To What End? Policy Objectives and US Public Support for Political Warfare

Abstract

Research regarding military interventions suggests that the objective of an intervention (such as whether it is carried out for humanitarian or power-political ends) substantially influences public support. It is unknown, however, whether that holds for political warfare strategies such as electoral interventions. In a series of survey experiments, we evaluate how different objectives affect US audiences’ support for foreign electoral interventions, both overall and in relation to other forms of political warfare, such as sanctions and covert regime change. Humanitarian objectives elicit greater support than those carried out to stave off threats to US leadership or to promote US economic interests. Furthermore, religious identity affects support for interventions undertaken for humanitarian ends, with the congruence between the identity of respondents and victims substantially influencing support. Finally, support for partisan electoral interventions and other forms of political warfare varies in relation to each other and in relation to interventions’ objectives.

La recherche relative aux interventions militaires suggère que l'objectif d'une intervention (si elle sert des fins humanitaires ou politiques/de pouvoir, par exemple) a une incidence importante sur le soutien de la population. Néanmoins, l'on ne sait pas si la même relation s'applique aux stratégies de lutte politique comme lors d'interventions électorales. Dans une série d'expériences de sondage, nous évaluons l'effet de différents objectifs sur le soutien de publics américains aux interventions électorales à l’étranger, tant dans l'ensemble qu'en relation avec d'autres formes de lutte politique, comme les sanctions ou le changement de régime en secret. Les objectifs humanitaires recueillent un soutien plus important que les interventions dans le but de repousser des menaces pour le leadership des États-Unis ou de promouvoir les intérêts économiques américains. En outre, l'on observe une relation entre l'identité religieuse et le soutien aux interventions à visée humanitaire, la congruence entre l'identité des personnes interrogées et des victimes ayant un effet conséquent sur le soutien. Finalement, le soutien aux interventions électorales et à d'autres formes de lutte politique varie en fonction des personnes concernées et par rapport aux objectifs des interventions.

La investigación en materia de intervenciones militares sugiere que el objetivo de una intervención (por ejemplo, si se lleva a cabo con fines humanitarios, políticos o de poder) influye sustancialmente sobre el nivel de apoyo público. Sin embargo, se desconoce si eso también se aplica a las estrategias de guerra política, como son las intervenciones electorales. Utilizamos una serie de encuestas, las cuales nos permiten estudiar la forma cómo los diferentes objetivos influyen sobre el apoyo del público estadounidense con relación a las intervenciones electorales extranjeras, tanto en general como en relación con otras formas de guerra política, tales como las sanciones y el cambio de régimen encubierto. Los objetivos humanitarios suscitan mayor apoyo que los que se llevan a cabo con el fin de evitar posibles amenazas contra el liderazgo de EE. UU. o para promover los intereses económicos de EE. UU. Además, cabe destacar que la identidad religiosa también influye sobre el apoyo a las intervenciones emprendidas con fines humanitarios, y que la congruencia entre la identidad de los encuestados y la de las víctimas influye sustancialmente sobre este apoyo. Por último, concluimos que el apoyo a las intervenciones electorales y a otras formas de guerra política varía entre ellas y con relación a los objetivos de las intervenciones.

Introduction

In 2018, Fox News host Laura Ingraham interviewed former Central Intelligence Agency (CIA) director James Woolsey about investigations into Russia's role in the 2016 US election. She asked Woolsey if the United States had “ever tried to intervene in other countries’ elections.” Woolsey admitted that it had “probably” done so in Italy and Greece in the past. Ingraham pressed him, asking “whether the U.S. does it now, mess around in other countries' elections.” Woolsey mumbled and grinned approvingly before stating, seriously and clearly, that the United States undertakes such interference “only for a very good cause” (“The Ingraham Angle” 2018). Woolsey's evasion and admission were both telling. As Woolsey presumably knows, since the Second World War, the US government has frequently intervened in other countries’ elections (Levin 2020, 2024). Yet Woolsey's initially cagey response underlined that even a former CIA director might hesitate to endorse such interventions publicly. After all, transgressing a core principle of democratic self-determination (Finnemore 2009, 79) might seem to require, as Woolsey hinted, a very good reason.

What, if anything, does the American public see as “a very good cause” that could justify partisan electoral interventions? The issue is not merely academic. Policymakers who engage in electoral interventions must at least consider whether the public will view their justifications as compelling. If one envisions electoral interventions as mostly, or even intrinsically, covert or clandestine, this may seem surprising. Yet that impression is misleading. More than 60 percent of electoral interventions in the post-Cold War era have been conducted overtly (Levin 2024). Furthermore, there is always a chance that at least some secret electoral interventions will become public. Either way, officials must be ready to justify these actions.

This paper examines two ways that a “good cause” could affect public support: first, how the objective for which the electoral intervention is ostensibly undertaken could influence support for election meddling and, second, how the purpose of an intervention may affect public evaluations of election meddling relative to other tools of political warfare. Although scholars have long studied the substitutability of foreign policies, electoral interventions, and public support for various foreign policy strategies separately, none have compared how public support for electoral interventions stands relative to other tools of political warfare. By investigating these topics, we thus contribute to the study of foreign policy and public opinion as well as the study of electoral intervention and political warfare.

Our first track explores how the objectives of an electoral intervention could influence public support. In the interview, Woolsey offered some clues about what US officials viewed as potentially convincing: promoting the “interests of democracy” and US national interests, such as stopping the spread of Communism. Recent research supports his claims, as Americans offer greater support for electoral interventions both to preserve another country's democracy and to counter rival great powers’ interventions (Levin and Musgrave 2023; see also Brands 2024). The literature on other forms of foreign policy interventions, including the use of military force and foreign-induced regime changes, suggests that other policy objectives, like a humanitarian goal or preserving US hegemony, might also increase support for election meddling. Yet it is worthwhile to test these hypotheses rather than assuming they extend to this domain. Earlier findings may not apply to electoral interventions. Unlike military operations, partisan election interventions do not usually raise the prospect of American casualties, for example. Consequently, public opinion might be less sensitive to objectives for a coercive but nonkinetic technique like an electoral intervention than for potentially costlier military interventions. On the other hand, the public may be uncomfortable with the skullduggery and intrusions that interventions like election meddling require. If the public sets higher ethical standards for electoral interventions compared to other, less underhanded methods, the objective might have to be particularly worthy to win public approval. Understanding whether the public's evaluations differ along such dimensions constitutes a major theoretical question, and one unexplored by previous research on electoral interventions.

