Loading [Contrib]/a11y/accessibility-menu.js
Huang, K.-P. (2023). Still Aquamarine: China Factor and the 2020 Election Revisited. Asian Journal for Public Opinion Research, 11(2), 77–106. https://doi.org/10.15206/ajpor.2023.11.2.77
Download all (4)
  • Figure 1. Comparisons of Party Identifiers’ Stances on Traditional Cleavages
  • Figure 2. Comparison of Vote Share in Presidential Elections and Party List Elections
  • Figure 3. Comparisons of Facebook Posts and Comments
  • Figure 4. Comparisons of Key Terms under the Policy and Criticism Theme (Comments)


The DPP’s victory over the KMT in Taiwan’s 2020 elections has been interpreted as a triumph for anti-China sentiment. However, the rise of political outsiders and their influence on voting behavior in this election were overlooked and underestimated. In this article, we examined different sources of data and found that supporters of these political outsiders mentioned sovereignty and cross-Strait issues less than the incumbent Tsai Ing-wen. However, when faced with the choice between Tsai and challenger Han Kuo-yu, voters who were concerned about governance chose Tsai, contributing to her winning a record number of votes. This article suggests that economic and governance issues had a considerable role in the election’s result and will probably be the main focus of the 2024 presidential election. With the potential for a conflict in the Taiwan Strait increasing, anti-China sentiment is unlikely to be the deciding factor this time around.

The 2020 presidential election in Taiwan saw President Tsai Ing-wen successfully re-elected with a record number of votes. In her press conference following the confirmation of her victory, Tsai called the election result a victory for democracy and Taiwan’s sovereignty and claimed that it clearly showed that Taiwan’s citizens rejected China’s claims over the island (Chen, 2020). The international press also interpreted the results as a rejection of Chinese aggression against the island (BBC News, 2020; Chang & Jennings, 2020; Tiezzi, 2020). Several studies of the election concluded that the China factor was the determinant of the election results, and Taiwan’s democracy was resilient enough to withstand the interference of China in the election (Fell, 2021; C.-L. Liu, 2020; Templeman, 2020). In short, Tsai’s victory demonstrated Taiwanese solidarity against China.

In this article, we argue that the strategic choices made by third-party supporters and independents also contributed to the election result. For these voters, leader capability was more important than unificationindependence stances. In an analysis of different sources of data, we demonstrate how the winner-take-all effect, the incomplete observation effect, and the mentality of path dependency led to the conclusion that the election result was a victory for anti-China sentiment. Based on the findings, it would be inaccurate to assume that Taiwan voters would continue to vote DPP because of its anti-China stances aided by Sino-US confrontation. No matter how divided the opinions between the blues and greens are concerning cross-Strait relations, Taiwan’s future will still be determined by the majority who prefer the status quo.

The 2020 Presidential Election


The run-up to Taiwan’s presidential and legislative elections on January 11, 2020, was marked by several notable developments. First, the Kuomintang (KMT) nominated Han Kuo-yu, who had recently been elected mayor of Kaohsiung, to run for the presidency. Second, former Foxconn chairman Terry Gou joined the KMT primary election, which he lost to Han. Third, Taipei Mayor Ko Wen-je, once thought to be likely to run in the presidential election, instead chose to establish the Taiwan People’s Party (TPP) to participate in legislative competitions. These developments made the 2020 elections unprecedented; except for Tsai, who represented the ruling Democratic Progressive Party (DPP), the other three potential contenders were all atypical political figures. Gou and Ko did not build their careers as politicians and were considered political outsiders. Even though Han had previously served as a legislator, he had disappeared from politics for nearly twenty years. He was a marginal figure in the KMT — a maverick to use Barr’s terminology — and thus to a certain extent also a political outsider.[1]

The rise of Han in the 2018 local elections has been understood as the result of anti-establishment sentiment among Taiwan voters. Big data analysis has revealed that many Internet users expressed displeasure toward Taiwan’s two mainstream parties, represented by the colors blue (KMT) and green (DPP) respectively, and “dislike of both blue and green” was a widespread sentiment during the election (L.-W. Liu, 2022). When Han claimed that Kaohsiung, a southern port city long ruled by the DPP, was “old and poor” and vowed to bring jobs back to the city so young people who had left for better job opportunities could rejoin their parents, the message struck a chord with not only Kaohsiung voters but also people around the island. The resulting “Han wave” was so overwhelming that the KMT won most of the city and county mayoral posts in the 2018 local elections. Subsequently, three atypical political figures, Han, along with Ko and Gou, became potential rivals to Tsai ahead of the 2020 presidential election.

The Rise of Political Outsiders in Taiwan’s 2020 Election

The rise of these atypical political figures or political outsiders was closely associated with anti-establishment sentiment in the context of popular discontent about political gridlocks, poor governance, and policymaking under both the DPP and KMT (Lin & Hsiao, 2021). These developments in Taiwan have a mirror image in Eastern Europe, where anti-establishment parties won elections due to popular dissatisfaction with economic management and widespread corruption (Pop-Eleches, 2010). Anti-establishment parties and political outsiders enjoy the advantage of not having the baggage of the past, and their support comes from voters who are dissatisfied with mainstream parties and vote to protest against the status quo (Hanley & Sikk, 2016; Sikk, 2012). Although Taiwan is rated as one of the most liberal countries in East Asia by Freedom House, the support of ordinary people for democracy was less than 50% between 2005 and 2018, and satisfaction with the way democracy works hovered around 60% over the same period (Chu et al., 2020, p. 172). Research has found that the low level of democratic support is associated with affective polarization between DPP and KMT supporters, although independents’ antipathy toward the two mainstream parties led to stronger support of democracy (Hsiao & Yu, 2020). Given these findings, anti-establishment sentiment may not be harmful to democracy in Taiwan since it reflects demands for better democratic governance: a functional government without self-interest-seeking policymakers (Hibbing & Theiss-Morse, 2002). In this context, the 2020 election could have been a critical election in which the anti-establishment atmosphere allowed a political outsider to win the presidency. Instead, following Tsai’s eventual victory, the DPP’s anti-China stance was interpreted as the main factor determining the outcome.

How Did China Help Tsai to Win the Re-Election?

