SPI twice-monthly election brief | A first look at current support shares and changes among the same respondents. Full analysis will follow after the next survey is completed.
September’s tracking showed Puma Shen closing the gap in Taipei and Ko Chih-en’s support share rising in Kaohsiung. In October, answers from the same respondents show a narrower gap between Taipei’s two leading candidates, while Ko’s share did not continue the increase seen in the previous comparison.
This comparison includes people who answered in both 16–30 September and 1–9 October, a different paired group from the September article. Across the six cities, 13.9% changed their answer and around a quarter still had no stated preference. We first look at current support shares, then at how the same people’s choices changed. Each city’s results are small-sample observations and are not enough to establish a trend.
Six-city snapshot, 1–9 October: who has the higher share, and how many still have no stated preference?
Swipe left or right to read the table
| City | Candidate | Candidate | No stated preference |
|---|---|---|---|
| Taipei | Puma Shen 29.2% | Chiang Wan-an 37.1% | 15.7% |
| New Taipei | Su Chiao-hui 26.7% | Li Ssu-chuan 28.6% | 27.6% |
| Taoyuan | Chang San-cheng 20.0% | Huang Shih-chieh 12.5% | 45.0% |
| Taichung | Ho Hsin-chun 33.9% | Chiang Chi-chen 30.5% | 25.4% |
| Tainan | Hsieh Lung-chieh 24.0% | Chen Ting-fei 42.0% | 28.0% |
| Kaohsiung | Lai Jui-lung 22.4% | Ko Chih-en 34.2% | 30.3% |
Among respondents in the six cities on 1–9 October, Chiang Wan-an recorded 37.1% and Puma Shen 29.2% in Taipei; Ko Chih-en recorded 34.2% and Lai Jui-lung 22.4% in Kaohsiung. The two leading candidates’ shares were close in New Taipei and Taichung. Taoyuan had the highest share with no stated preference, at 45.0%; Tainan and Kaohsiung were also close to three in ten.
This table describes the choices of the current respondents. To understand movement in support, we must compare the same people who answered in both surveys. We cannot simply subtract the percentages in the earlier article, which used a different respondent group.
Changes among the same people: 16–30 September → 1–9 October
Swipe left or right to read the table
| City | Candidate choice: 9/16–9/30 → 10/1–10/9 | Candidate choice: 9/16–9/30 → 10/1–10/9 | No stated preference: 9/16–9/30 → 10/1–10/9 |
|---|---|---|---|
| Taipei | Puma Shen 29.8% → 31.6% | Chiang Wan-an 43.9% → 40.4% | 10.5% → 12.3% |
| New Taipei | Su Chiao-hui 30.3% → 30.3% | Li Ssu-chuan 33.3% → 31.8% | 19.7% → 24.2% |
| Taoyuan | Chang San-cheng 16.7% → 16.7% | Huang Shih-chieh 16.7% → 16.7% | 50.0% → 55.6% |
| Taichung | Ho Hsin-chun 29.7% → 29.7% | Chiang Chi-chen 27.0% → 35.1% | 37.8% → 24.3% |
| Tainan | Hsieh Lung-chieh 19.4% → 22.6% | Chen Ting-fei 38.7% → 35.5% | 35.5% → 35.5% |
| Kaohsiung | Lai Jui-lung 22.0% → 22.0% | Ko Chih-en 34.0% → 32.0% | 26.0% → 30.0% |
In this Taipei group, Puma Shen moved from 29.8% to 31.6%, while Chiang Wan-an moved from 43.9% to 40.4%. Their gap narrowed from 14.0 to 8.8 percentage points. Yet the share with no stated preference rose from 10.5% to 12.3%. A narrower gap does not mean undecided respondents are converging on one candidate. These are observations among the same respondents; the limited base is not enough to establish that the gap is on a downward trend.
In this Kaohsiung group, Ko Chih-en moved from 34.0% to 32.0%, Lai Jui-lung remained at 22.0%, and the share with no stated preference rose from 26.0% to 30.0%. Ko had the higher share among all current respondents, but the people who answered both surveys did not continue the increase seen in the previous comparison.
Across the six cities, the same respondents’ share with no stated preference moved from 25.5% to 26.3%. The 13.9% who changed their answer include people choosing a different candidate, moving from no stated preference to a choice, or returning to a wait-and-see position. They did not all switch to another candidate. The detail by age, party background and gender remains insufficient for a reliable judgment.
Among the same respondents across the six cities, the share choosing Democratic Progressive Party candidates remained at 28.6%, while the share choosing Kuomintang candidates remained at 32.0%. In Taipei and New Taipei combined, the former moved from 30.1% to 30.9%, and the latter from 38.2% to 35.8%. These are changes in this group’s answers and are not yet enough to establish a trend. A chosen candidate’s party is also not the respondent’s own party identification.
Who is still undecided? Women have a higher share without a stated preference than men
Among current six-city respondents, 33.0% of women were undecided, compared with 21.0% of men. Women accounted for 61.1% of those without a stated preference, above their 49.9% share of all respondents. Women therefore make up a larger share of this group. These percentages alone cannot tell us which issues matter to them or why they are still waiting.
The full article will follow after the next survey is completed, with more tracking results and a complete explanation of the method.
Further reading: Taipei’s gap narrows, Kaohsiung remains contested—what will undecided voters choose next?
Methodology and data notes
This self-funded online survey was conducted by ShareParty Insight, led by Tatt Chen (陳彥廷), among SPI members aged 20 and above who live in Taiwan and have voting household registration in one of the six cities. SPI follows the same members over time, linking each person’s answers to compare changes in their choices. The same-person table includes only those who answered both surveys, avoiding the mistake of treating a change in respondent composition as a change in preference. Fieldwork took place on 16–30 September and 1–9 October 2026. The 16–30 September comparison uses the complete archived data, rather than the preliminary cutoff used in the earlier article. The current mayoral question has 419 valid answers: Taipei 89, New Taipei 105, Taoyuan 40, Taichung 59, Tainan 50 and Kaohsiung 76. Of these, 259 can be compared across both surveys: Taipei 57, New Taipei 66, Taoyuan 18, Taichung 37, Tainan 31 and Kaohsiung 50; Taipei and New Taipei together account for 123. Taipei’s paired group differs from the 53-person comparison in the earlier article. The population reference is the Ministry of the Interior’s national household-registered population aged 20 and above at the end of August 2026, 19,640,702; it is not a six-city sampling frame. Candidate options were added in mid-September, so some changed answers may reflect changes in the options offered. These are answer changes, not confirmed vote switches or content effects.
This is a preliminary small-sample release. Percentages are unweighted, skipped answers are not imputed, and the figures are neither citywide support estimates nor election forecasts. Under a simple random sampling calculation, the maximum 95% reference errors for the cities in the order above are approximately ±10.4, ±9.6, ±15.5, ±12.8, ±13.9 and ±11.2 percentage points. The actual sample is non-probability, and those figures do not cover participation bias. The 1–9 October results were extracted on the evening of 9 October; the exact state of the data at 18:00 cannot be reconstructed. Full tracking analysis and methodology will follow after the next survey is completed.