A second way to explore the meaning of Woolsey's “good cause” phrase would focus on how the objective of a policy influences the public's preferences among alternative means of conducting interventions—that is, whether a “good cause” implies a good reason to use an electoral intervention rather than another means of accomplishing a goal. Electoral interventions form part of a broader portfolio of potential ways to accomplish US goals. Scholars have long identified policy instruments and strategies as being substitutable—that is, different policies that could accomplish the same ends vary in their tradeoffs of costs, benefits, and capabilities (Most and Starr 1989). The recent intensification of great-power competition has led to a renewed interest in particular forms of statecraft, which George Kennan (1991, 302–5) called “political warfare”: overt or covert operations short of war to achieve national objectives. This capacious category includes electoral interventions, sponsoring coups, fostering insurgencies, and sanctions. Investigating how the public evaluates the substitutability of electoral interventions relative to other forms of political warfare is thus particularly important.

One implication of theories of policy substitutability is that a given objective could constitute a good cause for using a particular tool but that another objective might incline policymakers, and the public, to favor a different tool. Yet earlier research on both substitutability and political warfare has neglected how the public views electoral intervention. In particular, it has not analyzed how the public evaluates election meddling compared to potential substitutes or how different humanitarian and hegemonic objectives—prominent in the theory and practice of other interventions—might influence support for electoral interventions relative to other options. As the United States frequently employs many forms of political warfare, it is urgent to understand how facets of political warfare, particularly their objectives and means, affect public support for strategies like electoral interventions. Investigating how the context and goals of an intervention influence preferences over which tools to employ could help clarify when we should expect states (or at least the United States) to use one tool instead of another.

Both questions carry substantial academic interest and policy relevance. The United States was recently and (in)famously a repeated target of election meddling by Russia. Despite (or, perhaps, because of) this experience, some prominent US commentators and politicians have called for the US government to conduct such interventions itself (Robinson et al. 2019; Brands 2020, 2024). These endorsements likely reflect continuing, less visible preferences of parts of the US government and the foreign policy establishment (Brands 2020; Inboden 2022). Such preferences occasionally surface. In an interview during the 2020 primaries, for instance, Joseph Biden stated that the United States could “embolden [elements of the Turkish leadership] to be able to take on and defeat Erdogan. Not by a coup, not by a coup, but by the electoral process” (New York Times 2020). Not only is a greater understanding of the public's evaluation of foreign policy valuable in its own right, but it is particularly important if, as recent research has demonstrated, the foreign policy decision-making of elites (including elected officials) can be influenced by public opinion (Lin-Greenberg 2021; Chu and Recchia 2022).

This paper reports results from three survey experiments exploring how the objectives of potential interventions affect how the American public evaluates US partisan electoral interventions and other forms of political warfare. We find that some “good causes” did indeed generate more support for election meddling. Humanitarian objectives increase support for interventions compared to a baseline, as do power-political objectives (albeit more weakly). Not all humanitarian interventions are created equal: their effect on public support is strongly affected by respondents’ identification with the victims on whose behalf the intervention is conducted. Similarly, objectives influence approval for political warfare more generally, with interventions undertaken for humanitarian or economic ends eliciting greater support than those that aim to preserve US leadership in a region. When we turn to substitutability among intervention forms, we find that electoral interventions are usually preferred over forceful methods of political warfare, such as coups, as well as some other coercive but nonviolent policy options. We also find evidence that the objective of an intervention influences relative support for different techniques. A humanitarian goal, for example, produces substantial increases in support for using an electoral intervention and still larger increases for sanctions and coups.

Overall, these findings sum to a key contribution: we demonstrate that objectives matter for electoral interventions and political warfare—but not in the same way for all audiences and not to the same extent for all forms of interventions. We also demonstrate evidence of substitutability among different foreign policy strategies, as changes in relative approval are not parallel for all options.

Objectives and Public Support for Political Warfare

We begin by briefly surveying earlier work investigating how the objective of a foreign policy action influences public support. Earlier researchers have demonstrated that the principal policy objective of a military intervention substantially influences public opinion (Jentleson 1992; Jentleson and Britton 1998; Herrmann, Tetlock, and Visser 1999; Eichenberg 2005; Kreps and Maxey 2018). In particular, Americans are likelier to support the use of force against other countries to achieve humanitarian objectives or to resist undesired behaviors like aggression than to promote regime change or maintenance. Some researchers have found evidence that objectives matter for nonkinetic strategies as well. There is some evidence of significant US public support for using sanctions to punish or prevent human rights violations (Nielsen 2013; Peksen, Peterson, and Drury 2014). Heinrich and Kobayashi (2020) find that Americans prefer providing foreign aid to a “nasty” country when doing so yields instrumental benefits, especially on counter-terrorism efforts. In a partial contrast, Doherty et al. (2020) find that Americans are significantly less likely to support US aid aimed at enhancing a partner's military capability than promoting its economic development or improving population health. Chung et al. (2022) find that emphasizing either buttressing American leadership/hegemony or helping foreigners suffering from infectious diseases, but not emphasizing an economic goal, can increase support for the US government to match a rival's foreign aid contributions. Bugden and Brazil (2024) find that promoting international cooperation rather than rivalry increases support for providing foreign climate aid. Yet even though electoral interventions have been a major tool of US foreign policy in the postwar era, no study has explored how policy objectives affect support for electoral interventions.