The DPP suffered poor local election results in 2018, including the unexpected victory of Han in the Kaohsiung mayoral election on a populist and anti-elite platform. The DPP and Tsai’s fortunes, however, began to shift in early 2019 when Xi Jinping described the future of Taiwan under the “one country, two systems” framework. In June of the same year, millions of people in Hong Kong took to the streets to protest Beijing’s tightening grip over the special administrative region. The situation in Hong Kong made Taiwanese people especially wary of China. Although China has always been a factor influencing Taiwan’s presidential elections, in the context of these developments, as well as the ongoing trade war between the United States and China, the defeat of the KMT (perceived as a pro-China party) has been interpreted in the global atmosphere of anti-China sentiment (BBC News, 2020). According to these interpretations, although most of Han’s supporters were attracted by his populist rhetoricHan claimed to represent the common people who were neglected by political elitesthis anti-elite sentiment was overshadowed by the China factor (Fell, 2021). Han’s populist appeal was also rejected by the majority of voters as the DPP successfully painted him as a “phony populist” (Batto, 2020). However, such interpretations overlooked and downplayed the forces driving the anti-establishment sentiment, which inadvertently also contributed to Tsai’s victory.

Arguments and Hypotheses

The article argues that three factors contribute to the common interpretation of the 2020 election results. The first is the winner-take-all effect. Parties and candidates campaign on issues where they have advantages over their opponents and, through this issue ownership, seek the votes that determine the election outcome (Budge, 2015; Budge & Farlie, 1983). Issue ownership has been found to affect election results in several Western countries though issue priming, issue salience, and the interaction between issue ownership and party ideology also play a role in the causal chain (Bélanger & Meguid, 2008; Petrocik, 1996; van der Brug, 2004). In Taiwan, the most salient issue separating parties and their supporters is national identity and the unificationindependence issue (Fell, 2005; Hsieh & Niou, 2005). The DPP’s 2020 campaign slogan is “Taiwan to win in 2020” and the KMT’s is “Safe Taiwan, rich people.” As both sides implied, this victory was significant for cross-Strait relations; the DPP’s victory was a confirmation of Taiwan’s sovereignty and the KMT’s win would have shown support for peaceful engagement with China. In the end, the campaign issues of the winning side are likely to be viewed as the issues that voters bought into, determining the electoral outcome. Such a perception is also likely to be reinforced when exogenous events make certain issues salient or primed. Issues closely tied to party ideology are also reinforced—the electoral outcome is not only an endorsement of the winning party’s campaign messages but also its political orientation. However, election results are aggregate outcomes, and there are many factors that can affect the results. We may see the winner’s polling rate improve over time, but we do not know whether it is due to the success of the winner’s campaign issues or the failure of the loser’s campaign messages to convince undecided voters, especially if they campaign on different issues. Our first step will be to test whether issue priority drives vote choice.

Hypothesis 1 (the winner takes all): The China factor is the primary issue of the DPP and Tsai’s supporters but is de-emphasized by her challengers and their supporters.

The second factor is the sole focus on one type of election without considering other concurrent electoral results. Many countries hold concurrent executive and legislative elections which take place under different electoral rules. For the executive office, the winner is usually determined by the first-past-the-post rule (FPTP), but for legislative elections, there are various systems in use, including FPTP, proportional representation (PR), mixed electoral systems, and single transferable voting (STV) etc. Although legislative elections held concurrently with presidential elections are likely to be affected by the presidential coattail effect in which a popular presidential candidate also boosts the legislative seats of their party (Golder, 2006), voters’ behavior in concurrent elections is still conditioned by the different electoral rules in use, for instance, voting sincerely in PR elections but strategically in FPTP elections (Duverger, 1954). In this case, votes for the winner of the presidential election partially come from strategic voting, and the motivation for these strategic voters is unlikely to be driven by the winner’s issues (otherwise, these voters should be categorized as sincere voters), and instead likely to be propelled by the loser’s inability to capitalize on these voters’ top concerns. To sort out what drives these strategic voters to lend their votes to the winner, we need to examine not just presidential elections but also legislative elections. To test the argument, Hypothesis 2 is proposed.

Hypothesis 2 (incomplete observation): Supporters of new parties vote strategically based on their issue priorities.

The final factor concerns the mentality of path dependency. When party systems and social cleavages converge, party systems become “frozen” (Lipset & Rokkan, 1967). Given this logic, party systems are subject to change when new social cleavages emerge. The emergence of new parties representing new cleavages or issues is a sign of such a party system change (Hooghe & Marks, 2018). However, such changes are usually slow and resisted by mainstream parties and their supporters as they struggle to adapt or perish. Party systems as an institution are less likely to face a sudden breakdown (this happens only in extreme cases); from a historical perspective, party systems are likely to evolve in a path-dependent way (Thelen & Conran, 2016). As long as mainstream parties still win elections, new cleavages or demands tend to be underestimated. Despite this, mainstream parties still have to respond and adapt to the new challenges. They may even try to suppress new cleavages by accommodating the issues “owned” by new parties and thus reduce the salience of these issues (Meguid, 2005). The resistance and adaptation of mainstream parties and their supporters are likely to create an impression of “business as usual.”

The above discussion explains why the China factor is interpreted as the primary determinant of the election outcome through three potential effects: the winner-take-all effect, the incomplete observation effect, and the path dependency effect. We empirically tested the first two effects through the hypotheses outlined above. Support for these two hypotheses can, in turn, explain the third effect.


This article relied on multiple sources of data. The first source was Facebook posts and comments of the four potential contenders in the 2020 presidential election and their followers, collected between March 1 and August 31, 2019. By August, it was clear that Tsai and Han would be the DPP and KMT nominees, while Gou and Ko would not join the race. Although James Soong was the third candidate in the election, given that he entered the race at the last minute and performed poorly (obtained only 5% of the votes), he was excluded from the analysis. The purpose of analyzing Facebook posts and comments was to understand the issues “owned” by each potential contender and the kinds of messages that received the most responses from their followers. We also uncovered possible issue competition besides national identity through text mining. Most politicians in Taiwan use Facebook to connect with their supporters and advocate their policies. In August 2019, Tsai and Ko each had around 2,000,0000 Facebook followers, while Gou had around 600,000 followers and Han had around 500,000 followers.

