increased by only 7.0%P from 78.6% in the second survey to 85.6% in the third. This indicates that a significant consolidation of support for opposition candidates had already occurred by the second survey, resulting in minimal shifts in voter sentiment by election day.

In contrast, among the 'Anti-Opposition Anti-MB Judgment Proponents,' only 45.5% supported Han Myeong-sook/Yoo Si-min in the second survey, falling short of a majority. These voters held anti-government sentiments against the MB administration but simultaneously harbored distrust towards the existing opposition parties. In the second survey, support for Oh Se-hoon/Kim Moon-soo was 21.3%. However, in the third survey conducted a week later, support for Oh Se-hoon/Kim Moon-soo decreased by 8.3%P to 13.0%, while support for unified opposition candidates, including Han Myeong-sook/Yoo Si-min, increased significantly by 16.8%P to 62.3%. This is interpreted as the result of ambivalent voters, particularly those hesitant to support existing opposition parties, strategically voting for specific opposition candidates.

[Table 5] Changes in Supported Candidates by Attitude Towards Opposition Parties Among Anti-MB Judgment Proponents in the Seoul Metropolitan Area (Seoul/Gyeonggi)

The More Ambivalent the Voters, the Closer to Election Day They Decide Their Vote

If the upset in Seoul and Gyeonggi was not due to hidden votes but rather to ambivalent voters who hesitated in their decision-making finally leaning towards opposition candidates in the final stages of the election, then their decision-making period should have been around the week before the election. Examining the timing of voter decisions by respondent type indeed supports this hypothesis.

In [Table 6], among 'Pro-MB Stability Proponents,' 48.2% had decided on their candidate a month before the election, and an additional 22.8% decided during the candidate registration period two weeks prior, totaling a remarkable 71.1% who had already chosen their candidate. However, 'Anti-MB Judgment Proponents,' including a significant number of anti-opposition voters within this group, often made their candidate decisions closer to the election. Responses indicating a decision made within a week before the election were 24.8%, 22.0% decided 2-3 days prior, and 6.3% decided on election day, with over half (52.1%) making their decision within the final week. For ambivalent 'Pro-MB Judgment Proponents,' 18.4% decided a week before the election, 23.7% decided 2-3 days prior, and 10.3% decided on election day, with a majority also deciding within a week before the election.

Specifically, among 'Anti-MB Judgment Proponents' who were ambivalent and did not support any particular opposition party, 23.2% decided their candidate a week before the election, 27.7% decided 2-3 days prior, and 12.9% decided on election day. Approximately 64% of this group made their decision within a week before the election. This result indicates that their voting preferences were not hidden but were actively mobilized towards supporting opposition candidates in the week leading up to the election (see [Table 7]). [Table 6] Timing of Candidate Decision by Voter Type

[Table 7] Timing of Candidate Decision by Support for Opposition Parties Among Anti-MB Judgment Proponents

Conclusion: Causes of Late-Stage Mobilization and Political Implications of Election Results

The voter sentiment shifts observed in the final stages of the election offer important political implications for both the ruling and opposition parties. The ruling party needs to deeply reflect on why ambivalent voters, who held both positive and negative views towards the government and ruling party, ultimately leaned in a direction unfavorable to them in the final election period. Attention should be paid to the changes in the Grand National Party's campaign strategy from two weeks before the election, after the primary elections were completed, until just before the vote. In the early and mid-stages of the campaign, the Grand National Party's strategy was one of restraint, emphasizing the difficulty of the race and expressing caution about the prominence of the 'government judgment theory.' However, around the time when Yoo Si-min, a pro-Roh candidate, was decided as the unified opposition candidate in Gyeonggi Province, and buoyed by news of President MB's approval rating surpassing 50% in internal Blue House surveys, the campaign shifted to a more aggressive stance in the mid-to-late stages, characterized by a three-pronged attack against the opposition.

This included countering the 'MB government judgment theory' with a 'former government judgment theory,' and confronting the opposition's 'free school meals' policy with an 'ideological and political judgment of the Korean Teachers and Education Workers Union.' The government and ruling party, which had maintained a cautious and measured approach, emphasizing 'avoiding premature judgments,' launched an 'security offensive' after the 'Cheonan incident announcement' on the 20th. These changes, perhaps unintentionally, acted as a catalyst for centrist and ambivalent voters to engage in a balancing act, pushing them towards the opposing side.

When asked whether the government and ruling party had political intentions behind the Cheonan incident announcement, a staggering 90.7% of 'Anti-MB Judgment Proponents' responded affirmatively, and 66.8% of 'Pro-MB Judgment Proponents' agreed. Only 31.8% of 'Pro-MB Stability Proponents' agreed, showing a lower rate. Regarding the dismissal of teachers affiliated with the Korean Confederation of Trade Unions (KCTU) and the Korean Teachers and Education Workers Union (KTU), 88.8% of 'Anti-MB Judgment Proponents' viewed it as undesirable, and a majority (50.5%) of 'Pro-MB Judgment Proponents' also disagreed. Only among 'Pro-MB Stability Proponents' was the opposition to this measure lower, at 31.7%. Meanwhile, the 'pro-Roh judgment theory' was met with widespread disagreement across all groups, particularly among 'Pro-MB Stability Proponents,' with a significant 65.2% expressing negative views.

Ultimately, the offensive strategies employed by the Grand National Party in the mid-to-late stages of the election appear to have provoked antipathy and a sense of restraint among groups beyond their core supporters. The strategy of pressuring the 'ideological opposition' by invoking the previous administration for governance failures, the KTU for educational issues, and North Korea for security matters, while holding power in the executive, legislative, and local branches, amplified concerns about the government and ruling party's excessive conservatism and unilateral governance.

Meanwhile, the opposition parties should not rest on their laurels but should also reflect on the message conveyed by voter sentiment in this election. The analysis indicates that the Democratic Party likely would have faced considerable challenges in these local elections had the Grand National Party not made strategic errors in the latter half of the campaign, or had ambivalent voters not engaged in a balancing act against the ruling party's offensive strategy in the final decision-making phase. Most importantly, the Democratic Party failed to present policy alternatives and appeals that could independently secure voter support throughout the election campaign. Consequently, it was confirmed that even core supporters like 'Anti-MB Judgment Proponents' hesitated to support the Democratic Party until the final stages of the election. Therefore, rather than focusing on diagnostic assessments that indicate a lack of captured opposition-leaning voter support in pre-election polls, the opposition parties need to seriously consider why voters with anti-MB sentiments were withholding their support for the current opposition parties. It is hoped that both ruling and opposition parties will avoid getting lost in the 'hidden vote' debate and miss the core message that voters intended to convey to the political establishment and society. It is a time when both parties need the wisdom to look at the moon towards which the voters' fingers are pointing, rather than just the fingers themselves.■

[Figure 3] Voter Type Evaluation of Grand National Party's Major Campaign Strategies: Seoul/Gyeonggi, Third Survey Wave