The earlier literature is largely limited in another way: it neglects the role of substitutability. Many earlier studies of foreign policy and objectives have focused on a single type of intervention or political warfare, such as sanctions or aid, rather than investigating how different objectives could change relative support for strategies. Since Most and Starr (1989), many theorists have recognized that decision-makers rarely face a binary choice between doing nothing or employing a single policy option. Instead, policymakers can choose from a range of plausible policy substitutes with different combinations of costs, benefits, moral tradeoffs, and so on. Theorists have identified how many factors affect the ability of leaders to substitute among different policy options, including the preferences of the public (Milner and Tingley 2011), making public opinion potentially of great interest to understanding substitutability. The results of empirical analyses of substitutability have been mixed, due partly to the empirical difficulty of collecting data and specifying the range of options (Diehl 1994; Morgan and Palmer 2000; Starr 2000; Clark, Nordstrom, and Reed 2008; Fariss 2010; Salehyan 2010).

Experimental designs could help overcome many of these challenges, but they have been employed rarely to study substitution. One study examines the US public's evaluation of a hypothetical president's choices in response to an aggressive country that invaded its neighbor after the president issued a threat to use US ground troops to stop the invasion (Lin-Greenberg 2019). Respondents were significantly less likely to punish a president who imposed sanctions or conducted airstrikes rather than send US ground troops. Leal and Musgrave (2023) examine how varying levels of cyberattack severity and other conditions influence the US public's preferences regarding retaliation choices. They find an intuitive link between higher casualties and support for more severe responses from a constant list of potential options. Although generative, papers such as these leave many questions unanswered, particularly regarding how varying the objectives of such interventions could affect both the absolute and relative support for policy tools.

Studying partisan electoral interventions offers a way to advance both lines of research. Over the past several years, a recent wave of scholarship has improved our understanding of partisan electoral interventions’ dynamics. This work has largely followed two paths. One examines the causes of such interference (Bubeck and Marinov 2019; Levin 2020; Bubeck et al. 2022). The other investigates the effects of such meddling, from immediate effects on the target's electoral results (Levin 2016, 2020; Jamieson 2018; Walter et al. 2018), to their medium-term effects on the target's democratic regime and welfare (Shulman and Bloom 2012; Levin 2019; Tomz and Weeks 2020; Bush and Prather 2022; Justwan, Baumgaertner, and Curtright 2022; Palsson 2022), to longer-term effects, such as the willingness of the target to cooperate or trade with the intervener or third parties’ inclination to ally with the target (Bush and Prather 2020; Levin 2021; Goldsmith and Horiuchi 2022; Palmer 2023). A new third direction has begun assessing how intervening states’ publics view such interventions. Levin and Musgrave (2023) found that Americans, while overall open to the use of electoral interventions abroad, are more likely to support US electoral interventions that promote both democracy and US national interests. Levin and Musgrave (2021) found that Americans usually prefer overt to covert forms of US electoral interventions and oppose the use of inherently objectionable methods of intervention (such as spreading fake news about the undesired candidates).

There is still much to learn about how the public evaluates partisan electoral interventions. Earlier researchers did not explore how major policy objectives frequently discussed in the literature (like humanitarian or economic rationales for interventions) influenced support for electoral interventions. Nor did they compare electoral interventions to other tools, such as instigating coups or imposing sanctions. Yet studying political warfare could help us understand how the public evaluates nonkinetic interventions given varying objectives and potential policy substitutability because it constitutes a set of interventions with exactly the variations in tradeoffs that should prompt different levels of support. Crucially, these variations affect many dimensions. Electoral interventions, for instance, are neither as intrusive as supporting coups nor as comparatively morally unobjectionable as foreign aid. Investigating how the public evaluates electoral interventions relative to other forms of political warfare could thus also advance our understanding of policy substitutability, an issue of substantial theoretical—and policy—interest.

Theory

In this section, we describe arguments related to public opinion and electoral interventions. We describe arguments related to two major categories of objectives for which the US government may employ partisan electoral interventions, humanitarian goals, and the preservation of hegemony as well as how they may choose among substitutable policy forms.

Pursuing Humanitarian Objectives

Humanitarian purposes have become a prominent justification for interventions in the contemporary era. US audiences are substantially more likely to approve of interventions undertaken for humanitarian reasons than other goals (Jentleson and Britton 1998; Eichenberg 2005; Kreps and Maxey 2018; Kiratli 2024). The category is broad: Jentleson and Britton (1998, 399–400) define a “humanitarian intervention” as “provision of emergency relief through military and other means to people suffering from famine or other gross and widespread humanitarian disasters,” but their specific examples include the use of airstrikes in Bosnia to prevent mass killings. The import of the objective, however, is more easily stated: All else being equal, a large share of the US public does not want foreign nationals to suffer significant crimes against humanity or severe deprivations and mass deaths due to disasters. Much of the US public even believes that the US government has a moral obligation to stop or prevent atrocities. With a (perceived or actual) post-Cold War decrease in the cost of US kinetic operations abroad, acting on that general preference has been perceived as feasible by much of the US public since the early 1990s (Finnemore 2013; Kreps and Maxey 2018).

After the Cold War, the United States not only took part in disaster-relief missions but also regularly conducted military interventions for humanitarian purposes, from establishing no-fly zones in Kurdish areas of Iraq to intervening in Libya in the early 2010s. Activists, legislators, and officials invoked humanitarian discourses and evidence of major atrocities to lobby for those interventions and others that did not take place (Murdie and Peksen 2013). Indeed, humanitarian justifications for the use of force are so attractive that US presidents universally invoke such reasoning (Maxey 2020).

Although perhaps most prominent in justifying kinetic operations, humanitarian objectives have also supplied prominent justifications for electoral interventions. The US intervention in the 2000 Yugoslav election aimed to remove Slobodan Milosevic partly because of fears he would again conduct humanitarian atrocities (Shimer 2020, 9). One ethicist has even argued that preventing humanitarian atrocities justifies election interventions despite their other normative drawbacks (Fabre 2018). These observations lead to H1:

 

Yet the story is not so simple. Humanitarian practice is not always as universalist as humanitarian rhetoric. When humanitarian interventions first emerged in nineteenth-century Europe, “wide-scale and very visible atrocities against Christians” could put interventions “on the agenda of major powers” (Finnemore 2013, 60), but atrocities committed against Muslims elicited little condemnation (and sometimes instead even produced approbation) by European states. Similar dynamics may influence US public opinion toward humanitarian interventions today (Grillo and Pupcenoks 2017). The US public is significantly more willing to use military interventions to aid Christian victims than Muslim ones (Hanania and Trager 2021; Chu and Lee 2023). Americans are also significantly more willing to give foreign aid to Christian-majority countries then to Muslim-majority countries (Blackman 2018) and to support going to war with Muslim countries than Christian ones (Lacina and Lee 2013). In contrast, earlier work finds that the effects of the victim group's race on support for humanitarian interventions to be either nonexistent or much weaker and highly conditional (Hanania and Trager 2021; Chu and Lee 2023).