During the study period between March 1 and August 31, 2019, the four contenders had varying levels of activity on Facebook and received differing levels of engagement from their followers. We analyzed all posts but focused on posts that attracted the most engagement to analyze the comments. The most popular posts, logically speaking, should represent the issues that a politician’s supporters care about the most. The threshold was set at 100,000 interactions for a post, including likes and other emotional reactions as well as total shares. If no posts received more than 100,000 interactions, we selected the most popular posts to ensure comparability. Our next step was to crawl the comments on these selected posts. Table 1 reports the total posts and the number of selected posts (for analysis of comments) as well as the total number of comments for each contender. Tsai published the most posts, followed by Ko and Gou. While Han published fewer posts, his followers responded with the most comments.

Table 1.Comparison of Politicians’ Facebook Posts
Tsai Han Ko Guo
Total posts 387 147 242 187
Selected posts 33 31 34 31
Total comments 171,855 575,980 80,045 150,863

The second source of data was survey data collected by the Taiwan Election and Democratization Study (TEDS). Fieldwork was conducted after the 2020 presidential and legislative elections.[2] A face-to-face survey was carried out between January 13 and May 30, 2020. A total of 1,680 interviews were conducted with a response rate of 14%. We used these data to understand the strategic behavior of third-party supporters, in particular, supporters of Ko’s Taiwan People’s Party (TPP) and New Power Party (NPP). The TPP was founded in August 2019, and in its first legislative election in 2020, the party won five seats, all through the PR tier.[3] The TPP, like the NPP founded in 2015, is not a splinter party from the DPP or KMT. The NPP and TPP were founded by activists from the 2014 Sunflower movement and mainly attracted young and educated voters. However, the majority of NPP supporters are pro-independence, while the majority of TPP supporters opt for the status quo. Figure 1 compares the positions of party identifiers on national identity and unificationindependence issues, the traditional cleavages separating Taiwan’s parties. Although on average, most respondents identified themselves as Taiwanese, respondents who identified with the KMT and pan-blue parties were relatively likely to support unification with China, whereas respondents who identified with the DPP/pan-green parties and NPP strongly supported independence. As a result of Taiwanese identity and strong support for the status quo, TPP supporters were more aligned with the independents on traditional cleavages than other party supporters.

Figure 1
Figure 1.Comparisons of Party Identifiers’ Stances on Traditional Cleavages

Note. The vertical lines represent the means of the variables.

Election data then were published by Taiwan’s Central Election Commission. As Figure 2 shows, since Taiwan’s presidential and legislative elections were first held concurrently in 2008, the vote share of the winning president and the PR vote share of the president’s party has tended to be close, providing evidence of presidential coattail effect as found in the 2016 elections (Rich, 2018). However, the 2020 election was an exception; there was an 18% difference between the president’s vote share and her party’s PR vote share. Meanwhile, third parties gained a record PR vote share. Since neither the NPP nor TPP nominated any presidential candidates, Tsai’s record number of votes likely came from the TPP and NPP supporters. Here then comes the question: did TPP and NPP supporters vote for Tsai due to national identity or other issues?

Figure 2
Figure 2.Comparison of Vote Share in Presidential Elections and Party List Elections

Source: Central Election Commission. Percentages are recalculated after excluding parties below 5% of vote share.

Results and Discussion

In order to test Hypothesis 1, our first step was to analyze all the Facebook posts of the four politicians during the study period and compare the proportion of certain keywords in their posts. Since text is unstructured data, it needs to be processed to allow meaningful analysis of the messages conveyed. This study used Latent Dirichlet Allocation (LAD) for topic modeling to classify the posts and comments to determine whether the issues that politicians and their supporters care about differed. As this study only analyzed Facebook posts and comments, it was assumed that these commenters are ardent supporters of political figures, and their preferences may differ from the general public and such commenters mostly only post positive messages (Sweetser & Lariscy, 2008). Therefore, the results of the analysis cannot be extrapolated to the general electorate due to these limitations. Additionally, we were unable to identify fake accounts in the analysis, which can be used to boost the popularity of specific Facebook pages. This type of post tends to be short with encouraging words, so it is grouped into one category in the analysis. It does not affect our ability to understand why these politicians are attracting followers by exploring the issues they care about.

Figure 3 compares the themes of politicians’ posts and issues their followers are concerned about. We identified four topics for politicians’ posts and five topics for followers’ comments. Based on the top terms shown in each topic (see Table A in the Appendix), we named the four topics for politicians “democracy,” “economy and policy,” “municipal affairs,” and “publicity.” The upper panel in Figure 3 shows that over 50% of Tsai’s posts were categorized as “democracy,” as they mostly talked about defending Taiwan’s political system against China’s aggression and its intrusion into Hong Kong’s two-system arrangement. By contrast, most of the posts of Han and Ko talked about municipal affairs and related issues, reflecting their positions as mayors of Kaohsiung and Taipei. Most of Gou’s posts (besides publicity) were focused on the economy and policy, areas which he identified as relative strengths compared to Tsai.

For the comments beneath politicians’ posts, we identified five themes. Most comments were positive, demonstrating followers’ support for particular politicians. The lower panel of Figure 3 shows that most of the comments beneath Tsai’s posts were related to cross-Strait relations, echoing Tsai’s “democracy” theme. Comments beneath the posts of the remaining three politicians were mostly categorized as policy and criticisms. Han and Gou were competing in the KMT’s primary election, and thus, a large proportion of comments were related to this intra-party contest. Overall, Tsai and her supporters had much in common as they focused on democracy and anti-China sentiment as the most important issues. By contrast, these issues were relatively de-emphasized by the other three politicians and their followers.

