- Split View
-
Views
-
Cite
Cite
Ronald R Krebs, Robert Ralston, Aaron Rapport, Why They Fight: How Perceived Motivations for Military Service Shape Support for the Use of Force, International Studies Quarterly, Volume 65, Issue 4, December 2021, Pages 1012–1026, https://doi.org/10.1093/isq/sqab033
- Share Icon Share
Abstract
What shapes public support for military missions? Existing scholarship points to, on the one hand, individuals’ affiliations and predispositions (such as political partisanship and gender), and, on the other hand, factors that shape a rational cost–benefit analysis (notably, mission objectives, the prospects for victory, and the magnitude and distribution of costs). We argue that public opinion is also shaped by beliefs about why soldiers voluntarily enlist. Using novel survey data and an experiment, deployed to a nationally representative sample of Americans, we test how four conceptions of soldiering affect support for a prospective military operation. We find, in observational data, that believing that a soldier is a good citizen or patriot bolsters support for the mission, while believing that a soldier has enlisted because he wants the material benefits of service or has “no other options” undermines support. These results support our causal argument: Americans’ attitudes toward military missions are shaped by their perception of whether the soldier has consented to deployment rather than by feelings of social obligation. This article has implications for debates on the determinants of public support for military missions and the relationship between military service and citizenship in democracies.
¿Qué factores determinan el apoyo público a las misiones militares? En estudios existentes, por un lado, se señalan las afiliaciones y predisposiciones de las personas (como el partidismo político y el género), y, por otro lado, los factores que determinan un análisis costo-beneficio razonable (en especial, los objetivos de las misiones, las posibilidades de que sean exitosas y la magnitud y distribución de los costos). Sostenemos que las creencias de las personas sobre los motivos por los que los soldados se alistan voluntariamente también afectan la opinión pública. Utilizamos una prueba y los datos de una encuesta innovadora, implementados en una muestra representativa a nivel nacional de ciudadanos estadounidenses, para evaluar la manera en que cuatro conceptos influencian el apoyo a una posible operación militar. Mediante los datos de observación, descubrimos que si las personas consideran que un soldado es un buen ciudadano o patriota, el apoyo a la misión se reafirma, mientras que si consideran que se alistó por los beneficios del servicio o porque “no tuvo otra opción,” el apoyo disminuye. Con los resultados experimentales, obtuvimos evidencia a favor de nuestro argumento causal: La actitud de los ciudadanos estadounidenses en relación con las misiones militares se verá más afectada por el consentimiento del soldado sobre el despliegue que por la sensación de obligación social. Este artículo presenta conclusiones para debatir sobre el apoyo público a las misiones militares y la relación entre el servicio militar y la ciudadanía en las democracias.
Par quoi le soutien public des missions militaires est-il façonné? Les recherches existantes indiquent que ce sont d'une part les affiliations et prédispositions des individus (notamment leurs convictions politiques et leur genre), et d'autre part, des facteurs qui constituent une analyse coût/bénéfice rationnelle (particulièrement les objectifs de mission, les perspectives de victoire et la magnitude et la répartition des coûts). Nous soutenons que l'opinion publique est également façonnée par les convictions relatives aux raisons pour lesquelles les soldats s'engagent volontairement. Nous nous sommes appuyés sur des données issues d'une nouvelle enquête et sur une expérience menée auprès d'un échantillon nationalement représentatif d'Américains pour évaluer la manière dont quatre conceptions de l'engagement en tant que soldat affectaient le soutien d'une opération militaire potentielle. Dans nos données d'observation, nous avons constaté que le fait de croire qu'un soldat est un bon citoyen ou un bon patriote renforce le soutien de la mission, alors que celui de croire qu'il s'engage car il souhaite profiter des avantages matériels du service ou car il n'a « aucune autre option » mine ce soutien. Les résultats de notre expérience offrent des preuves soutenant notre argument causal : les attitudes des Américains envers les missions militaires sont motivées par leur perception du consentement du soldat au déploiement plutôt que par des sentiments d'obligation sociale. Cet article a des implications pour les débats sur les déterminants du soutien public des missions militaires et sur la relation entre le service militaire et la citoyenneté dans les démocraties.
Why do mass publics support or oppose military missions? Existing scholarship points to the mission objectives (Jentleson 1992), the prospects for victory (Feaver and Gelpi 2004), and the mission's costs in especially blood (Mueller 1973). Scholars also suggest that public support for war depends on how its costs are distributed. As military conscription has become rare across the West, with smaller, long-serving, volunteer, professional forces replacing the conscripted mass armies of the nineteenth and twentieth centuries, a common view, reinforced by experimental research (Horowitz and Levendusky 2011), is that Western publics have become desensitized to casualties and more likely to endorse the use of military force (Moskos 1977; Segal 1989; Abrams and Bacevich 2001; Burk 2007; Cohen 2001). Yet, in recent decades, Western publics have nevertheless proved sensitive to military casualties among professional, volunteer soldiers. Edward Luttwak, observing the outcry following the October 1993 Battle of Mogadishu in which nineteen US service members had been killed, inferred that, in this era of “post-heroic warfare,” battlefield deaths would not be tolerated (Luttwak 1995; see also Burk 1999). In the 2000s, NATO countries often placed restrictions on their forces in Afghanistan because they worried about public reaction to soldiers’ death and injury (Auerswald and Saideman 2014). Western publics have thus been surprisingly resistant to putting professional, volunteer soldiers’ “boots on the ground.” There is clearly much we still do not understand about public opinion on the use of force.