One prominent explanation for this selectivity concerns in-group identification (Grillo and Pupcenoks 2017; Chu and Lee 2023, 4–6). In this paper, however, we are less interested in explaining the determinants for humanitarian intervention support than whether humanitarian impulses universally or selectively influence support for partisan electoral interventions and other tools of political warfare. Given the repeated finding that religion, rather than other potential identities (such as race), is particularly salient in this domain, a religious manipulation seems best suited to test the limits of universality (Mutz 2021). Such a treatment is also particularly plausible because many countries or regions where humanitarian interventions have taken place or have been proposed, such as Kosovo and the Central African Republic, contain substantial populations of both Muslims and Christians. Accordingly, we present H1.1:

 

Preserving US Hegemony

Another commonly invoked “good cause” for US political warfare has been the preservation of US hegemony (or, more politely, “leadership”). During the Cold War, some regions have seen such “good causes” frequently invoked: Latin America may have seen a large share of covert foreign-imposed regime changes (FIRCs) because US leaders feared that the advent of a hostile leader in any Latin American country, regardless of its size, could threaten its position in the region and globally (O'Rourke 2018,115–6). Administrations from both parties have pursued such goals throughout the post-Cold War era (for instance, White House 2022; 2017). Contemporary threats to US hegemony may emerge as other countries entertain offers from alternative patrons, undermining the infrastructure of US leadership (Cooley and Nexon 2020). Because rival parties or coalitions may each favor a different potential patron, an electoral intervention carried out to support US-aligned parties may prove particularly tempting.

Preserving hegemony may be perceived as valuable on its own terms. According to some international relations scholars, being a hegemon makes a country significantly more secure than other countries in the international system (Mearsheimer 2001). Given that some Americans possess “folk realist” beliefs about the nature of the international system (Drezner 2008; Kertzer and McGraw 2012), they may also intuitively see hegemony as providing substantial benefits to the United States and thus see potential threats to US hegemony as being a threat to US national security as well.¹

 

Willingness to support meddling may vary based on the perceived strategic importance of the region in question, in line with some international relations scholarship (Gholz, Press, and Sapolsky 1997; Posen 2014). For example, the US public might perceive US hegemony in some areas as especially important due to major defense commitments, substantial economic relations, or other factors. Accordingly, the US public may be more willing to intervene in elections to counter potential threats to US regional hegemony when a region is perceived as strategically important than when it is not, separate from the degree of the threat posed to US leadership in that region.

 

Exploring the Substitutability of Strategies

US decision-makers contemplating political warfare can choose from many strategies. Some options are violent. One violent option is a covert regime change operation, which the US government employed in Chile between 1970 and 1973 and Iran in 1953 (Gustafson 2007; Abrahamian 2021). Other options do not involve violence. If the “wrong” side wins an election, the US government could impose post-election sanctions to coerce them to change their policies, as it did after Juan Perón won the 1946 Argentine elections (Escudé 1989). Alternatively, the United States could seek a modus vivendi if the candidate or party it opposes wins. Those negotiations could include significant concessions, a tack favored by Washington after its unsuccessful intervention in Brazil's 1955 elections.²

It seems plausible that the US public's preferences over which foreign policy tools should be used are separable from their judgment regarding whether the situation requires “doing something.” It further seems plausible that contextual factors, including the objective of the intervention, could affect preferences over which tool to employ. Earlier work suggests how the public's evaluation of electoral intervention might vary relative to potential substitutes. For example, prior research has found that electoral interventions designed to defend democracies, especially longstanding ones, elicit much greater support from American respondents (Levin and Musgrave 2023), in line with earlier findings about a general US public preference for supporting democracy (Smith 2012). Given this apparent preference for preserving democratic institutions, we expect Americans to prefer nonviolent (but still coercive) options over violent ones. As the targets of an electoral intervention are, definitionally, all democratic regimes, the tradeoffs involved in choosing some other forms of political warfare could be severe. For example, a covert FIRC would involve toppling a democracy and perhaps even shedding the blood of democratically-elected officials and democratically-minded civilians. Many Americans might find such a FIRC repugnant even if undertaken for objectives they would otherwise support. The exposure of US government involvement in plots against the democratically elected Chilean president Salvador Allende led to a major outcry by the American public against the US government's conduct of such activities.³ A partisan electoral intervention that (in the short term, at least) avoids a democratic breakdown or such bloodshed in the target is likely to be seen as far more preferable.

Second, we expect respondents to favor other nonviolent alternatives over partisan electoral interventions. Neither sanctions nor policy concessions infringe on the right of the targeted country's public to freely choose its own leader, unlike partisan electoral interventions. Furthermore, the US public tends to approve of other coercive but nonviolent strategies, like sanctions, that do not involve such transgressions (Jentleson 2022). This produces H3:

 

It seems plausible, however, that respondents’ preferences over substitutions may be influenced by the objectives of a potential action. The moral imperative of a humanitarian objective, for instance, may outweigh concerns about intervening in an election. Ideas like the (Western-inspired) Responsibility to Protect norm suggest a strong (Western) belief in a prerogative, even an obligation, to violate the sovereignty of other countries to prevent severe human rights violations (Fabre 2018). Extending that logic to substitutability raises the prospect that offering policy concessions to leaders who severely violate human rights (or are expected to do so) may be morally repugnant—even if the same option would be viewed more favorably in the pursuit of, say, security interests. As a result, electoral interventions may be preferred by respondents over other nonviolent options.

