The 2026 Kaohsiung mayoral race has entered a heated phase, with the latest poll released on the 10th by JU News Network showing DPP-nominated Lai Jui-long at 47.78% support, closely followed by KMT's Ko Chih-en at 46.14%, a narrow margin of just 1.64 percentage points—within the margin of error. Senior media analyst Huang Yang-ming noted that Kaohsiung's fundamental political structure remains 'greener than blue,' but Ko Chih-en's ability to narrow the gap stems from securing not only the KMT's core base but also gaining favor among centrist voters and supporters of the Taiwan People's Party. However, he cautioned that it's too early to speak of 'flipping Kaohsiung,' with the biggest variable being whether the DPP can achieve genuine internal integration.

Is the 1.64-point gap between blue and green accurate? Huang first emphasized that the polling methodology differs from some traditional surveys. He pointed out that JU News Network collaborated with Nielsen for this survey, which excluded respondents with no voting intention or undecided voters before calculating candidate support among those expressing intent. As a result, the declared support rate appears naturally higher. 'They eliminated all undecided respondents, so the declared rate is almost always elevated,' Huang explained.

Can Ko Chih-en replicate the 2018 'Han Wave'? With only a 1.64-point gap, KMT supporters are once again hopeful about 'flipping Kaohsiung.' However, Huang stressed that it's still mid-August, and the race is far from decided. Citing the 2018 Kaohsiung mayoral election where Han Kuo-yu's momentum shifted in late August, Huang warned against directly applying that scenario to Ko Chih-en. 'Does Ko Chih-en have the same energy as Han Kuo-yu's reversal? Honestly, the two election contexts are completely different,' Huang stated, adding that the current poll closeness only proves competitiveness, not that Ko has already recreated the 'Han Wave.'

What could enable Ko's comeback? Huang believes the key to breaking Kaohsiung's long-standing green-dominated political structure lies in the DPP's internal cohesion. He noted that if factional rifts remain after the DPP's primary and nomination process—leading some supporters to vote passively—Ko would naturally gain ground. Conversely, if the green camp unites, the DPP's massive base will remain Lai Jui-long's biggest advantage. 'Ko Chih-en heavily depends on DPP's fragmentation. If the DPP's factions are divided and disunited, then Ko certainly has a chance,' Huang said. Analyzing support structures, he identified three main pillars for Ko: 'centrist voters, TPP supporters, and KMT loyalists'—the reason she's narrowed the gap to 1.64% even in traditionally green-leaning districts.

In contrast, Lai Jui-long is currently supported primarily by the green base. With Kaohsiung's 'greener than blue' structure still evident, even with Ko's edge among centrists and white camp voters, Lai maintains a slight lead. 'At least from this poll, Ko's support comes from centrists, TPP, and KMT; Lai relies entirely on green supporters. Since the green base in Kaohsiung is large enough, he can still lead,' Huang observed.

Can one poll predict a reversal? Lai's camp has questioned the survey methodology, but Huang noted that this approach isn't unprecedented. A similar method was used in the 2022 Kaohsiung mayoral election, though Chen Chi-mai's larger lead over Ko Chih-en at the time sparked less controversy. He emphasized that different polling agencies use different sampling and calculation methods, leading to natural variations. Therefore, the key isn't focusing on a single data point but tracking ongoing trends. 'The Kaohsiung race still holds multiple variables: DPP's unity, centrist voter flow, TPP supporters' final choices, and whether Ko can further break into the green base. In short, Kaohsiung remains worth watching,' Huang concluded.

The survey was commissioned by JU News Network and executed by Pearson Data, funded by JU News Network. It targeted Kaohsiung residents aged 20 and above with internet access, conducted over six days from July 27 to August 1, 2026, yielding 1,608 valid samples. At a 95% confidence level, the sampling error is within ±2.40%.

FACT BOX

  • Source: PR Times
  • Category: Survey