This article explores the possibility that an important factor has been overlooked: the public's beliefs about why soldiers voluntarily enlist. Perhaps—in line with Charles Moskos’ contention that, with the end of the draft in the United States, military service had become just a “job” (Moskos 1977)—people see professional soldiers as employees who are compensated for their dangerous work. Or perhaps, in line with republican models of citizenship, they view professional soldiers as model citizens, making sacrifices for the common good (Krebs 2006). These beliefs may have powerful political effects, and we explore two theoretical possibilities. The first we call the logic of social obligation. If, notwithstanding the reality of market-based recruitment, people subscribe to the discourse of the citizen-soldier, that may evoke a feeling of social indebtedness and heighten their stake in these soldiers’ fate, depressing their support for military missions. The second we call the logic of consent. If people believe soldiers serve for reasons other than deep devotion to country—say, because soldiers want the material benefits service bequeaths or because soldiers see the military as a way to escape adverse life circumstances—they may infer that soldiers have not freely consented to deployment, and they may therefore be reluctant to send soldiers into harm's way.
We test these logics using novel observational survey data, supplemented by a survey experiment to pin down the causal direction and mechanism. Although the logic of obligation is intuitive and consistent with the conventional wisdom, both the observational and experimental data point more powerfully toward the logic of consent. Our survey of a nationally representative sample of Americans finds, at odds with the logic of obligation, that respondents who believe that soldiers join the armed forces primarily for intrinsic reasons—patriotism and good citizenship—are more likely to support a prospective military mission. In contrast, and in line with the logic of consent, respondents who believe that soldiers enlist for extrinsic reasons—because they want the material benefits of service or because they have “no other options”—are generally less likely to support the mission. In the survey experiment, we expose respondents to a vignette that portrays a fictional soldier who has joined the armed forces for one of these four reasons. The experiment produces results largely consistent with the observational findings. Crucially, respondents exposed to an intrinsic treatment are not only more likely to support the prospective mission, but also more likely to believe that the fictional soldier approves of the mission too. Respondents exposed to an extrinsic treatment, especially those told the fictional soldier enlisted out of desperation, are more likely to believe the opposite and not to support the mission. A causal mediation analysis confirms that the experimental treatments’ effects flow almost entirely through respondents’ beliefs about whether the fictional soldier himself approves of the mission.
These findings have important implications for ongoing debates regarding public opinion and the use of force. They suggest that public judgments regarding military force are the product not just of mission-related inputs—as the existing literature emphasizes—but also of how members of the public think about the soldiers charged with pursuing that mission. Emphasizing how soldiering is represented, this study complements recent scholarship that explores how intervening institutions like the media affect what information the public receives about war (Baum and Groeling 2010), highlights how political ideology and ethnic affiliation shape the public's response to information (Berinsky 2009), and attributes substantial power to elite rhetoric in shaping public perception of war (Berinsky and Druckman 2007).
These findings also have significant policy implications. It is an American public ritual to thank soldiers for their service, honor them for their sacrifice, and hail them for their patriotism (Mann 1988; Enloe 2000; Lutz 2001; Millar 2019). The nation's politicians routinely engage in such rhetorical genuflection (Krebs 2009; Millar 2021). Some contend that militaristic elite rhetoric has troubling consequences—promoting deference to the uniform and undermining democratic civil–military relations (Fallows 2015; Krebs, Ralston, and Rapport 2021) and, by boosting soldiers’ social prestige, allowing the government to lower the costs of a large military (Bacevich 2007). If the logic of obligation had merit, such militarism would have a hidden benefit: it would render Americans more reluctant to send these societal exemplars into danger. But our research suggests yet another reason for disquiet. Ubiquitous militaristic rhetoric downplaying the market-based roots of recruitment and saluting soldiers’ selflessness ironically encourages Americans to support military missions. It may, therefore, help feed the United States’ penchant for using military force.
This article proceeds in five sections. First, we review the literature on public opinion and the use of force. Second, we introduce the contending theoretical logics of social obligation and consent. Third, we describe the survey and its main variables of interest. Fourth, we present and discuss the key observational findings, followed by the experimental results. The conclusion returns to a broader discussion of conceptions of soldiering and the use of force and elaborates opportunities for future research.
