Published by The Lawfare Institute
in Cooperation With
Tensions with North Korea have reinvigorated long-standing debates over when and how the United States should use military force. Legal experts have offered sometimes conflicting views on how domestic and international law limit potential military action against the Kim Jong Un regime—but expert legal opinion is only a small part of the overall policy debate surrounding the use of force.
Among other factors, the elected political leaders who make such decisions are likely to be sensitive to public opinion. Indeed, research has shown that public opinion can have a substantial impact on when and how political leaders choose to use force. If public opinion and relevant legal restrictions align, the two are likely to reinforce each other. Where they depart, however, political leaders may feel more license to push against legal boundaries, especially when other mechanisms for enforcing legal requirements are weak. And this is arguably the case with U.S. decisions to use force, as the executive branch’s control over the military leaves other actors and institutions with limited input on when and how to initiate the use of military force.
With this in mind, we set out to design a set of surveys to measure the extent to which public opinion—or, perhaps more accurately, the public’s moral intuition—aligns with legal considerations regarding the use of force. To do so, we used the online Google Surveys platform to ask a panel of more than 1,000 respondents whether they believed the United States would be justified in using military force in a number of hypothetical scenarios. We then changed aspects of those scenarios to isolate certain key factors and measure their impact on public perceptions of legitimacy. (Additional details regarding our methodology and its limitations are discussed in the final section of this post.)
Our results surprised us. The relationship between both domestic and international law and U.S. public views regarding the legitimacy of the use of military force is fairly robust. But introducing other variables complicated this picture, as did breaking down our results along certain demographic lines. Ultimately, some of these trends may pose significant challenges to the system of laws and institutions that regulate the use of force by the United States.
We’ll be discussing these results in a series of posts over the next several weeks. This first post, which is based on our “control” survey as discussed below, addresses the extent to which differences in certain key variables in the domestic and international legal analysis of the use of force impact public perceptions of legitimacy. In subsequent posts, we will examine the impact of other factors, including: the threat of nuclear weapons; the involvement of specific actors, specifically President Trump and North Korea; the demographics of the respondents; and, finally, the respondents’ partisan political affiliations.
Analyzing Public Opinion on the Use of Force
So how well do legal considerations regarding the use of force align with public perceptions of legitimacy?
To answer this question, we focused our analysis on three factors that are widely seen as critical in the legal analysis of decisions to use military force. For international law, we focused on the principles of necessity (i.e., whether a state has exhausted any available alternative means such as diplomacy or sanctions prior to using force for addressing a given threat) and proportionality (i.e., whether the damage caused by a use of force is proportional to the threat that it is addressing)—both of which are key elements of jus ad bellum, the international law governing when states may resort to the use of force. For U.S. domestic law, we focused on whether there was congressional authorization for a given use of force. Each scenario was otherwise framed as a situation of U.S. self-defense, circumstances in which the use of force is generally more easily justified under both domestic and international law. This allowed us to avoid other considerations that might bear on legality and legitimacy on non-self-defense circumstances.
Of course, the exact relationship between these factors and the legality of a given use of force is not always clean-cut. While most actors—including the U.S. Department of Defense—acknowledge that necessity and proportionality are key factors in determining whether a proposed use of force is consistent with international law, each requires fact-specific analysis that may be subject to substantial differences of opinion. The degree to which uses of force require congressional authorization under U.S. law is even more hotly contested, as some maintain that congressional authorization is required for any use of military force that is not responding to an imminent threat, while others—including the executive branch—contend that advance congressional authorization is only truly required for certain major conflicts. We attempted to avoid ambiguity by drawing sharp contrasts for each variable across questions. The resulting lack of nuance makes it difficult to translate the results of our survey into complexity-laden real-world scenarios. But it also lets us more easily gauge the extent to which public opinion is sensitive to our target considerations.
For each question in our survey, we asked respondents to indicate whether they believed the United States would be justified in using force against an unnamed foreign state if that other state were to take one of six actions, which were presented in random order. Respondents were free to select as many responses as they felt were appropriate, or could select “None of the above.” These actions were designed to measure necessity by presenting different levels of imminence in the threat posed, from “Attempt a military strike”—as the most imminent—to “Improve its military strike capability”—as the least imminent. As states have more and more opportunities to try other responses short of using force in response to less and less imminent threats, the use of force is generally considered to grow less consistent with the principle of necessity.
