Kaohsiung 50-50 Poll 'Divine Prediction'? Only 1.64% Gap — Experts Reveal Three Major Doubts: Could Become a Self-Cheering Tool for Specific Camp

The Kaohsiung mayoral election is entering its final stretch. Kaohsiung City Councilor Chang Po-yang recently hinted that a '50-50 online poll' might emerge this week. Yesterday (10th), a web media outlet indeed released a commissioned online survey conducted by Pearson Data, showing Lai Jui-long at 47.78% and Ke Chih-en at 46.14%—a mere 1.64 percentage point difference. However, as this is not a formal poll and exhibits several anomalies, it has sparked debate.

A local expert familiar with poll interpretation points out that beyond the conveniently '50-50' outcome, the data itself raises at least three concerns.

First, the survey explicitly targets 'internet users aged 20 and above' and uses online self-selection distribution, differing from standard random-digit dialing and interviewer-administered telephone surveys. In other words, it does not qualify as a statistically valid public opinion poll.

Second, the combined support for Lai and Ke reaches 93.92%, with only 4.24% undecided. With 110 days still until voting day, a 94% declared support rate significantly diverges from typical polling and actual voter behavior.

In June, multiple Kaohsiung mayoral polls showed much higher undecided rates. The New Taiwan Think Tank reported Lai at 46.2%, Ke at 28.4%, and 25.4% undecided. The Mirror News poll showed Lai at 45.1%, Ke at 24.8%, with 29.3% undecided. Generally, undecided rates hover near 30%, aligning more closely with real-world turnout. Pearson’s 4.24% is a stark outlier.

Third, Pearson claims Ke Chih-en leads significantly among the 23.14% who 'choose candidates over parties or are neutral,' framing this 'independent voter' bloc as Ke’s strength—a key talking point for the Blue Camp. However, the same report notes this group is '53.87% anti-Green, 31.30% anti-Blue' and adds, 'not truly neutral.'

The expert notes inconsistency: the report first labels them 'neutral voters,' then later defines 'true centrists' as only 7.34%. Additionally, the poll contains a page mistakenly labeled 'New Taipei City,' ran for six days, and was released nine days after completion—deviations from standard polling practices.

Pearson’s March poll in Tainan showed Chen Ting-fei at 45.96% and Hsieh Lung-chieh at 40.57%—a 5.39-point gap. Yet, TVBS, using phone interviews less than two weeks later, found Chen at 53% and Hsieh at 30%, a 23-point lead. Such a drastic shift in two weeks is implausible, raising suspicions of biased data to generate headlines.

The expert warns that online surveys lack random sampling, attracting highly motivated, ideologically clear supporters, leading to 'self-selection bias' in statistics. Thus, such polls risk becoming self-congratulatory tools for specific political camps rather than accurate reflections of public sentiment.