Public Opinion and the Use of Force
Although the decision to deploy military force rests with decision-makers in government, the views of the public are hardly irrelevant. Before and during military missions, leaders often worry about whether they can cultivate and sustain public support. Concern that a war will be costly and unpopular affects how force is deployed, how missions are legitimated, how extensively the public is mobilized, and how the executive and legislative branches interact (Sobel 2001; Howell and Pevehouse 2007; Kriner 2010). This concern is not limited to democratic regimes. As the Soviet experience in Afghanistan reveals, even authoritarian governments have difficulty maintaining military missions in the face of active public opposition. Consequently, political leaders devote substantial energy to trying to shape mass opinion on prospective and ongoing military missions, framing campaigns in ways they believe will garner public support and downplaying aspects they worry will arouse opposition (Entman 2004; Berinsky 2009; Baum and Groeling 2010).
Existing research has established that individuals’ characteristics, affiliations, and predispositions affect whether and to what extent they favor the use of force—among others, their hawkishness, moral values, political ideology and partisanship, ethnicity and race, and gender (Gartner and Segura 2000; Nincic and Nincic 2002; Berinsky 2009; Brooks and Valentino 2011; Kertzer et al. 2014; Eichenberg 2019). However, much literature seeks to show that public opinion also responds rationally to some combination of costs in blood (and to a lesser extent treasure), benefits, and the likelihood of success (see, relatedly, Eichenberg 2006). Mueller (1973) argues that public opposition rises along with casualties (costs). Jentleson (1992) maintains that publics are willing to tolerate high casualties if a mission is seen as securing vital interests (benefits), but otherwise show little tolerance for casualties—especially when humanitarian goals are concerned. Feaver and Gelpi (2004) aver that the prospect of victory—that is, the likelihood of reaping the benefits—trumps all. While these evaluative criteria are often cast as competing, they are in fact part of the same rationalist framework: costs, benefits, and the likelihood of success should all feature in a rational public's calculus regarding initiating, continuing, or expanding military operations. Rooted in this framework, subsequent scholarship has explored the conditions under which particular groups of citizens are tolerant of casualties. Especially in volunteer armies, the costs of war are not evenly distributed across the population (Kriner and Shen 2010). Citizens believe the use of force to be costlier if casualties are more salient to them, touching their local community (Althaus, Bramlett, and Gimpel 2012; Kriner and Shen 2012), ethnic or racial group (Gartner and Segura 2000), or family (Gartner 2008; Fazal 2021).
From this rationalist model, it follows that certain recruitment systems render the public as a whole more tolerant of casualties and more supportive of armed force. This is what the turn from the mass conscript army to a volunteer professional military is typically thought to have done. Mass armies draw from a broad swath of the population and the citizenry at large therefore feels the costs of large-scale military operations. In contrast, professional armed forces are composed of long-serving volunteers. Such militaries are not only smaller, but they typically recruit disproportionately from among particular classes, regions, or ethnicities and races. If citizens’ extended social networks include few active-duty soldiers, veterans, or their families, they may not have much awareness of or sensitivity to battlefield casualties. Consolidating widespread opposition to the use of force is therefore harder when the military is a professional, volunteer force. Consistent with this logic, experimental evidence finds that the draft decreases US public support for military missions (Horowitz and Levendusky 2011), at least among Democrats if the draft is fairly egalitarian (Kriner and Shen 2016; see, relatedly, Levi 1997). Democracies, which are more responsive to public opinion, are therefore especially likely to invest in capital-intensive military formats and strategies (Caverley 2014) and are less likely to adopt conscription (Asal, Conrad, and Toronto 2017), except in extremis (Levi 1997).
The mass army has been declining in the West for decades (van Doorn 1975; Haltiner 1998; Moskos 2000). Countries retaining conscription have recruited shrinking portions of each cohort, and they normally have not deployed conscripts overseas (Forster 2006). The turn to volunteer, professional militaries should have made both military commanders and civilian political leaders more sensitive to casualties: the former because they serve with their lower-ranking comrades for longer stretches and share a common identity with them as military professionals, and the latter because highly trained, expert soldiers have become a scarce labor resource (Cornish 2012, 565; Horowitz, Simpson, and Stam 2011). However, the above logic suggests that the shift away from the mass army should have rendered Western mass publics relatively insensitive to military casualties and more supportive of the use of military force. This is the conventional wisdom among military sociologists and historians (Moskos 1977; Segal 1989; Abrams and Bacevich 2001; Cohen 2001; Moran 2003; Burk 2007). Professional armed forces are seen as conducive to what Michael Mann acidly called “spectator sport militarism” (Mann 1988; see also Shaw 1991; Shaw 2005).
However, how publics respond to the prospect and reality of military missions may also be a product of how they think about soldiers and their social role. Moskos famously observed that, with the end of mass conscription, homo economicus—motivated by the skills, salary, and educational benefits of military service, rather than by patriotism or obligation—took the place of the citizen-soldier (Moskos 1977). Segal summarizes it well: with the end of the draft, military service in the United States was redefined “from being an obligation of citizenship in a community to being an obligation of national citizenship and, most recently, to being a job” (Segal 1989, 45). Burk thus concludes that, in embracing the all-volunteer force (AVF), the United States “abandon[ed] the ideal of the citizen soldier” (Burk 2007, 444). Or, as Abrams and Bacevich assert, in the United States, “the mythic tradition of the citizen-soldier is dead” (Abrams and Bacevich 2001). These scholars rightly suggest that the decline of the mass army is as much a cultural phenomenon as an institutional one.