As reflected in the table below, respondents were more likely to view the use of force as justified in situations where the threat posed was more imminent. For example, 47.2 percent of respondents viewed the use of military force as justified where the foreign state had already attempted a military strike—a response that was statistically significant, meaning that there is 95 percent probability that we’d receive the same winning answer upon running the survey again—compared to only 7.9 percent if the foreign state simply improved its military strike capability.
This ranking of responses tracks the imminence of the threat posed by the different responses, from most to least. Moreover, this pattern repeated itself in the responses to all the questions that we asked as part of our survey. The only variation we observed was between “[a]ttempt a military strike” and “[n]one of the above”, which frequently changed places as the most popular response across questions. This consistent pattern shows that respondents were indeed sensitive to factors of necessity in a manner that roughly tracks legal considerations.
To measure the impact of proportionality, we changed our baseline question to inquire specifically about whether the United States would be justified in using nuclear weapons, not just military force. As the damage caused by nuclear weapons is likely to be perceived as substantially greater than that caused by conventional military force, respondents who are sensitive to proportionality concerns should be less likely to view the use of nuclear weapons as justified than they were to hold that view about the use of conventional military force. The results are illustrated below.
Comparing these results to those of the prior question underscores the significant differences between the two:
When respondents were asked about the United States’ use of nuclear weapons, “None of the above” was the most common response, given by nearly half (49.2 percent) of respondents—once again, a statistically significant winning result. Even with congressional authorization, a full 12.9 percent fewer respondents found that the United States would be justified in responding to an attempted military strike with nuclear weapons than conventional military force, and the response rates for the other remaining responses dropped significantly as well. While 34.3 percent of respondents view the use of nuclear weapons as a justified response to the most imminent threat presented—an attempted military strike—less than 20 percent of respondents felt that the use of nuclear weapons would be justified in response to actions that are less imminent than an attempted military strike. All told, these results show that the respondents’ views were also sensitive to considerations of proportionality.
Finally, we asked respondents variants of the above two questions in which the U.S. action in question explicitly lacked congressional authorization. The results with regard to the use of military force are shown below.
Once again, comparing these results to the results of our earlier question relating to the use of conventional military force with congressional authorization shows several differences.
Only 4.3 percent fewer respondents viewed the use of conventional military force in response to “Attempt a military strike”—the most imminent threat among the various responses—as justified when it lacked congressional authorization. But a full 8.2 percent fewer respondents felt using force absent congressional authorization was justified in response to “Take steps to initiate a military strike,” a threat that may not be deemed as imminent. There was a similar drop in the response rate for the other remaining responses as well. Interestingly, this arguably tracks expert views regarding the need for congressional authorization under U.S. law, where many see authorization as increasingly less necessary the more obviously imminent the nature of the threat.
We found similar results when the U.S. use of nuclear weapons without congressional authorization was at issue. The results for that question are shown below:
As before, comparing these results to those for the prior question relating to the use of nuclear weapons with congressional authorization shows some differences. However, these differences are lower in magnitude than the differences relating to the use of conventional military force.
Respondents were still less likely to choose any response except “None of the above” when a U.S. response had congressional authorization. But the declines in support for various responses were significantly less dramatic as compared to a U.S. response using only conventional weapons, ranging from 3.9 percent for “Take steps to initiate a military strike” to just 1.4 percent for “Improve its military strike capability”. Perhaps surprisingly, while respondents were somewhat sensitive to considerations of congressional authorization in relation to the use of nuclear weapons, the increase in support for the use of nuclear weapons with congressional authorization, as compared to without authorization, was less substantial than the increases in support for the use of conventional military force with congressional authorization as opposed to without.
It’s not clear why this is the case. Perhaps public opinion is simply less sensitive to questions of congressional authorization in the context of using nuclear weapons. This would run against the grain of most legal analyses, which generally find that congressional authorization is more important where U.S. uses of force are more severe in nature. (Notably, the United States gives the president operational control over the use of nuclear weapons, but this does not bear on the legality of that decision.) Another possible explanation may be is that there is a substantial overlap between those respondents who find uses of conventional military force less justified when lacking congressional authorization and those who find the use of nuclear weapons less justified regardless of congressional authorization.
Finally, one consistent feature across the responses to our questions warrants specific mention: the consistently high proportion of respondents who answer “None of the above” when asked whether the use of force would be justified, implying that the respondent did not believe the use of force would be justified in any of the situations presented. If taken at face value, this outcome is at times quite striking: for example, a full 32.5 percent of respondents responded that they do not believe that the United States would be justified in using congressionally authorized military force against another state even if that other state were to attempt an attack the United States, a scenario in which both domestic and international law accept the use of force as lawful.