 

Methodology

Like other recent studies (Bush and Prather 2020; Tomz and Weeks 2020, 2022), we employ survey experiments to test our hypotheses.⁴ We chose to explore Americans’ views regarding foreign electoral interventions with multiple vignettes rather than alternative approaches, such as employing conjoint experiments, because of the early stage of research into electoral interventions, the resources that conjoint experiments require, and the fact that a conjoint experiment can only explore one type of scenario at a time. Furthermore, using multiple experiments may increase external validity (McDermott 2011, 37–8). As the majority of electoral interventions in the post-Cold War era have been conducted overtly (Levin 2024) and were even occasionally subjects of internal debate within the United States while they were being conducted,⁵ we concluded that vignettes describing interventions that had not yet been carried out would be a reasonable way to examine the determinants of US public attitudes.

One consideration in designing our experiments concerned whether to incorporate retaliation by a targeted country or, instead, to treat US foreign electoral interventions and other tools as instances of asymmetrical international relations (Musgrave 2019), in which strong and weak countries possess dissimilar abilities and a target would thus be less likely to be able to retaliate. Empirically, post-election retaliation by the target against a foreign country that has meddled in its elections is historically rare and, according to some scholars, may have never happened after a US electoral intervention (Levin 2020). Because it appears that the United States can intervene in weaker countries without fear of specific retribution (even if it may suffer diffuse reputational or other costs), we do not incorporate retaliation here. We do note, however, that a comparative lack of external constraints makes domestic politics potentially more important as a determinant of when electoral interventions may take place, thus rendering this study still more valuable.

We preregistered our analysis plans with the Open Science Foundation. Table 1 lays out which experiments test which hypotheses.

Experiment 1: Humanitarian Objectives

Experiment 1 was administered to 1,515 adult US citizens on the Prolific Academic online platform in early November 2022. Prolific Academic is designed specifically for academic research and features subject pools fully aware of their possible recruitment into academic studies (rather than into commercial surveys or various microtasks). It also employs a payment method designed to discourage low-quality submissions (Palan and Schitter 2018). Prolific respondents have been found to be significantly more attentive, careful, honest, and reliable in their responses than respondents on many other widely used survey platforms (Eyal et al. 2021). We used quotas and targets to ensure diversity on several dimensions, including gender, racial identification, age, education, and party identification. Full details are available in the Online Appendix (Table A3).

The experiment manipulated the objective of a potential electoral intervention. Respondents read a brief vignette. Two-thirds of respondents were informed the goal of an intervention was stopping a potential human rights atrocity (the humanitarian condition) while one-third were told the goal was stopping a potential change in the country's foreign policy direction from an ally to an opponent (a national-interest condition). We employed the latter condition as a baseline treatment since it proved difficult to create a plausible reason why the United States would intervene without some rationale, making a pure control (with no motivation) implausible. Within the humanitarian variations, respondents were also randomly assigned to be told the victims were either Christian or Muslim. As described above, we varied the religious identity of the victims to test whether support for a humanitarian intervention was selective or universal.

The scenario began: “Imagine that it's several years from now. A foreign country will soon hold elections to choose its government. There are two parties, the Turquoise Party and the Orange Party.” The scenario then informed respondents of independent observers' assessments of the policy of the Turquoise Party, specifically that if elected the party would

• Humanitarian: enact highly discriminatory policies against the country's [Christian/Muslim] minority.

• Interests: enact a new foreign policy direction that would turn the country from a US ally to a staunch opponent of the United States.

The scenario further stated that experts feared the consequences of a Turquoise victory would include

• Humanitarian: “using the military and police to force millions of [Muslims/Christians] to leave their ancestral homes and flee to other countries. Thousands could be killed.”

• Interests: damage to US foreign policy interests.

By contrast, the scenario stated, “independent observers believe that the Orange Party, if elected, would not enact any such plans.” To indicate that intervention could be decisive, we informed respondents that “Polls indicate the election will be close.”

Our primary dependent variable was support for electoral intervention generally. We asked, “Would you approve of the US government taking actions to reduce the Turquoise Party's chances of winning the elections?” Respondents could choose from a seven-point scale (strongly disapprove to strongly approve).⁶

Support for Intervention by Objective Type (H1)

Figure 1 displays the results of a regression analysis of the primary dependent variable, which serves as a test of H1 (respondents will support an electoral intervention to prevent crimes against humanity). The raw distribution of support demonstrates substantially higher levels of support for interventions to prevent a humanitarian disaster than for raison d'etat. Further, a plurality of respondents supported US interference in pursuit of the humanitarian goal in both the Christian and Muslim victim conditions. The second part of the figure reports the results of several ordinal logistic regressions using pre-registered specifications. We find that the humanitarian condition (pooling both Christian and Muslim victims) elicits substantially stronger support than the national interest condition. As the final part of the figure demonstrates, these effects are substantively large, with support for approval increasing by about 13 percentage points in the humanitarian condition compared to the interest-based condition. We thus find full support for H1, that respondents will be likelier to support interventions meant to prevent crimes against humanity. (Full results are given in Table A4 of the Online Appendix.)

The Online Appendix (Figure A2) shows that these results broadly hold across (pre-registered) subgroups, including by education, employment status, and gender. Respondents with highly hawkish predispositions (as measured by an index of responses to common attitudinal measures) displayed no differences between interest-based and humanitarian treatments, while doves and those in the middle were swayed by the humanitarian treatment. The effect was weaker among Republicans and strongest for Democrats.

Selective Support for Humanitarian Intervention (H1.1)

We next test H1.1, which proposed that respondents would be likelier to support interventions to protect in-group members (Christians) rather than out-group members (Muslims). Figure 2 shows support for this claim. On the surface, the coefficient for “All” respondents shows no statistically significant difference in support for interventions between respondents in the Christian and those in the Muslim treatment groups. Although our study cannot exclude the possibility of small differences between those evaluations, our study is sufficiently powered to detect a difference large enough to be substantively important.