However, these observations mistakenly identify a specific recruitment format with a particular discursive ideal. How soldiers are recruited to service can be quite independent of the culture of service. Even if soldiers enlist in response to market incentives, their fellow citizens may hold very different beliefs about why soldiers serve and what the political community owes them. For instance, the Israel Defense Forces has long maintained mandatory service for Jewish males, but Levy records how a militarist republican logic of citizenship seized the center of Israeli discourse, lost that position, and then regained it (Levy 2007). In the United States, despite the turn to the AVF in 1973, soldiers and officers bolstered their social prestige, and they are today firmly associated in public discourse with duty, love of country, honor, and self-sacrifice—values at odds with the logic of the market. In US politicians’ rhetoric and in the American public imagination, the mythic tradition of the citizen-soldier is alive and well (Krebs 2009). Explaining why one conception of soldiering is prevalent at a given time and in a given place is beyond the scope of this article. What matters is that these conceptions may be independent of how soldiers are recruited and may have political consequences. In this article, public opinion on military service and the use of force takes a discursive and cultural turn.
Conceptions of Soldiering and Their Consequences
We begin with a typology of conceptions of soldiering. This typology, which we have developed at length elsewhere (Krebs and Ralston 2020), is grounded in two theoretically significant dimensions: the source of soldiers’ motivation for military service and the degree to which their decision to volunteer is unconstrained. These two dimensions derive from our core theoretical concepts, obligation and consent. Soldiers may be seen as having voluntarily joined the military for either intrinsic reasons—that is, due to personally held values associated with military service—or extrinsic reasons—that is, for the wages and other benefits they receive in exchange for service (Moskos 1977; Janowitz 1983; Burk 2000; Cohn 2015; Woodruff 2017). Soldiers’ enlistment may also be seen as having been driven either by high constraint—that is, as a result of substantial pressure, whether internal or external and whether material or social in origin—or low constraint—that is, as the product of unencumbered choice.
In combination, represented in table 1, these two dimensions constitute four soldier ideal-types. Employees (extrinsic motivation, low constraint) freely choose to join the military. Recruited on the open labor market, they enlist because they want the material benefits promised in exchange for military service. For the desperate (extrinsic motivation, high constraint), military service is also a means to a material end: they too are attracted to the financial benefits of military service. But they sign up because they have (or believe themselves to have) little other way to escape what seems to be their destiny. Due to their socioeconomic origins, race, or place of birth, their options are very limited. Patriots (intrinsic motivation, low constraint) freely volunteer for military service out of love of country. This is only one of many ways they could conceivably choose to express their patriotism. Good citizens (intrinsic motivation, high constraint) join because they see it as their duty to the political community. Like patriots, their principles move them to enlist, but, unlike patriots, they narrate military service as an ethical obligation, which they feel enjoined to fulfill.1
. | . | Degree of constraint . | |
---|---|---|---|
. | . | Low . | High . |
Intrinsic | “Patriot” (Love of Country) | “Good Citizen” (Duty/obligation) | |
Motivation for joining the military | |||
Extrinsic | “Employee” (Pay, benefits) | “Desperate” (No other options) |
. | . | Degree of constraint . | |
---|---|---|---|
. | . | Low . | High . |
Intrinsic | “Patriot” (Love of Country) | “Good Citizen” (Duty/obligation) | |
Motivation for joining the military | |||
Extrinsic | “Employee” (Pay, benefits) | “Desperate” (No other options) |
. | . | Degree of constraint . | |
---|---|---|---|
. | . | Low . | High . |
Intrinsic | “Patriot” (Love of Country) | “Good Citizen” (Duty/obligation) | |
Motivation for joining the military | |||
Extrinsic | “Employee” (Pay, benefits) | “Desperate” (No other options) |
. | . | Degree of constraint . | |
---|---|---|---|
. | . | Low . | High . |
Intrinsic | “Patriot” (Love of Country) | “Good Citizen” (Duty/obligation) | |
Motivation for joining the military | |||
Extrinsic | “Employee” (Pay, benefits) | “Desperate” (No other options) |
Two plausible logics link these conceptions of military service to individuals’ reactions to prospective military missions. Each can apply to both prior beliefs (observational) and primed beliefs (experimental) regarding why soldiers enlist.
