When the Institute for National Defense and Security Research (INDSR) released its latest public opinion survey on June 18, the results moved immediately through Taiwan's political commentary and produced, within hours, three incompatible headlines.
The DPP pointed to the finding that 44.9% of respondents chose "strengthening self-reliant defence capabilities" as the most important approach to cross-strait security. KMT-aligned commentary and Beijing-friendly media fixed instead on the 29.7% who preferred "extending goodwill toward mainland China." The South China Morning Post framed the story as evidence that more Taiwanese favour goodwill over defence ties with the United States.
All three readings were accurate. None was informative.
Many Taiwanese hold several security preferences at once, but surveys often require them to pick just one. Politicians then cite those results as proof of public backing, and the policies shaped by those battles often endure long after the survey cycle has passed.
The May 2026 defence budget is the clearest recent illustration of what that process produces in practice. Divided democracies under external pressure are especially susceptible to this dynamic, and Taiwan's case is distinctive because the distortion feeds directly into legislative contests with material security consequences.
The Flaw and Its Exploitation
A poll that forces respondents to choose one option from several they may support at the same time leaves out much of how they actually think. Someone who prioritises self-reliance but also supports engagement with China and cooperation with the United States is recorded the same way as someone who supports self-reliance and rejects the other two approaches. The 44.9% figure cannot tell them apart.
In practice, self-reliance, cross-strait engagement, and US cooperation have existed side by side in Taiwan's security policy for decades. Asking voters to pick only one produces results that fit neatly into political narratives but capture only part of a much more layered public view.
The May 2026 survey adds a complication that neither camp chose to acknowledge. Fieldwork ran May 28 to 31, two weeks after Trump's Beijing summit with Xi Jinping. Non-aligned voters watching their primary security patron conduct independent diplomacy with the adversary had rational grounds to hedge toward engagement that week.
The post-summit timing may have pushed some voters toward engagement. Whether the survey captured a lasting preference or a reaction to a specific geopolitical moment is difficult to know. Both camps nevertheless cited the results as evidence of settled public opinion.
The exploitation runs symmetrically. The DPP pointed to the 44.9% who prioritised self-reliance and presented Lai Ching-te's proposed NT$1.25 trillion defence budget as reflecting majority opinion.
Other polling suggests a more divided picture. Independent voters, who make up roughly 32% of the electorate in current TVBS tracking, and represent the largest single bloc, were almost evenly split on the budget, with 48% supporting it and 41.3% opposed.
The KMT highlighted the 29.7% who favoured goodwill toward mainland China and used the result to justify its resistance to the budget. Yet a Brookings Institution survey of 1,195 Taiwanese voters conducted in April 2026 found that 54% supported increased defence spending when asked directly about the budget.
The difference between 29.7% and 54% suggests that responses can change considerably depending on how the question is framed. A voter may prefer engagement as a strategic approach while still supporting higher defence spending in practice.
This exploitation does not occur in a closed system. A randomised field experiment published in the American Political Science Review found that exposure to pro-Beijing media can increase scepticism toward the United States among less politically engaged Taiwanese voters. Brookings data suggests the scale of that divide. Among politically attentive respondents, 55% believed the United States would intervene militarily in a cross-strait conflict. Among those who paid little attention to politics, the figure was 21%.
These are also the voters most vulnerable to media influence and the hardest to capture accurately through single-choice survey questions. Both dynamics tend to affect the same segment of the electorate.
What the Fuller Data Actually Reveals
The Brookings trust data captures something the strategic-priority question cannot. Only 24% of Taiwanese voters overall regard the United States as a trustworthy ally. Among DPP voters, the cohort most favourably disposed toward Washington and toward increased defence spending, the figure reaches only 45.3%.
Taiwan's ruling party has built its entire security posture on an external commitment that fewer than half of its own voters find reliable. The INDSR forced-choice design obscures this almost completely, reducing US partnership to a named option that 11.8% of respondents selected as their single top priority in a direct comparison with two alternatives. That figure says nothing about the far larger number who would accept US cooperation as part of a layered approach while remaining privately uncertain that Washington's commitment will hold under real pressure.
When you set aside the forced-choice constraint and look across five surveys conducted between late 2025 and early 2026, a more coherent picture emerges. Different surveys have produced similar findings on defence willingness, pointing to a stable underlying preference.
TVBS November 2025 polling found 57% supporting increased defence spending, while 35% opposed it. Brookings April 2026 found 54%. INDSR August 2025 data placed over 60% of respondents willing to fight in an invasion scenario. The American Portrait survey from January 2026 found 58.7% willing to resist at all costs even without US military deployment.
The consistency across these figures is itself analytical information: it points to a stable floor of defence commitment that partisan framing consistently overstates as contested.