Figure 3
Figure 3.Comparisons of Facebook Posts and Comments

Since most comments under the posts of the three potential challengers to Tsai concerned policy and criticisms, Figure 4 compares the key terms appearing in this theme across the four politicians. It is apparent that “economy” dominated the comments under Han and Gou’s posts. While Han’s supporters talked about the struggle of the subaltern common people and policy issues in Kaohsiung, Gou’s supporters commented on job opportunities. Ko’s supporters talked about policies in Taipei, for instance, the dome stadium, which was a topic of policy debate during Ko’s term as mayor, and the work ethics and governance Ko advocates, including openness and transparency. Anti-establishment sentiment also appeared, as the terms “blue-green” and “rubbish” indicate. Tsai’s supporters strongly praised human rights policies such as indigenous policies and same-sex marriage, showing a difference from the supporters of the other three politicians. It is generally held that Tsai and her supporters emphasize Taiwan’s sovereignty and democracy, which are viewed as two sides of the same coin. The supporters of challengers, by contrast, tended to emphasize the economy and governance, as well as anti-establishment and anti-elite sentiments. These comments demonstrate that supporters of political outsiders hope to break away from value competitions and shift the focus to economic or governance issues.

Figure 4
Figure 4.Comparisons of Key Terms under the Policy and Criticism Theme (Comments)

In the above analysis, supporters of challengers rarely mention cross-Strait issues, but does this mean that they are less concerned about sovereignty? We argue that it is a matter of issue priority. Supporters focus more on their political figures’ strengths. For example, the economy was often mentioned by Han and Gou because it is their issue (and Tsai’s weakness). Since the messages were posted on the Facebook pages of individual politicians, there was an echo chamber effect, so posts and comments tended to mirror each other. Supporters of politicians clearly identified the issues “owned” by the politicians that they support, and the issues owned by the other three politicians were distinct from traditional cross-Strait and national identity issues.

Accordingly, the results of the 2020 presidential election should be interpreted cautiously. Tsai stated in her victory speech that the Taiwanese people had chosen democracy and progress, words which reflected the position and attitude of her Facebook followers. However, this statement ignored the issues that other voters cared about. As Gou and Ko did not run in the election, their supporters had to choose between Tsai and Han. The election results indicate that most of them voted for Tsai, giving her the largest number of votes for a president in Taiwan’s history. Despite this, democratic and progressive values are not these voters’ primary concerns; the economy and governance are. Thus, it can be hypothesized that supporters of Ko and Gou used the issues of the economy and governance to assess the capabilities of Tsai and Han, and they ultimately chose Tsai. To show that it is the case, we used electoral and survey data to understand how Tsai was able to win the most votes.

Our first analysis, shown in Table 2, examined electoral data to understand whether the vote share in the PR tier for each party had a positive or negative impact on the difference in vote shares between Tsai and Han. A positive impact indicates that more votes cast for a party in the legislative election (PR tier) are correlated to more votes for Tsai in the presidential election at the township level. All models also controlled for region and the turnout in the presidential election. Since all models have the same dependent and control variables except for the PR vote share of different parties, it is likely that the disturbances of each equation are correlated, which violates the assumption of OLS (Zellner, 1962). To solve the problem, we used seemingly unrelated regression (SUR) to simultaneously estimate the equations.

The results showed that more votes for the DPP in the PR tier was positively related to Tsai winning more votes than Han (Model 1), whereas the PR vote for the KMT produced the opposite effect (Model 2). With regard to the impact of the PR vote share of the TPP, which did not nominate any presidential candidate, Model 3 shows that more votes for the TPP in the PR tier clearly increased Tsai’s votes at the township level (Model 3). This finding was repeated for the NPP (Model 4). The results remained intact and robust when we used a pooled sample (see Table C in the Appendix). The NPP’s stance on the unificationindependence issue is closer to the DPP’s position (as shown in Figure 1), so NPP supporters choosing Tsai is not surprising. TPP’s stance, on the other hand, is closer to the KMT’s position, yet most TPP supporters did not vote for Han. As the analysis of Facebook comments shows, Ko’s supporters care about governance issues more than national identity and cross-Strait relations. The 2020 electoral results further demonstrated that the old cleavages had little influence on TPP supporters’ vote choice. Since Guo did not form a party after losing the KMT primary, we can only infer his supporters’ vote choices from polls conducted before the election. TVBS conducted a poll after the KMT primary in July 2019, and found that 74% of those who identify as KMT would vote for Han, while 24% would vote for Guo. In terms of the independents (those who do not identify with a party), 34% would vote for Guo, higher than Han (28%), and Tsai (26%) (TVBS Poll Center, 2019). Clearly, Guo’s supporters are largely the independents, and so we will examine their voting choices to understand how Guo’s supporters would vote in the following section.

Table 2.The Determinants of the Difference in Vote Shares between Tsai and Han
Difference in Vote Share between Tsia and Han
(1) (2) (3) (4)
PR vote share 2.296*** -1.896*** 0.649** 1.075**
(0.044) (0.024) (0.242) (0.330)
Turnout president 0.667*** 0.032 1.320*** 1.314***
(0.096) (0.069) (0.219) (0.213)
South 0.020 0.034** 0.328*** 0.320***
(0.018) (0.012) (0.039) (0.038)
Central-South 0.084*** 0.031* 0.384*** 0.387***
(0.020) (0.014) (0.043) (0.042)
Central 0.094*** -0.009 0.196*** 0.209***
(0.018) (0.013) (0.041) (0.040)
Central-North 0.034 -0.047** -0.038 -0.038
(0.021) (0.014) (0.047) (0.046)
North 0.078*** -0.035** 0.192*** 0.197***
(0.018) (0.013) (0.041) (0.040)
Outlying islands 0.149*** -0.057* 0.123 0.134
(0.031) (0.022) (0.071) (0.069)
Constant -1.181*** 0.806*** -1.102*** -1.110***
(0.064) (0.051) (0.144) (0.141)
Observations 379
R2 0.9116 0.9571 0.4721 0.4802
Adjusted R2 0.9096 0.9562 0.4607 0.4690
Note. Base=Eastern region. Standard errors are in the parentheses. Significance level *p < .05; **p < .01; ***p < .001.
Model: Seemingly Unrelated Regression (SUR)

Table 3 further examines the determinants of voting for Tsai for different party supporters using survey data. Note that party supporters in this analysis were measured as those casting a PR vote for the party as well as its party identifiers to expand the number of observations, especially for the third parties. As supporters of challengers are hypothetically concerned about the economy and governance issues, we used two questions to represent these concerns. The first is the evaluation of candidates’ capability (What do you think of the overall capabilities of the candidates in this general election?), and the second is the belief that candidates understand the needs of ordinary people. These two questions were asked separately for each presidential candidate, and the scores ranged from 0 to 10. We also controlled for national identity and unificationindependence views.