When we split our samples by religious identification, however, we find evidence that this apparent null effect for all respondents obscures significant heterogeneities. In particular, we find that Christian-identifying respondents were substantially less likely to support intervention on behalf of Muslim victims than Christian ones.⁷ We also find that non-Christian respondents were more likely to support interventions on behalf of Muslim victims. (Full results are given in Online Appendix Table A5.) Christian respondents’ support for an electoral intervention on behalf of Muslim victims is predicted to be 10.9 percentage points percent lower than for Christian victims, while non-Christian respondents would support an intervention for non-Christian victims by 9.4 percentage points more than for Christian victims.

Further tests (Figures A3 and A4 in the Online Appendix) demonstrate that it is Christian identity, rather than religious attendance or importance, that explains these results, in line with earlier research (Taydas and Olson 2022). These results partly confirm the findings of Chu and Lee (2023) that Christian Americans prefer to intervene to support Christian victims. Our findings partly disagree with theirs, however, as their findings indicate a general preference rather than one concentrated among Christian-identifying respondents.⁸

Substitutability and Strategy Preferences (H3 and H3.1)

Other hypotheses proposed ways of exploring how partisan electoral interventions could be substituted for other forms of interventions. Specifically, H3 proposed that respondents will prefer electoral interventions to violent options and nonviolent options to electoral ones and H3.1 proposed that humanitarian objectives will lead to a preference over other nonviolent alternatives. To test these hypotheses, we asked our respondents to rate their approval or disapproval (7-point scale, from strongly disapprove to strongly approve) of five options:

• An electoral intervention: Contribute funds to the Orange Party election campaign (“Contribute”).

• A pre-election concession: secretly negotiate a compromise with the Turquoise Party on another issue in exchange for them not carrying out their policies if elected (“Concession Before”).

• A coup: If the Turquoise Party wins, secretly encourage a coup to remove them from power (“Encourage a Coup”).

• A post-election concession: If the Turquoise Party wins, offer its leader a major policy concession they desire in return for them not carrying out their other plans (“Concession After”).

• Post-election sanctions: If the Turquoise Party wins, publicly threaten to impose severe economic sanctions if they carry out their policies (“Threaten”).

The raw results of Figure 3 show some evidence in support of H3. Coups are the least widely supported alternative in both the humanitarian and interest conditions, while post-election sanctions were the most popular in each arm. Yet respondents did not view all nonviolent options more favorably than electoral interventions, as the electoral intervention (“contribution”) was more popular than concessions. (The electoral intervention was the most popular option in the humanitarian condition.) H3.1 conjectured that respondents would prefer electoral interventions to other forms when the objective was humanitarian. The raw results contradict this hypothesis, as sanction threats after an election but before the humanitarian disaster received the highest levels of support.

Figure 3 also reports regression results. There were no statistically significant differences in support for pre- or post-election concessions, while there were relatively large increases in support for contributions, coups, and sanctions in the humanitarian condition compared to the interests condition. Interestingly, the single largest change came from the drop in “Strongly disapprove” regarding coup promotion in the humanitarian compared to the interest-based scenario, even as the overall level of support for this strategy remained low. These results partly support H3.1, as electoral interventions were preferred over pre-and post-election concessions (and saw a substantial increase in support in the humanitarian rather than the interest condition) but, in contrast to H3.1, (nonviolent) sanction threats are more supported than electoral interventions.

Experiment 2: Regional Hegemony

Experiment 2 was fielded on the pre-election wave of a 1,000-respondent module of the Cooperative Election Study (CES, formerly the Cooperative Congressional Election Study or CCES) during late September to late October 2022. (Full details about the population are available in Online Appendix Table A1.) It explored what conditions affect US public support for interventions designed to preserve US hegemony (or “leadership”). Respondents read the following passage:

Imagine that it's several years from now. A/an [REGION] country will soon hold elections. Its two main parties are the pro-American BLV Party and the anti-American PLQ Party. The country has long been friendly to the United States. Independent experts believe that a victory for the PLQ Party would turn its foreign policies strongly against the United States. However, experts believe that there is a [THREAT] chance that a victory for this party could lead the country's neighbors to turn against the United States and end U.S. leadership in the region. By contrast, a victory by the BLV Party would not have those effects. Polls indicate the election will be close.

The threat treatment let us test H2, regarding the relationship between threats to US leadership and support for an intervention. The treatment could take one of two conditions: high or low chance. The threat treatment manipulated whether it was likely or not that such a challenge would come to pass, mirroring the finding that electoral interventions to counter possible threats to democratic regimes found to be significant by Levin and Musgrave (2023).⁹ Note that this treatment essentially differs from the interest-based treatment in Experiment 1, as the treatment concerns the likelihood of a so-called “domino effect” throughout an entire region rather than preventing the switch in allegiance for a single country.

The region manipulation let us test H2.1, regarding the importance of a challenge to US leadership. The region treatment could take one of four conditions: an African, an Asian, an Eastern European, or a Latin American country. This treatment sought to vary the strategic importance of the region. The United States has substantial economic relations with Asia and strong cultural ties with Europe, as well as treaty allies in both regions. It has many fewer such relations with Africa and Latin America, although the proximity and historical ties of the United States to other hemispheric countries likely make Latin America more important to the United States. To validate these intuitions, we conducted a pretest asking respondents to rank how important they found various world regions. They rated Asia as highly important, Sub-Saharan Africa at a low level of importance, and Eastern Europe and Latin America in between (see Online Appendix, section 3). We excluded the Middle East because of the region's paucity of democracies; we excluded Western Europe because a domino effect seemed implausible at the time.

We specified threat level and region without a pure control to generate a plausible vignette. Without a threat, there would be no reason to consider intervening in the first place. Leaving the region name unspecified could have introduced additional problems relating to information equivalence, for instance, if respondents varied in imagining a particular region as apt to be the target of interventions (Dafoe, Zhang, and Caughey 2018; Suong, Desposato, and Gartzke 2023).

We measured our dependent variable with the question “Would you approve or disapprove of the US government taking actions to reduce the PLQ Party's chances of winning the elections?” Respondents could indicate approval or disapproval on a five-point (strongly disapprove to strongly approve) scale.

