Public surveys about Taiwan's willingness to fight an invasion by China don't reveal as much as one might think.
As the world watches Russia’s invasion of Ukraine, pundits and politicians worldwide are again making a comparison to the situation in Taiwan. There are surface similarities: Taiwan, too, faces the threat of invasion posed by a larger and militarily powerful adversary. Yet few of these commentaries offered much in the way of informed understanding of the actual challenges facing Taiwan. Among the most problematic discussions concerns opinion polling around the so-called “willingness to fight” among Taiwan’s public in a war with China.
In December 2021, two organizations in Taiwan separately published opinion polls supposedly addressing whether Taiwanese public would be willing to “fight on the battlefield” should China invade. One organization’s poll had 62 percent of respondents say yes and 27 percent say no, while the other survey, with a slight difference in its wording asking whether “you or your family” would be willing to fight, found only 40 percent said yes and 51 percent said no.
This led to an ironic scene where Taiwan’s media, pundits, and international commentators cherrypicked whichever of the two polls that best suited their partisan narrative and declared either that the Chinese People’s Liberation Army (PLA) will have to fight an all-out “people’s warfare” should they invade Taiwan, or that Taiwanese society will surrender in a heartbeat at the first sign of war.
It is no surprise that the first poll was published by Taiwan Foundation for Democracy (TFD). The TFD is an organ of the Taiwan government funded with taxpayers’ money, but it is actually controlled by ruling Democratic Progressive Party appointees to conduct partisan messaging.
The other poll was done by Global View Monthly, a Taiwanese magazine which did not release details about its methodology. But the TFD survey came with a brief description about its sampling which revealed, interestingly, that it only surveyed respondents from Taiwan’s main island and excluded Kinmen and Matsu. These two outer islands of Taiwan are much closer to China and therefore the likely “frontline” if a war breaks out. This leaves one to question why the poll excluded them from the survey even though their residents are no doubt also citizens of Taiwan.
Any institutional biases aside, how could two polls conducted around the same time produce a 20-percentage point difference, much less on a question so widely assumed to be vital to the success or failure of Taiwan’s defense?
Issue Preferences vs. Behavioral Predictions
In the world of opinion polling, pollsters often aim to answer one (or both) of two questions. First, what are people’s preferences on a given issue of public interest? And second, how would people realistically act concerning a future event?
One example of the first type is a poll asking respondents if they support a particular public policy, such as lifting a restriction on certain imports. Whether the cabinet or parliament would actually consider the proposal is another matter, but the poll would tell us something about the distribution of public opinion on the issue. This leads to a more informed, evidence-based discussion on public policy.
In the second type of polls, pollsters ask respondents how they would behave in a given situation; for example, which political party or candidate they will vote for in the upcoming election. In order to conduct these surveys, pollsters must shrink the scope of the surveyed population from the general population to those participating in the event. For example, Taiwan’s constitution historically limited voting to adults aged 20 years or older. Because of this, our election polls used to exclude those younger than 20, as it did not make sense to include responses about voting behavior from non-adults who cannot vote. Many pollsters will also attempt to estimate how likely a person is to vote, thus gaining an even more specific prediction about voter turnout and the election result. With these tools in mind, pollsters can project a reasonably accurate picture of the likely distribution of people’s choices on the actual voting day.
However, elections are known events with a finite set of pre-determined choices. Asking respondents how they would act with regards to a hypothetical event is more difficult, as a pollster must be very careful in crafting the survey and interpreting the result. The more hypothetical a given scenario is, the less common understanding people would share concerning its objective reality, and the lower the predictive power of the poll.
War! Polls! What are they good for?
This problem of hypothetical scenarios is a real problem when it comes to questions of war and warfighting. The Taiwanese public, as is true for other societies that have been at peace for decades, has no experience with war. This means that the public has little if any understanding of what a war would entail before it actually occurs. In addition, given their varying experiences, people across demographic groups most certainly have diverse and unique conceptions of what a commitment to “fight on the battlefield” entails. Some might recall images of World War I trenches they’ve only seen in history textbooks, while others might presume the smallest contribution, such as participating in a bucket brigade, counts as “fighting.”
In any case, a very small proportion of Taiwan’s population would have a realistic chance of participating in combat in most conceivable China-Taiwan war scenarios. The vast majority of Taiwan’s population is not of military age, and their self-declared willingness (or unwillingness) during peacetime to “fight on the battlefield” has little impact on Taiwan’s defense planners or on the invading Chinese PLA.
If most Taiwanese are unlikely to be engaged in the fighting, who would be? It’s a surprisingly small slice of the population. Taiwan’s military currently has at most 188,000 active soldiers, or 0.7 percent out of a total population of 23.5 million. It might be tempting to imagine that Taiwan could mobilize the supposedly two million strong reserve force, but as I have explained in a previous report such a “reserve force” exists only on paper. For one thing, the military’s stockpile of spare service rifles could barely equip more than a few brigades, or about 10,000 to20,000 of reservists concurrently. It is also widely debated among Taiwan’s defense analysts how useful, if at all, these hastily organized infantry units armed with only small arms would be facing the onslaught of the modern PLA in a war expected to be conducted predominantly over air and sea with missiles and other means of long-range fire projection.
Even if such a poll were to survey only active service soldiers and military-age reservists, it would still be fatally flawed in the real world. By Taiwanese law, soldiers are expected to obey orders rather than to “choose” what to accept, and soldiers or reservists who fail to report to duty or order of mobilization are subject to criminal prosecution and punishment. A poll asking soldiers whether they would obey orders is not likely to be any more reliable than, for example, a poll conducted in a church after a Sunday mass asking the congregation whether they consider themselves good Christians.
Measuring the Right Data for the Right Question
In reality, a country’s ability to field an effective military in the modern age has more to do with its combined economical, technological, and human resources and the level of commitment its political system and leadership is willing to invest in it. This determines the quantity and quality of the troops and whether they would have functioning weapons, equipment, and ammunition as well as sufficient training, organization, and leadership personnel. As a defense researcher myself, my investigations have concluded that the most serious shortcoming of the Taiwanese military today is not a lack of morale or motivation among the general public, but a lack of political will to reform and reorganize the military in a way that can realistically deal with the ever-increasing threats posed by the PLA.
Many might find it disappointing that opinion polling cannot be a reliable indicator to gauge the extent to which Taiwan’s population would fight to defend themselves. But it is important to remember that all research methods, including public opinion surveys, have their limitations. As the United States’ own intelligence agencies have found, predicting the will to fight is extremely difficult—contrast US predictions of a long fight by the Afghan government and a rapid defeat of Ukraine, neither of which bore out in reality. It is always better to acknowledge these limitations than to make bold yet unreliable projections which could mislead the public’s understanding of the reality of war.