Table 3.The Determinants of Voting for Tsai
Voting for Tsai
(1) (2) (3) (4) (5)
Tsai's capability 0.576*** 0.361*** 0.162 0.337* 0.126
(0.146) (0.098) (0.166) (0.143) (0.197)
Han's capability -0.252* -0.227** -0.231* -0.270** -0.248
(0.115) (0.078) (0.107) (0.097) (0.139)
Tsai's understanding 0.236 0.208* 0.334 0.333* 0.526*
(0.140) (0.095) (0.184) (0.136) (0.244)
Han's understanding -0.412*** -0.215** -0.241* -0.097 -0.098
(0.120) (0.078) (0.101) (0.101) (0.170)
Unification-independence 0.548 0.303 0.595 0.497 1.282*
(0.309) (0.231) (0.387) (0.275) (0.582)
National identity 0.836* 0.247 0.320 0.341 -0.280
(0.339) (0.242) (0.388) (0.343) (0.669)
Constant -5.806*** -2.927** -3.080* -3.405** -3.537
(1.251) (0.925) (1.225) (1.293) (2.143)
Observations 461 441 185 699 130
McFadden R2 0.4917 0.3594 0.3374 0.2958 0.3523
Note. Standard errors are in the parentheses. Significance level *p < .05; **p < .01; ***p < .001. No ID includes Independents and voters not casting a PR vote.
Model: Logistic regression

For KMT and pan-blue supporters, there was a trade-off between evaluations of Tsai and Han’s capabilities. As expected, positive evaluations of Tsai were associated with a greater probability of voting for her, while positive evaluations of Han had the opposite effect. The same pattern was observed for independents and DPP/pan-green supporters. However, the voting behavior of TPP supporters was only affected by positive evaluations of Han’s capability; evaluations of Tsai’s capability did not affect the voting behavior of TPP supporters. In terms of the perception of understanding people’s needs, the trade-off effect was observed among the independents. For DPP and NPP supporters, positive perceptions of Tsai increased the probability of voting for her, while perceptions of Han had no effect. KMT and TPP supporters showed the opposite pattern, with positive perceptions of Han decreasing the probability of voting for Tsai, while evaluations of Tsai had no effect on vote choice. NPP supporters’ voting behavior was also largely affected by their unification–independence disposition. For supporters of other parties, the old cleavages did not appear to affect their voting decisions. These results are likely due to a lack of variations on the traditional issues among supporters of particular parties. For our robustness checks, we ran separate analyses for the supporters of established parties, supporters of new parties, and independents. Our analysis was limited to those who identified with and voted for the party in the PR tier in order to exclude party identifiers who split votes. Additionally, we excluded national identity since it has a small variation across party identifiers and is positively related to the unificationindependence stance (Pearson correlation 0.4161). Considering that partisans are ardent party supporters, their vote choice is more likely to be influenced by the party’s issue ownership.

The findings show that DPP and KMT supporters held opposing views toward the overall capabilities of Tsai and Han, and these attitudes determined their voting decisions (see Model 1 in Table D in the Appendix). When party ID is included, the effect of unificationindependence stance disappears, and the effects of capabilities are also reduced. On the other hand, for supporters of TPP and independents, the significant effects were the evaluations of Han’s capability, while NPP supporters tend to give Tsai positive evaluations (see Table E in the Appendix). The findings largely conform to the results reported in Table 3.

Taking the findings together, TPP supporters showed a distinct pattern of voting behavior as their motives were more closely related to their assessments of Han rather than Tsai. KMT supporters also showed a tendency to vote based on their assessment of Han’s capability and understanding of people’s needs. Aside from Han’s staunch supporters, there are Gou’s supporters and urban, educated voters within the KMT who had reservations about Han. As this article argues, Tsai won over these voters not because of her anti-China stance but because of Han’s failure to convince these voters that he would be capable of solving the economic and governance issues they prioritize.

Our analysis shows that the supporters of political outsiders care about issues outside of traditional cleavages and that they voted based on these issues (H1 and H2 had empirical support). However, one question remains: Why have traditional cleavages still dominated explanations of electoral outcomes? The mentality of path dependence is reinforced by the winner-take-all and incomplete observation effects, as the issues “owned” by the election winner are given credit for the election victory in the presidential competition, eclipsing signs of changes in the party system in the legislative election. In the context of Taiwan, mainstream parties are also protected by the electoral system in which FPTP with single-member districts (SMD) benefit large parties, meaning that new parties are only able to win seats in the legislature through the PR tier unless they coordinate with mainstream parties in the SMD tier. Because only 30% of seats (34 seats) are allocated to the PR tier, new parties face major obstacles to achieving an electoral breakthrough.

Moreover, policy adaptation also plays a role. During his bid to win the KMT primary, Gou announced a plan to lessen childcare costs for parents of young children. This policy was later adopted by Tsai with the slogan: “the state [and parents] raise children together” (National childcare policy for ages 0–6). In his campaign speeches, although Han often focused on the grievances of the economically marginalized, he seldom talked in detail about how he planned to improve living standards. By contrast, Tsai’s speeches were less centered on the struggles of ordinary people, but she followed the scripts prepared by her staff to lay out welfare programs for seniors and parents and young people. For voters who care about policies (such as the independents and TPP supporters), Tsai appeared to be a more reliable choice than Han. In other words, the forces behind the rise of political outsiders around the world also appeared in Taiwan, but the failure of Han to convince these voters that he would make a difference prevented him from winning the election. As a result, DPP winning the competition along the old cleavages (the China factor) was interpreted as the most important factor determining the electoral outcome. However, this article reveals undercurrents that, if left unattended, could create another wave of political outsiders challenging the status quo in the future.