Social Obligation
Alvin Gouldner argued that societal rules of reciprocity generate obligations. In ordered societies, in which a norm of reciprocity is well-established, a substantial amount of time can intervene between an individual or group's provision of some benefit and the recipient's payment. During this period, the relationship in question lies under “the shadow of indebtedness” and imposes obligations on the debtor (Gouldner 1960, 174). Soldiers provide the public good of security to their fellow citizens, under conditions of diffuse reciprocity that cast the shadow of indebtedness widely. Members of the polity are expected to repay that debt by, among other things, displaying a healthy respect for soldiers’ lives and well-being by taking care over what military operations they authorize. This logic is implicit in the common idea—embodied in prominent British debates of the 2000s (Ingham 2014)—that soldiers serve the nation under a “covenant,” which entails tacit mutual obligations, rather than an explicit contract.
Divergent conceptions of military service accentuate society's debt to soldiers and thereby activate the logic of social obligation to different degrees. First, intrinsic motivation for service increases society's debt and corresponding obligation. When soldiers are thought of as intrinsically motivated, their fellow citizens acquire soldiers’ services at a discount. In contrast, when soldiers are considered extrinsically motivated, their service is provided under explicit contract and therefore does not incur much debt: their labor has been compensated at the negotiated terms. The logic of social obligation thus predicts that respondents should be less supportive of military missions when soldiers are thought of as patriots or good citizens than when they are considered employees or the desperate (Hypothesis 1a). Second, high-constraint soldiers reinforce that sense of debt and obligation. Soldiers impelled to enlistment by their internal sense of duty or by external material pressures have less motivation (the good citizen) or capacity (the desperate) to negotiate favorable terms of employment. Therefore, the state acquires their services at a greater discount compared to soldiers who have more choice in joining, and respondents should feel a greater debt to these high-constraint soldiers. Per the logic of social obligation, when soldiers are thought of as desperate, respondents should be less supportive of missions relative to when soldiers are thought of as employees (Hypothesis 1b). Likewise, respondents should be less supportive of military missions when soldiers are thought of as good citizens, moved by moral compulsion, relative to patriots, who freely choose a military life (Hypothesis 1c). Put another way, within intrinsic and extrinsic motivational categories (rows of table 1), soldiers face more or less constraint in deciding to join the armed services (columns of table 1), which we hypothesize generates social debt among respondents (Hypotheses 1b and 1c).
Respondents who believe soldiers join the armed forces primarily for intrinsic reasons (or have been so primed) are less likely to favor a prospective military mission than are respondents who believe soldiers enlist primarily for extrinsic reasons (or have been so primed).
Respondents who believe soldiers join the armed forces primarily out of desperation (or have been so primed) are less likely to favor a prospective military mission than are respondents who believe soldiers enlist primarily for the pay and benefits associated with service (or have been so primed).
Respondents who believe soldiers join the armed forces primarily out of good citizenship (or have been so primed) are less likely to favor a prospective military mission than are respondents who believe soldiers enlist primarily out of patriotic motivations (or have been so primed).
Consent
Alternatively, the logic of consent suggests that people are more likely to support potentially costly measures—such as sending troops into battle—when those who would bear the costs do so willingly. The logic of consent rests on the “golden rule”: people prefer that they not be compelled to do something they would rather not do, and they similarly prefer not to coerce reluctant others into action. This preference gives rise to motivated bias, in that people are more likely to associate freely taken actions with positive outcomes (Stillman, Baumeister, and Mele 2011). Both the “golden rule” and the logic of consent are rooted in the variable human capacity for empathy. Empathy involves cognitive and emotional identification with another. It requires imagining how that person thinks or feels given her situation and adopting her perceptions and emotions as, to some extent, one's own. Empathetic individuals are therefore reluctant to coerce others, and they even feel impelled to offer assistance (Eisenberg and Strayer 1987; Batson 2017; Stocks and Lishner 2018). Grounded in the psychology of empathy, consent theory expects that individuals are disinclined to favor the use of force when they think that the deployed soldiers are themselves not likely to support the mission—that is, because they believe soldiers have dovish preferences—and that individuals are inclined to favor the use of force when they believe soldiers themselves support the mission, because they believe soldiers have hawkish preferences.
Different conceptions of military service have divergent implications for soldiers’ presumed consent. Consent is, intuitively, inferred from what observers think they know about how an action aligns with an actor's desires (Monroe and Malle 2010). Extrinsically motivated soldiers—employees and desperate—are less devoted to, and identify less closely with, the military and its values than the patriots and good citizens who serve for intrinsic reasons. Survey research on the US military confirms this intuition: “intrinsic enlistment motives/goals have a strong positive effect on [soldiers’] perceptions of the organization, social satisfaction, organizational identification, and discretionary pro-organizational behaviors” (Woodruff 2017, 579). It follows that extrinsically motivated soldiers deploy on combat missions less willingly: they deploy not because they want to, but because their contract demands it. In contrast, intrinsically motivated soldiers go off to fight more readily: because they are devoted to the organization and its mission, not just because they are contractually bound. Put differently, respondents associate soldiers’ extrinsic motivations with dovish preferences and soldiers’ intrinsic motivations with hawkish preferences. Consequently, if consent theory holds, observers who think of soldiers as extrinsically motivated are less likely to support prospective military missions (Hypothesis 2a).