Figure 4 displays the results of the experiment (full results are available in the Online Appendix, section 3.1). The raw results show an increase in the level of threat to US hegemony increases support for interference in all regions except for Asia (where the threat does reduce opposition substantially). Nevertheless, intervening to preserve hegemony seems unpopular overall in both threat conditions and in every region. To test the differences by region and threat, we again used ordinal logistic regression. Region treatments did not attain statistical significance, leading us to reject H2.1 regarding regional variation. The degree of threat that US leadership faced did, however, achieve statistical significance in line with H2, with respondents likelier to support intervention in the high-threat than low-threat condition (a gain of about 7 percentage points).

Public support for defending US hegemony via electoral interventions is clearly driven more by perceived threat than by regional importance. Figure A6 in the Online Appendix presents treatment effects by subgroup. It demonstrates that respondents who have little interest in political news respond more strongly to regional context than to threat, while those with some or high interest in political news respond to threat rather than regional context. Similarly, respondents who have some political knowledge (ability to name the majority party of both houses of Congress) respond more strongly to threats to hegemony than those who are unable to name either party.

Experiment 3: Substitutability

Experiment 3 was fielded on the post-election wave of a 1,000-respondent module of the 2022 CES during the second half of November (with about 830 completions due to panel attrition). It explored how varying objectives affected relative support for different intervention strategies. As it would be jarring to simply ask whether the United States should intervene in an election without any cause, we chose the leadership objective to serve as a baseline condition. We also included a third objective that some scholars claim lies behind many other types of US interventions: protecting US economic interests.

Respondents read a scenario that asked them to “Imagine it's several years from now. A foreign country will soon hold elections.” Each treatment condition continued differently:

• Humanitarian: “If elected, the extremist Orange Party plans to enact highly discriminatory policies against the country's religious minority. Experts fear that this could include using the military and police to force millions of members of this minority to leave their ancestral homes and flee to other countries. Thousands could be killed.”

• Leadership: “If elected, the extremist Orange Party plans to turn the country's foreign policy strongly against the United States. Experts fear that could motivate other countries in the region to oppose the United States as well, bringing about the end of U.S. leadership in the region.”

• Economic: “If elected, the extremist Orange Party plans to pursue radical economic policies that would include seizing U.S. businesses in that country and sharply reducing trade with the United States. Experts fear that could lead to layoffs of many workers in the United States.”

Each passage continued by informing the reader that “By contrast, the moderate Turquoise Party will not enact these changes.”

Respondents were then informed that “Polls indicate the election will be close. The White House is debating what policies to adopt toward this foreign country.” We then measured their preferences in two ways. For one, respondents were informed that “Experts have prepared a list of possible options by which US policymakers could respond to this situation.” Respondents then indicated their approval of each option on a 5-point scale (strongly disapprove to strongly approve):

• Contribute funds to the Turquoise Party's election campaign.

• Before the election, secretly offer the Orange Party a major policy concession they desire in return for them not carrying out their plans if elected.

• If the Orange Party wins, secretly encourage a coup to remove them from power.

• If the Orange Party wins, offer its leaders a major policy concession they desire in return for them not carrying out their plans.

• If the Orange Party wins, publicly threaten to impose severe economic sanctions if they carry out their plans.

This measurement enables us to test H3 (a preference for electoral interventions compared to violent ones but not compared to other nonviolent ones regardless of objective) and H3.1. (greater support for electoral interventions relative to alternatives when the objective is humanitarian). We chose campaign funding as the method used in our partisan electoral intervention treatment because it was the most commonly used method for conducting electoral interventions by the United States and other interveners between 1946 and 2014 (Levin 2020, 2024). The temporal descriptions of the various nonelectoral options similarly follow how that the United States has used these various tools in practice.¹⁰ By specifying a range of different objectives (normative, material, and national security/realpolitik), we allow for more general and cleaner evaluations of respondents’ overall views of different options.¹¹ Further, because these choices are specified ex ante, they are not affected by any surprising revelations about the relative strengths of domestic actors in the target country, such as an unexpected surge of support for one party.

Figure 5 displays the raw results and ordinal logistic regression analysis. (See full results in section 4.1 of the Online Appendix.) We find partial, limited support for H3. As predicted, electoral interventions are significantly more popular option than coups (the least favored option). In contrast to H3, electoral interventions were more popular than nonviolent options. Electoral interventions are significantly more popular than both pre- and post-election concessions, and were the second most popular option overall. However, threatening sanctions was the most popular option by a significant margin, mirroring their real-world status as “weapons of first resort” (“The Treasury 2021 Sanctions Review” 2021). Still, even when shown a full menu of other policy tools, many respondents were willing to consider electoral interventions, demonstrating an absence of a “taboo.”

Our other measurement asked respondents to indicate which of four choices they most preferred:

• Intervene before the election to try to stop the Orange Party from winning.

• Intervene only if the Orange Party wins the election.

• Intervene both before and after the election—whatever it takes to stop the Orange Party.

• Do not intervene at all.

The first option corresponded to an electoral intervention, while the second option excluded that option. The third option (intervene before and after) included an endorsement of an electoral intervention and other measures. The last intervention was the nonintervention choice—a rejection of an electoral intervention and all other options. This measure was designed to enable tests of respondents’ preferences concerning the timing of intervention and across policy objectives.

Figure 6 shows the key results of analysis of this question. (Full results are available in Table A12 of the Online Appendix.) Throughout all conditions, “no intervention” was the most favored response. Yet respondents’ preferences over other strategies varied by objective. The most preferred intervention for a humanitarian objective was an intervention at any time, while an electoral intervention was most favored for the economic condition. Overall, an intervention in support of a humanitarian objective was far more popular than an intervention to promote US leadership (or hegemony), with the economic objective ranking between the two. If we combine the “before” and “anytime” categories, 29.9 percent of respondents would contemplate measures that would include an electoral intervention for a humanitarian objective and 26 percent would endorse similar measures for an economic objective, but only 20 percent would do so to defend US leadership.