The purpose of this article is to demonstrate how issues that were “owned” by political outsiders and their supporters affected voters’ choice between established political figures and political outsiders. Based on an analysis of Facebook posts and comments, we showed that Tsai and her Facebook followers prioritized democracy and progressive values, while political outsiders and their followers emphasized the economy and governance. The differing issue ownerships led to the interpretation that the China factor and anti-China sentiment helped Tsai win the presidential election. However, in reality, the China issue was rarely an important issue of concern for supporters of political outsiders. The winner-take-all effect can explain how the election results were interpreted according to the winner’s story. Further analysis of electoral and survey data showed that TPP supporters tended to vote for Tsai, not because of her stance on national identity and unification–independence issues, but due to Han’s perceived incompetence as a leader. When we fail to take the results of legislative elections into account, incomplete observations are likely to lead to incorrect conclusions. Finally, the mentality of path dependency has reinforced the belief that electoral competition continues to be centered on old social cleavages and mainstream parties still rely on these cleavages to win elections. We argue that this conclusion oversimplifies the choices of voters.

The rise of political outsiders and challenger parties indicates that voters are tired of political gridlocks and disagreements over old cleavages and are looking to politicians to pay attention to issues of livelihood and governance. Many people are wary of changing the status quo as a result of the Russian-Ukraine war and the Chinese military exercises near Taiwan following Tsai’s meetings with Nancy Pelosi and Kevin McCarthy, the two U.S. Speakers of the House. The government’s reaction to the situation also showed that it was not prepared for such a conflict. Consequently, Taiwanese citizens reconsidered whether an anti-China stance was the best strategy to ensure the country’s safety. As implied by the findings of this article, economic and governance issues, as well as avoiding cross-Strait conflict, are likely to be the main topics of discussion in the upcoming presidential election in early 2024. As of writing this article, Ko has announced his candidacy for president, and Guo once again sought nomination from the KMT, but lost to New Taipei Mayor Hou Yu-ih, who is a fringe figure within the party. All three are expected to challenge the established figure, the DPP’s nominee William Lai, for the 2024 election, a combination similar to the 2020 election. However, with the looming threat of war with China, all candidates will have to embrace the status quo, which is where the majority of Taiwan voters stand. This situation makes Huang and James’ 2014 statement even more applicable today: “Aquamarine – a combination of Blue and Green positions into a status quo of de facto sovereignty – is likely to be the lasting color of Taiwan” (Huang & James, 2014, p. 688).


The author would like to thank the two anonymous reviewers for their comments and suggestions. This work was supported by the Ministry of Education of Taiwan (108L893502).