What the same population holds alongside this is instructive. The Global Views end-2025 survey found 56.6% wanting to increase cross-strait interaction and 60.3% agreeing that political leaders on both sides should meet. Read against the defence data, this is not inconsistency.
A public that would fight to defend Taiwan while preferring to avoid the conditions that make fighting necessary is behaving rationally under uncertainty. The forced-choice design records this as a contradiction when it is a conjunction.
Differences over defence spending appear to stem less from identity than from competing views about deterrence. Brookings found a strong relationship between support for the special defence budget and belief that increased spending reduces the likelihood of Chinese military action.
Support remains high across party lines among those who accept that premise. Among those who do not, opposition remains strong regardless of broader threat perceptions. At its core, the debate concerns whether greater defence spending can meaningfully strengthen deterrence. That is a question surveys can explore more directly than they do today.
Current survey design cannot capture that disagreement. Asking respondents to choose a strategic priority reveals little beyond their preferred label. It does not show whether they believe the logic behind it.
The Lai administration has largely defended the special budget through threat perceptions and alliance solidarity, arguments that resonate with voters who already believe stronger defence spending deters conflict. They are less likely to persuade the 47.9% of respondents who, in a separate INDSR question, doubted Taiwan's ability to defend itself. That figure was virtually identical to the 47.5% who expressed confidence.
The divide runs through the middle of the electorate, yet it is largely absent from the political narratives that emerge after each survey is released.
The Concrete Cost
The budget Taiwan's legislature passed on May 8, 2026 is a useful register of what these dynamics produce in practice. A 59-0 vote with 48 DPP lawmakers abstaining approved NT$780 billion, approximately US$25 billion, in supplemental defence funding. The figure is regularly cited as evidence that Taiwan's political system can, when sufficiently pressured, deliver on defence commitments.
Look at what the bill actually funds. The approved budget covers only US arms purchases. It cuts the T-Dome layered air defence system, Taiwan's domestic drone program, and broader defence industry initiatives that the Lai administration originally proposed. Taiwan's defence expansion now depends almost entirely on US supply chains and Washington's willingness to deliver.
In a blockade scenario, disrupting those supply chains would be one of Beijing's most practical options. That makes dependence on external deliveries a central vulnerability of the budget. At the same time, the bill increases reliance on a partner that many Taiwanese remain uncertain about in a crisis, while reducing investment in programs intended to strengthen Taiwan's own long-term capacity.
The polling created ready-made narratives that neither side had to defend in detail. The KMT's months of legislative obstruction aligned with the survey finding that many respondents preferred goodwill toward China. The DPP's push for a larger defence budget aligned with the finding that self-reliance ranked highest among respondents. Both sides could point to numbers that supported their position.
Lost in the debate was another pattern visible across multiple surveys: many Taiwanese support stronger defence and greater engagement at the same time, and their views on defence spending often depend on whether they believe it will actually deter China. The survey offered little insight into those questions.
It showed what respondents selected when asked to choose one priority, but not how they balanced competing goals or what trade-offs they would accept. That left ample room for political interpretation.
Conclusion
Taiwan's November 2026 local elections and January 2028 presidential race will unfold under the same survey infrastructure. Each campaign cycle will generate new INDSR surveys and new rounds of selective citation, because the institutional incentive sustaining this is durable. Single-choice polling serves parties seeking mandate, and parties are the ones who commission the polling.
The NCCU Election Study Centre, which conducts the INDSR surveys, has already shown it can use more advanced polling methods. In June 2025, INDSR commissioned the Centre's Pollcracy Lab to run a conjoint experiment that asked respondents to evaluate different defense policy packages. These packages varied by security conditions, funding sources, spending priorities, and allied expectations.
The method already exists and has been tested. What remains unexplored is how it might reshape the strategic-priority question itself. Do respondents who favour goodwill toward China also support higher defense spending? Do those who prioritise self-reliance see engagement as compatible with deterrence, or as something that weakens it?
A similar approach could provide clearer answers than current single-choice surveys. It would help reveal where genuine disagreements lie and where voters hold multiple preferences at the same time. That would give future budget debates a firmer empirical foundation than competing claims of public endorsement.
One figure stands out. Three-quarters of Taiwan's population does not regard the United States as a fully trustworthy ally. Many Taiwanese appear to support stronger defence and greater engagement at the same time, a pragmatic response to uncertainty about how a crisis might unfold.
Surveys often reduce that judgment to a single choice and hand the result to politicians eager to claim public backing. The public's views are more complex than the numbers imply. That disconnect deserves more attention than the familiar divide between blue and green.
Meng Kit Tang
Meng Kit Tang is a Singaporean freelance analyst and commentator who works as an aerospace engineer. He graduated from the S. Rajaratnam School of International Studies (RSIS), NTU, Singapore in 2025. His research interests include cross-Strait relations, Taiwan politics and policy issues, and aerospace technology.