The same figure also reports results from a multinomial logistic regression. The increase in support for a humanitarian intervention in any phase is statistically and substantially significant; support for an economic purpose is higher in the pre- or post-election phases. Support for a humanitarian intervention is predicted to be 13.3 percentage points higher than for the leadership scenario, with support for an intervention any time 5.4 percentage points higher, for an intervention after an election 4.1 percentage points higher, and for an electoral intervention 3.8 percentage points. The increase in support for an electoral intervention is highest in the economic condition, with an 8.3 percentage point increase compared to the leadership condition. As the objective of an intervention, then, affects preferences over its timing and its overall extent, these results provide some limited indirect support for hypothesis 3.1.

Discussion

Overall, the balance of evidence suggests that the objective of an intervention substantially affects public support for political warfare. Experiments 1 and 3 show that electoral interventions for humanitarian objectives will enjoy greater support than similar interventions undertaken for other goals. Respondents are not immune to realpolitik concerns. Experiment 2 shows a significant threat to US hegemony elicits modestly stronger support for action and Experiment 3 suggests that economic interests may raise support for interventions as well. Likewise, respondents frequently preferred electoral interventions over some violent options and even over some other nonviolent policy options. Such findings are an extension of the principal policy objectives framework to tools of political warfare in which they had not yet been examined and provide further experimental evidence for a set of conclusions drawn from observational studies of public opinion.

Our results suggest that partisan election interventions are not viewed as singularly effective or morally reprehensible—as either a silver bullet or a taboo. Rather, they form part of a toolkit that Americans evaluate in comparison to other intervention strategies. Different objectives lead to different preferences for interventions, both within the set of potential interventions and compared to a preference for no intervention, as Experiment 3 shows perhaps most clearly.

Almost as important is what we do not find. These effect sizes are large but not overwhelming. They must be interpreted in the context of respondents’ modest baseline levels of support for such interventions, even for humanitarian interventions, which contrast with the survey experiments in Levin and Musgrave (2023). In the experiments in that paper, threats to the targeted state's democracy and various national interests consistently elicited plurality baseline support for election meddling across samples with similar demographics.

In the Online Appendix, we present additional analyses and specifications. In general, Republicans and independents appear somewhat less supportive of electoral interventions than Democrats. Racial identification does not seem to influence respondents’ support for interventions, and gender differences seem muted as well.

Conclusion

We conclude that the objectives for which political warfare is waged influence respondents’ support for the use of those tools. Our findings resemble earlier research concerning objectives and public support for armed interventions, but there are also important differences. Although interventions conducted for the humanitarian objectives were more popular than those pursuing other objectives, the relatively low absolute magnitude of support for political warfare among our respondents may suggest that the humanitarian objective may be less attractive as a justification for political warfare compared to the use of force (Jentleson and Britton 1998). Scholars examining support for the use of various policy tools should accordingly not assume a constant cross-domain effect but rather carefully investigate each option.

This paper's findings also contribute to the development of more specific theories of support for foreign policy techniques. Our results support others’ findings that the identity of the victim in a humanitarian crisis substantially influences support for a humanitarian intervention. The increased diversity and secularization of American society (Pew Research Center 2022), with only about three in five Americans now calling themselves Christians, may thus lead to major shifts in US public opinion compared to a generation ago. Instead of studying the general population's evaluations of foreign policy actions, future work ever more urgently needs to more finely distinguish who supports such actions and under what conditions (Kam and Trussler 2017). In this, we join other work, such as studies evaluating support for foreign aid in terms of distributional politics (Heinrich and Peterson 2020) or covert action against fellow democracies (Carnegie, Kertzer, and Yarhi-Milo 2022). We contribute to these efforts by taking seriously not only the objectives and potential beneficiaries but also the role of policy substitutability.

Our results also may inform policy. Our respondents clearly had little interest in repeating the lowlights of the Cold War. To the extent that resurgent great power competition may involve political warfare to preserve US leadership in particular regions, these results indicate that US public support will be conditional and face a low ceiling. In particular, our findings regarding the use of covert coups to preserve US hegemony (and other goals) should be a wake-up call to any future policymakers. Our results indicate that if a US administration tried to repeat the covert ouster of a Salvador Allende-like figure today, the US public would harshly rebuke their efforts. Suggestions to use covert coups should accordingly be relegated to a file drawer.

Scholars still have much to learn about how the US public views political warfare. Our approach demonstrates that they can make progress by extending traditional research programs regarding the use of force (and sanctions) into engaging with nonkinetic strategies, while also using methods that are more suited to causal inference. Future research should investigate additional topics, including the evidence we find suggesting possible US public support for meddling to pursue economic motives and whether the US public is exceptional or typical in its views of such tools.

Acknowledgments

The authors gratefully acknowledge the support of Tatishe Nteta and comments from participants in panels at the 2023 annual meetings of the International Studies Association and the Midwest Political Science Association, as well as a March 2023 Yale International Security Studies (ISS) workshop presentation. We also gratefully acknowledge suggestions from several anonymous reviewers. Author names are listed alphabetically. However, equal authorship is implied.

Author Biography

Dov H. Levin is an Associate Professor of International Relations at the Department of Politics and Public Administration, University of Hong Kong. His current areas of research include partisan electoral interventions/foreign election interference, US foreign policy, and strategic public diplomacy. His research has appeared in several journals including International Organization, American Political Science Review, International Studies Quarterly, and Journal of Conflict Resolution as well as a book in Oxford University Press: Meddling in the Ballot Box: The Causes and Effects of Partisan Electoral Interventions. He received his Ph.D. in Political Science from the University of California, Los Angeles.

Paul Musgrave is an Associate Professor of Government at Georgetown University-Qatar. He studies US foreign policy and international relations theory. His research has appeared in International Organization, International Studies Quarterly, Security Studies, Presidential Studies Quarterly, and Comparative Political Studies, and he has written for the Washington Post, Foreign Policy, and other outlets. He holds a Ph.D. in Government with a focus on International Relations from Georgetown University.

Notes

Authors’ Note: Author names are listed alphabetically. However, equal authorship is implied.

Footnotes

References