Accepted: May 26, 2023 KST


Barr, R. R. (2009). Populists, outsiders and anti-establishment politics. Party Politics, 15(1), 29–48. https://doi.org/10.1177/1354068808097890
Google Scholar
Batto, N. F. (2020, January 12). Opinion: Still, President Tsai of Taiwan wasn’t re-elected just because she’s tough on China. The New York Times. https://www.nytimes.com/2020/01/12/opinion/taiwan-election-tsai-han-populism.html
BBC News. (2020, January 11). Taiwan election: Tsai Ing-wen wins second presidential term. BBC. https://www.bbc.com/news/world-asia-51077553
Bélanger, É., & Meguid, B. M. (2008). Issue salience, issue ownership, and issue-based vote choice. Electoral Studies, 27(3), 477–491. https://doi.org/10.1016/j.electstud.2008.01.001
Google Scholar
Budge, I. (2015). Issue emphases, saliency theory and issue ownership: A historical and conceptual analysis. West European Politics, 38(4), 761–777. https://doi.org/10.1080/01402382.2015.1039374
Google Scholar
Budge, I., & Farlie, D. (1983). Explaining and predicting elections: Issue effects and party strategies in twenty-three democracies. Taylor & Francis.
Google Scholar
Chang, C., & Jennings, R. (2020, January 11). Tsai Ing-wen reelected, promises to defend Taiwanese sovereignty from China. Los Angeles Times. https://www.latimes.com/world-nation/story/2020-01-11/taiwans-president-sweeps-to-re-election-setting-the-stage-for-chilly-relations-with-china
Chen, C. (2020, January 21). Tsai’s victory speech at international press conference. Embassy of the Republic of China (Taiwan) to the Holy See. https://www.taiwanembassy.org/va_en/post/3411.html
Chu, Y.-H., Huang, K.-P., Lagos, M., & Mattes, R. (2020). A lost decade for third-wave democracies? Journal of Democracy, 31(2), 166–181. https://doi.org/10.1353/jod.2020.0029
Google Scholar
Duverger, M. (1954). Political parties. Wiley.
Google Scholar
Fell, D. (2005). Measuring and explaining party change in Taiwan: 1991–2004. Journal of East Asian Studies, 5(1), 105–133. https://doi.org/10.1017/s1598240800006275
Google Scholar
Fell, D. (2021). Challengers to mainstream parties in Taiwan’s 2020 elections: Continuity rather than the expected change. Taiwan Journal of Democracy, 17(1), 141–160. https://eprints.soas.ac.uk/35278
Google Scholar
Golder, M. (2006). Presidential coattails and legislative fragmentation. American Journal of Political Science, 50(1), 34–48. https://doi.org/10.1111/j.1540-5907.2006.00168.x
Google Scholar
Hanley, S., & Sikk, A. (2016). Economy, corruption or floating voters? Explaining the breakthroughs of anti-establishment reform parties in eastern Europe. Party Politics, 22(4), 522–533. https://doi.org/10.1177/1354068814550438
Google Scholar
Hibbing, J. R., & Theiss-Morse, E. (2002). stealth democracy: americans’ beliefs about how government should work. Cambridge University Press. https://doi.org/10.1017/cbo9780511613722
Google Scholar
Hooghe, L., & Marks, G. (2018). Cleavage theory meets Europe’s crises: Lipset, Rokkan, and the transnational cleavage. Journal of European Public Policy, 25(1), 109–135. https://doi.org/10.1080/13501763.2017.1310279
Google Scholar
Hsiao, Y.-C., & Yu, E. C.-H. (2020). Polarization perception and support for democracy: The case of Taiwan. Journal of Asian and African Studies, 55(8), 1143–1162. https://doi.org/10.1177/0021909620911150
Google Scholar
Hsieh, J. F.-S., & Niou, E. M. S. (2005). Measuring Taiwanese public opinion on Taiwanese independence. The China Quarterly, 181, 158–168. https://doi.org/10.1017/s0305741005000093
Google Scholar
Huang, C.-H., & James, P. (2014). Blue, green or aquamarine? Taiwan and the status quo preference in cross-Strait relations. The China Quarterly, 219, 670–692. https://doi.org/10.1017/s0305741014000745
Google Scholar
Lin T.-J., & Hsiao Y.-C. (2021). Wèishéme bù xǐhuān guómíndǎng yǔ mín jìn dǎng? Tànsuǒ táiwān mínzhòng fǎn zhèngdǎng qíngxù de láiyuán [Why do people dislike both the KMT and the DPP? Exploring the sources of anti-party sentiment in Taiwan]. zhèngzhì xué lùncóng [Taiwanese Journal of Political Science], 89, 89–120. https://doi.org/10.6166/TJPS.202109_(89).0003
Google Scholar
Lipset, S. M., & Rokkan, S. (1967). Cleavage structures, party systems, and voter alignments: An introduction. In S. M. Lipset & S. Rokkan (Eds.), Party Systems and Voter Alignments: Cross-National Perspectives (pp. 1–64). Free Press New York.
Google Scholar
Liu, C.-L. (2020). Cóng 2020 nián dàxuǎn kàn lán lǜ jīběn bǎnkuài de yídòng yǔ liǎng dàng zài xuǎnjǔ cèlüè de sīkǎo [The power shift between Taiwan’s ruling & opposition parties and their election strategies based on Taiwan’s 2020 presidential election]. Zhǎnwàng yǔ tànsuǒ yuèkān [Prospect & Exploration], 18(4), 52–62.
Google Scholar
Liu, L.-W. (2022). 2018 nián jiǔ hé yī xuǎnjǔ wǎngluò fǎn dǎng qíngxù: Dà shùjù de wéidù yǔ láiyuán [Anti-party sentiment on the Internet in the nine-in-one election in 2018: Dimension and source in big data]. shèhuì kēxué yǔ zhéxué zázhì [Journal of Social Sciences and Philosophy], 34(1), 75–116.
Google Scholar
Meguid, B. M. (2005). Competition between unequals: The role of mainstream party strategy in niche party success. American Political Science Review, 99(3), 347–359. https://doi.org/10.1017/s0003055405051701
Google Scholar
Petrocik, J. R. (1996). Issue ownership in presidential elections, with a 1980 case study. American Journal of Political Science, 40(3), 825–850. https://doi.org/10.2307/2111797
Google Scholar
Pop-Eleches, G. (2010). Throwing out the bums: Protest voting and unorthodox parties after communism. World Politics, 62(2), 221–260. https://doi.org/10.1017/s0043887110000043
Google Scholar
Rich, T. S. (2018). Coattails and mixed electoral systems: Evidence from Taiwan’s 2016 election. Journal of East Asian Studies, 18(1), 47–66. https://doi.org/10.1017/jea.2017.25
Google Scholar
Sikk, A. (2012). Newness as a winning formula for new political parties. Party Politics, 18(4), 465–486. https://doi.org/10.1177/1354068810389631
Google Scholar
Sweetser, K. D., & Lariscy, R. W. (2008). Candidates make good friends: An analysis of candidates’ uses of Facebook. International Journal of Strategic Communication, 2(3), 175–198. https://doi.org/10.1080/15531180802178687
Google Scholar
Templeman, K. (2020). How Taiwan stands up to China. Journal of Democracy, 31(3), 85–99. https://doi.org/10.1353/jod.2020.0047
Google Scholar
Thelen, K., & Conran, J. (2016). Institutional Change. In O. Fioretos, T. G. Falleti, & A. Sheingate (Eds.), Oxford Handbook of Historical Institutionalism (pp. 51–70). Oxford University Press. https://doi.org/10.1093/oxfordhb/9780199662814.013.3
Google Scholar
Tiezzi, S. (2020, January 12). Taiwan “shouts back”: President Tsai wins re-election despite China’s pressure campaign. The Diplomat. https://thediplomat.com/2020/01/taiwan-shouts-back-president-tsai-wins-re-election-despite-chinas-pressure-campaign/
TVBS Poll Center. (2019). Guómíndǎng chū xuǎn hòu 2020 nián zǒngtǒng dàxuǎn mín diào [Polls after the KMT primary for the 2020 presidential election]. https://cc.tvbs.com.tw/portal/file/poll_center/2019/20190718/92ae63afafc8669dd8cbd2de127cd5c4.pdf
van der Brug, W. (2004). Issue ownership and party choice. Electoral Studies, 23(2), 209–233. https://doi.org/10.1016/s0261-3794(02)00061-6
Google Scholar
Zellner, A. (1962). An efficient method of estimating seemingly unrelated regressions and tests for aggregation bias. Journal of the American Statistical Association, 57(298), 348–368. https://doi.org/10.1080/01621459.1962.10480664
Google Scholar


Table A.Top Key Terms of Politicians’ Facebook Posts
Topic Key terms
Democracy democracy, president, people, nation, Republic of China, U.S., international, politics, diplomacy, Hong Kong, China, Freedom, premier, protect, govern, freedom and democracy, peace, allies, military, cooperation, sustainable, safety, value, defense, sovereignty, democracy and freedom, persistence, unity
Economy and policy economy, industry, development, investment, agriculture, technology, world, nation, worldwide, enterprise, growth, assistance, farmer, innovation, chance, skill, cooperation, corporation, creation, transformation, young people
Municipal affairs city, councilor, culture, value, question, hope, sightseeing, information, progress, program, municipal governance, citizen, report, improve, glory, market, policy, subsidy, promote, typhoon
Publicity welcome, Line, education, children, hope, live streaming, film clips, national flag, night, thanks, channel, subscription, telephone, young people, activity, invite, join in, social media manager, link, make friends
Table B.Top Key Terms of Facebook Comments
Topic Key terms
Cross-strait relations Hong Kong, democracy, One country, two systems, mainland, CCP,
1992 consensus, re-elected, freedom, straw bag, independence, military, democracy and freedom, sovereignty, unification, defend, peace, make a fortune, consensus, guard, Communist Party, protect, Taiwan, cross-strait, Xi Jinping, Anti-Extradition Law Amendment Bill Movement
Policy & criticism economy, politician, tolerate, persist in, ballot, young people, govern, society, lead, life, job, love, choice, expectation, proud, live streaming, step down, policy, national flag, legislator, central government, reform,
Electoral issues election, telephone, firm, smear, take down, care, official document, Mazu, judicial, attack, show, watchful, conscience, corruption, protest,
smear Han, Pitiful, lie, shameless, fake news
Primary election primary election, poll, media, unity, common people, blue-green, touching, success, rubbish, vote, candidate, change, China Television Incorporation, sincere, wisdom, supporter, represent, disappointment, Chairman, dislike, rich and powerful
Support effort, elected, take care of yourself, happiness, voter, like, figure, rich,
safety, break away from the party, go down, nausea, smear, campaign,
take a rest, flood, criticize, thank you, flip, four years, grateful, attentively
Table C.The Determinants of the Difference in Vote Shares Between Tsai and Han (pooled sample)
Coefficient Standard error
DPP PR vote share 1.054*** (0.019)
KMT PR vote share -1.177*** (0.016)
TPP PR vote share 0.499*** (0.084)
NPP PR vote share 0.333** (0.118)
Turnout President 0.034 (0.025)
South 0.014** (0.005)
Central-South 0.014** (0.005)
Central 0.005 (0.005)
Central-North -0.025*** (0.005)
North -0.022*** (0.005)
Outlying islands -0.023** (0.008)
Constant 0.118*** (0.022)
Observations 379
R2 0.9955
Adjusted R2 0.9954