But we are hesitant to accept the proposition that nearly a third of Americans are such committed pacifists. A more likely explanation is that certain respondents are hesitant to endorse the use of force in response to our sometimes inartful questions and loosely sketched scenarios, erring against the use of force when in doubt. Alternatively, given the text-heavy nature of the questions, respondents may have picked “none of the above” simply to move on to the next question quickly. This possibility would make the raw number of respondents who selected “None of the above” a less reliable indicator of actual public opinion. Changes in the response rate across questions remains revealing, however—as we will examine in future posts.
Methodology, Errors, and Limitations
We ran the above survey until we were able to secure weighted results for 1,000 respondents in the United States. The version of Google Surveys that we chose to use administers its surveys through “survey walls” for websites on a set publisher network. Individual internet users that wish to visit that website may complete the survey in order to gain access. Pursuant to Google Surveys policies, each respondent was informed in advance that the survey would ask about sensitive subjects and permitted to opt out, which 77.2 percent of respondents chose to do. There was also attrition as respondents proceeded through the survey. As a result, the final results came from an initial pool of over 11,000 potential respondents. Thanks in substantial part to this large sample size, our final results had a margin of error of no more than +/- 2.6 percent for each response, and often substantially less.
Of course, our survey still has several limitations worth noting. Due to limits on the number of characters we can use in the Google Surveys platform, our questions were at times imprecisely worded and often lacked detail. These ambiguities that may have led respondents to read the questions differently. Nor do we know what background knowledge the respondents may be bringing to the table. For example, in asking about congressional authorization, certain respondents may not have been aware of the relationship between different American political institutions, skewing the results.
Other limitations are inherent to the Google Surveys platform, as they have explained at length. Most notably, as a result of how the survey was administered, the population of respondents may disproportionately include internet users and individuals inclined to visit the websites within Google Surveys’ publisher network. Google Surveys weights the results so that they proportionally reflect a representative population of adult (i.e., 18 or over) internet users according to age, gender, and geographic region in the United States, all of which are inferred through respondents’ browsing histories and internet protocol (IP) addresses. Google Surveys also seeks to ensure that respondents do not come disproportionality from any one website in the publisher network. Nonetheless, these measures do not constitute an absolute guarantee that our respondents are truly representative of the broader American public.
Conclusion and Future Analyses
Thus far, we have focused our analysis on three key variables—necessity, proportionality, and congressional authorization—that are relevant to whether a use of force is likely to be found lawful under domestic and international law. In each case, we found that public perceptions regarding the legitimacy of the use of force are also responsive to these variables, often to a substantial degree. While this does not mean that legal arguments and public opinion will always align—especially in real-world situations more nuanced than the hypotheticals we employ—it should give hope to lawyers and supporters of the rule of law that their arguments may resonate intuitively with a substantial portion of the American public. What’s more, it should hopefully remind policymakers that compliance with domestic and international law is not just a means of avoiding (sometimes relatively weak) enforcement measures, but can itself be a road to greater public legitimacy.
But our study does not stop at the above three variables. Instead, we designed it to collect information on a number of other variables as well. (We also provided respondents with an open field in which they could identify other factors that had a bearing on their decision-making. As this field did not yield any usable results, however, we do not plan to discuss it further.)
In addition to the four questions discussed above, we asked variants of the above questions addressing yet another hypothetical scenario: what if a foreign state threatens the United States with nuclear weapons?
We were also curious as to how respondents’ views of certain real-world actors might affect their responses. While the “control” survey discussed above listed the U.S. decision-maker as a generic “United States” and did not identify the foreign state whose actions are at issue, we also ran a separate “Trump impact” version that identifies the U.S. decision-maker as President Trump and a “North Korea impact” version that identifies North Korea as the foreign state. And we ran a fourth “North Korea & Trump impact” version that implements both changes, providing a sense as to how respondents’ views of the two actors affect each other.
Finally, Google Surveys allows us to break down our results along several demographic variables, including age, gender, and geographic region. We also inquired about respondents’ partisan affiliations, providing us a window into how political groups’ approaches to these issues may vary. Examining how results varied across these different groups reveals several interesting differences in how different parts of the American population view questions regarding the use of force.
All of these results will be the subject of future posts in this series.