Among extrinsically motivated soldiers only, consent is also a function of the degree to which they have freely chosen to serve in the armed forces: the more constrained that decision, the less observers can infer soldiers’ consent from their actions. Soldiers who choose military service freely are presumed to be more committed to the organization and the use of force, and therefore more supportive of combat deployments, than those whose enlistment reflects a lack of options. Whereas the logic of obligation sees constraint as operating similarly and consistently across both motivational categories, the logic of consent limits constraint's effects to extrinsically motivated soldiers: it suggests that individuals should be less supportive of military missions when soldiers are thought of predominantly as desperate and more supportive when soldiers are viewed chiefly through the pay-and-benefit lens (Hypothesis 2b).
Respondents who believe soldiers join the armed forces primarily for extrinsic reasons (or have been so primed) are less likely to favor a prospective military mission than are respondents who believe soldiers enlist primarily for intrinsic reasons (or have been so primed).
Respondents who believe soldiers join the armed forces primarily out of desperation (or have been so primed) are less likely to favor a prospective military mission than are respondents who believe soldiers enlist primarily for the pay and benefits associated with service or for either intrinsic reason (or have been so primed).2
Research Design and Variables
The preceding theoretical claims cannot be evaluated with existing observational data. Available surveys of US public opinion have very rarely asked respondents why they think US soldiers join the military (but see Avant and Sigelman 2010). Thus, to assess our claims, we fielded a survey, in September 2018, to 2,451 US-based respondents recruited by Lucid.3 Lucid weights data and supplies respondents using an iterative process that matches sex, age, education, race, Hispanic origin, state, and region to parameters from the 2015 Census Bureau's American Community Survey. On these characteristics, our sample leans slightly more educated, white, and Democratic, but is otherwise largely comparable to existing national benchmarks.4 With minor exceptions, respondents were equally and randomly distributed across the treatment groups.5
The survey includes both observational and experimental components. To avoid introducing post-treatment bias into the statistical analysis, respondents read all questions measuring control variables before their randomly assigned vignette. In addition, all respondents were, before the experimental treatment, asked about soldiers’ reasons for enlisting in the armed forces. All respondents were then presented with a vignette portrayed as an edited selection from a news article. The vignette describes an interview with a soldier awaiting deployment on a prospective military mission. Three control vignettes do not prime respondents regarding the soldier's motivation for service or degree of constraint,6 thereby allowing us to isolate the impact of respondents’ pre-existing beliefs about military service. With respect to this study's variables of interest, these respondents constitute the observational portion of the survey (1,049 respondents). In the four experimental treatments (1,402 respondents), the soldier is asked why he joined the military, and his answer is depicted as typical of his battalion mates.7 These respondents were thereby primed to one of four distinct portraits of the soldier in question, varying his service narrative corresponding to our typology (table 1).
The primary dependent variable is respondents’ support for the prospective mission, measured on a seven-point scale. The primary explanatory variable in the observational study is the respondent's pretreatment belief regarding why soldiers enlist. We asked respondents to choose which of four options—pay and benefits (corresponding to the employee rationale), no other way to escape desperate life circumstances (desperate), love of country (patriot), and duty as citizen (good citizen) rationales—best represents their belief as to why people primarily join the armed services. Descriptively, respondents most commonly subscribed to the pay-and-benefits service narrative (43 percent), while another 10 percent endorsed the view that service members primarily join because they have “no other options.” The remaining 47 percent opted for one of the two intrinsic narratives: 33 percent of respondents chose the patriotic service narrative, while 14 percent chose the citizenship service narrative.8
In the experimental study, the primary explanatory variable is the randomly assigned treatment depicting a fictional soldier's reason for enlisting. In the vignettes, an interviewed soldier, named “Michael Cameron,” explains why he signed up for military service and confirms that most of his battalion joined for the same reason. The article further claims that surveys indicate that Cameron's motives are typical of most service members. We estimate models in which each prime (pay-and-benefits, desperate, patriot, and citizen) is specified separately, as well as models in which intrinsic primes (patriot and citizen) and extrinsic primes (pay-and-benefits and desperate) are grouped together.9
The survey included common controls for demographic factors including age, sex, race, income, and education. It also included controls for party identification, political ideology, feelings toward the military, and hawkishness. We included batteries of questions that gauge a respondent's “blind patriotism,” “right-wing authoritarianism,” and “social dominance orientation”; all these are correlated with, but distinct from, political ideology (Van Hiel and Mervielde 2002). Finally, we controlled for personal and household military status.