Note: Base=East; significance level *p < .05; ** p < .01; *** p < .001.
Model: OLS

Table D.The Determinants of Voting for Tsai (DPP and KMT)
Voting for Tsai
Model 1 Model 2
Tsai’s capability 0.864*** 0.646***
(0.158) (0.184)
Han’s capability -0.582*** -0.297*
(0.122) (0.126)
Tsai’s understanding 0.080 -0.058
(0.138) (0.163)
Han’s understanding -0.230 -0.215
(0.123) (0.128)
Unification-Independence 0.866** 0.433
(0.316) (0.373)
DPP 3.215***
Constant -3.947*** -3.571**
(1.105) (1.309)
Observations 710
McFadden R2 0.7780 0.8248

Note. Standard errors are in the parentheses. Significance level *p < .05; ** p < .01; *** p < .001. Base=KMT and pan-blue
Model: Logistic regression

Table E.The Determinants of Voting for Tsai (TPP, NPP, and Independents)
Voting for Tsai
Model 1 Model 2
Tsai’s capability 0.383* 0.280
(0.181) (0.189)
Han’s capability -0.407** -0.418**
(0.133) (0.138)
Tsai’s understanding 0.089 0.126
(0.191) (0.198)
Han’s understanding -0.100 -0.101
(0.106) (0.107)
Unification-Independence 0.744 0.383
(0.403) (0.430)
TPP 0.341
NPP 1.788*
Constant -2.645* -2.001
(1.139) (1.188)
Observations 138
McFadden R2 0.3270 0.3605

Note. Standard errors are in the parentheses. Significance level *p < .05; ** p < .01; *** p < .001.
Base=Independents and No vote
Model: Logistic regression

Table F.Descriptive Statistics and Coding Scheme
Electoral Data min. mean max. SD No. of Obs.
Difference in Presidential vote share -0.717 0.115 0.557 0.286 379
PR vote share DPP 0.017 0.376 0.812 0.141 379
PR vote share KMT 0.045 0.331 0.538 0.109 379
PR vote share TPP 0.001 0.092 0.175 0.031 379
PR vote share NPP 0.004 0.065 0.125 0.019 379
Turnout presidential election 0.211 0.719 0.806 0.076 379
East 0 0.108 1 0.311 379
South 0 0.290 1 0.454 379
Central-South 0 0.106 1 0.308 379
Central 0 0.179 1 0.384 379
Central-North 0 0.090 1 0.286 379
North 0 0.185 1 0.389 379
Outlying islands 0 0.042 1 0.201 379
Survey Data min. mean max. SD No. of Obs.
Voting for Tsai 0 0.586 1 0.493 1485
Tsai's capability 0 6.348 10 2.521 1666
Han's capability 0 3.483 10 2.984 1667
Tsai's understanding 0 6.155 10 2.48 1666
Han's understanding 0 4.012 10 2.959 1666
Unification-independence 1 2.285 3 0.633 1629
National identity 1 2.604 3 0.566 1659
DPP and pan-green identifiers 0 0.345 1 0.476 1673
DPP and pan-green PR votes 0 0.395 1 0.489 1579
KMT and pan-blue identifiers 0 0.209 1 0.407 1673
KMT and pan-blue PR votes 0 0.280 1 0.449 1579
TPP identifiers 0 0.060 1 0.238 1673
TPP PR votes 0 0.102 1 0.303 1579
NPP identifiers 0 0.039 1 0.195 1673
NPP PR votes 0 0.077 1 0.266 1579
No ID (Independents) 0 0.345 1 0.476 1673
Did not cast a PR vote 0 0.147 1 0.354 1579
Coding scheme
East Yilan, Hualien, Taitung
South Pingtung, Kaohsiung, Tainan
Central-South Chiayi, Yunlin
Central Changhua, Nantou, Taichung
Central-North Miaoli, Hsinchu
North Taoyuan, New Taipei, Taipei, Keelung
Outlying islands Penghu, Lienchiang, Kinmen
DPP and pan-green identifiers DPP, Green Party, Pan-green, Taiwan Statebuilding Party
DPP and pan-green PR vote Taiwan Solidarity Union, Labor Party
Taiwan Statebuilding Party, Formosa Alliance, Taiwan Action Party Alliance
KMT and pan-blue identifiers KMT, New Party, People First Party, Republican Party
KMT and pan-blue PR votes KMT, New Party, People First Party, Congress Party Alliance, Stabilizing Force Party, Interfaith Union

  1. This article defines political outsiders as politicians not in the power center of mainstream parties. See Barr (2009).

  2. More details can be found on the Taiwan’s Election and Democratization Study website http://teds.nccu.edu.tw/intro2/super_pages.php?ID=intro11&Sn=166.

  3. Taiwan uses a mixed electoral system, with 79 legislative seats selected from single-member districts and 34 seats from party lists using proportional representation.