The only post-treatment variables, outside the dependent variables, were to estimate mediation effects. The survey posed questions that sought to gauge directly whether respondents endorse, in the abstract, the logics of obligation and consent. Respondents were asked how strongly they agree with two statements: “when soldiers would rather not be deployed on a mission, I am uncomfortable putting them in harm's way” (consent); “because soldiers make sacrifices on my behalf, I owe it to them to be especially cautious about putting them in harm's way” (obligation).10 We also asked questions about what respondents thought about the fictional soldier, Michael Cameron. If respondents exposed to intrinsic primes were more likely to think that Cameron “personally approved of the mission,” and if respondents exposed to extrinsic primes were less likely to think so, that would suggest that the logic of consent is at work. Relatedly, because conservatives are more likely to favor the use of military force, respondents’ beliefs about Cameron's political orientation carry implications for their perception of his approval of the mission. If respondents exposed to intrinsic primes were also more likely to think that Cameron was politically conservative, and if respondents exposed to extrinsic primes were less likely to think so, that would bolster the logic of consent.
Results
We find considerable support for the logic of consent and little support for the logic of obligation. Consistent with consent, and contrary to obligation, respondents who believe that soldiers serve primarily for intrinsic reasons were more likely to favor the prospective mission across a variety of specifications. These respondents were also more likely to believe that the fictional soldier Michael Cameron himself approved of the mission, bolstering our confidence in consent theory. Finally, respondents who endorsed soldiers’ consent as a normative principle were less likely to support the mission, while those who endorsed the logic of obligation were, in some models, more favorable toward the mission. The experiment reaches similar substantive conclusions: those primed to see soldiers as serving primarily for intrinsic reasons were more likely to favor the mission than those primed to view soldiers as enlisting primarily for extrinsic reasons. More importantly, however, the treatment effects were almost entirely mediated by respondents’ post-treatment beliefs about whether Cameron himself approved of the mission.
Observational Results
The observational analysis provides fairly strong support for consent theory and offers no support for the logic of obligation. In the bivariate data, as expected by Hypothesis 2b, respondents who believe soldiers join out of desperation were less likely to favor the mission. The group mean among these respondents was 2.99 (σ = 1.68), compared to 3.54 (σ = 1.35) among all other respondents—a difference of 18.4 percent. Substantively, the desperate believers “neither favor nor oppose” the prospective mission, while all other respondents “somewhat favor” the mission. Consistent with Hypothesis 2a, those respondents who believe soldiers join for pay and benefits (mean: 3.52; σ = 1.35) were also somewhat less favorable toward the mission than those who subscribe to a patriotic narrative of military service (mean: 3.62; σ = 1.32). These reported differences of means are all statistically significant at conventional levels. However, contrary to Hypothesis 2a, believers in the pay-and-benefits account were not significantly less supportive of the mission than those who believe soldiers enlist primarily out of a sense of duty.11
These descriptive results are largely borne out in multivariate statistical analysis. In OLS models, the belief that people join the armed forces chiefly due to desperate circumstances significantly depressed support for the prospective military mission—as expected by Hypothesis 2b.12 In table 2, in which “prior belief: desperate” is the base category, respondents who endorsed any other narrative of soldiering were significantly more favorable toward the mission than were those who believed soldiers enlist because they lack other options. These results hold across almost all model specifications, including those with full controls. Moreover, even though respondents were told virtually nothing about Cameron beyond his name and his possible deployment, they endorsed beliefs about Cameron consistent with both their own priors and the logic of consent. Respondents who believed that Cameron personally approved of the mission were more likely to favor the mission, and this result remained unchanged in a model with additional controls (table 2, Model 5). The logic of obligation, in contrast, does not yield any expectations about Cameron's own attitude toward the prospective mission. Consent theory performed somewhat less well with respect to the pay-and-benefits service narrative. When the desperate prior belief is the base category, in most models, the coefficient of the pay-and-benefits prior belief variable was smaller than that of the patriotic belief respondents—consistent with the bivariate findings and Hypothesis 2a—but the magnitude of that difference was small. Moreover, when the pay-and-benefits account is the reference category, believers in soldiers’ patriotic motivations were not significantly more favorable toward the mission, contrary to Hypothesis 2a—though, in most models, that variable's sign is positive, in line with the hypothesis.13
Ref Group: . | Priors only . | Priors and ideology . | Priors and Cameron's ideology . | Priors, Cameron approval, and obligation/consent . | Full model (abbreviated) . |
---|---|---|---|---|---|
Prior: Desperate . | (1) . | (2) . | (3) . | (4) . | (5) . |
Prior: pay-and-benefits | 0.53** | 0.48** | 0.48** | 0.47** | 0.42** |
(0.15) | (0.15) | (0.15) | (0.15) | (0.14) | |
Prior: patriot | 0.63*** | 0.54** | 0.54** | 0.51** | 0.28+ |
(0.16) | (0.16) | (0.16) | (0.15) | (0.15) | |
Prior: citizen | 0.40* | 0.36* | 0.38* | 0.33+ | 0.21 |
(0.18) | (0.18) | (0.18) | (0.18) | (0.17) | |
Political ideology | 0.55** | 0.54** | 0.22 | ||
(0.16) | (0.16) | (0.18) | |||
Cameron conservative | −0.06 | −0.16 | |||
(0.13) | (0.12) | ||||
Cameron liberal | −0.27+ | −0.37* | |||
(0.15) | (0.14) | ||||
Cameron moderate | 0.06 | −0.01 | |||
(0.11) | (0.10) | ||||
Cameron approves of mission | 1.09*** | 0.94*** | |||
(0.16) | (0.16) | ||||
Consent | −0.79*** | −0.76*** | |||
(0.16) | (0.16) | ||||
Obligation | 0.55** | 0.25 | |||
(0.19) | (0.19) | ||||
Constant | 3.99*** | 3.77*** | 3.80*** | 3.48*** | 3.38*** |
(0.14) | (0.15) | (0.17) | (0.22) | (0.27) | |
Additional controls? | No | No | No | No | Yes |
N | 1,049 | 1,049 | 1,049 | 1,049 | 1,049 |
R2 | 0.0160 | 0.0273 | 0.0780 | 0.0780 | 0.1603 |
Ref Group: . | Priors only . | Priors and ideology . | Priors and Cameron's ideology . | Priors, Cameron approval, and obligation/consent . | Full model (abbreviated) . |
---|---|---|---|---|---|
Prior: Desperate . | (1) . | (2) . | (3) . | (4) . | (5) . |
Prior: pay-and-benefits | 0.53** | 0.48** | 0.48** | 0.47** | 0.42** |
(0.15) | (0.15) | (0.15) | (0.15) | (0.14) | |
Prior: patriot | 0.63*** | 0.54** | 0.54** | 0.51** | 0.28+ |
(0.16) | (0.16) | (0.16) | (0.15) | (0.15) | |
Prior: citizen | 0.40* | 0.36* | 0.38* | 0.33+ | 0.21 |
(0.18) | (0.18) | (0.18) | (0.18) | (0.17) | |
Political ideology | 0.55** | 0.54** | 0.22 | ||
(0.16) | (0.16) | (0.18) | |||
Cameron conservative | −0.06 | −0.16 | |||
(0.13) | (0.12) | ||||
Cameron liberal | −0.27+ | −0.37* | |||
(0.15) | (0.14) | ||||
Cameron moderate | 0.06 | −0.01 | |||
(0.11) | (0.10) | ||||
Cameron approves of mission | 1.09*** | 0.94*** | |||
(0.16) | (0.16) | ||||
Consent | −0.79*** | −0.76*** | |||
(0.16) | (0.16) | ||||
Obligation | 0.55** | 0.25 | |||
(0.19) | (0.19) | ||||
Constant | 3.99*** | 3.77*** | 3.80*** | 3.48*** | 3.38*** |
(0.14) | (0.15) | (0.17) | (0.22) | (0.27) | |
Additional controls? | No | No | No | No | Yes |
N | 1,049 | 1,049 | 1,049 | 1,049 | 1,049 |
R2 | 0.0160 | 0.0273 | 0.0780 | 0.0780 | 0.1603 |
Notes: Standard errors in parentheses. +p < .10, *p < .05, **p < .01, ***p < .0001.
Model 5 includes controls for: right-wing authoritarianism (β = 0.33, p = .2), social dominance orientation (β = −1.39, p < .0001), blind patriotism (β = 0.83, p < .0001), military thermometer (β = 0.69, p < .0001), hawkishness (β = 0.26, p = .146), which render political ideology insignificant in this model. Model 5 also includes controls for age, education, income, household service, self-service, sex, and race. For the full table, see table A2.2 in the online appendix. For identical models with ordered logistic regressions, see table A2.1 in the online appendix.
Ref Group: . | Priors only . | Priors and ideology . | Priors and Cameron's ideology . | Priors, Cameron approval, and obligation/consent . | Full model (abbreviated) . |
---|---|---|---|---|---|
Prior: Desperate . | (1) . | (2) . | (3) . | (4) . | (5) . |
Prior: pay-and-benefits | 0.53** | 0.48** | 0.48** | 0.47** | 0.42** |
(0.15) | (0.15) | (0.15) | (0.15) | (0.14) | |
Prior: patriot | 0.63*** | 0.54** | 0.54** | 0.51** | 0.28+ |
(0.16) | (0.16) | (0.16) | (0.15) | (0.15) | |
Prior: citizen | 0.40* | 0.36* | 0.38* | 0.33+ | 0.21 |
(0.18) | (0.18) | (0.18) | (0.18) | (0.17) | |
Political ideology | 0.55** | 0.54** | 0.22 | ||
(0.16) | (0.16) | (0.18) | |||
Cameron conservative | −0.06 | −0.16 | |||
(0.13) | (0.12) | ||||
Cameron liberal | −0.27+ | −0.37* | |||
(0.15) | (0.14) | ||||
Cameron moderate | 0.06 | −0.01 | |||
(0.11) | (0.10) | ||||
Cameron approves of mission | 1.09*** | 0.94*** | |||
(0.16) | (0.16) | ||||
Consent | −0.79*** | −0.76*** | |||
(0.16) | (0.16) | ||||
Obligation | 0.55** | 0.25 | |||
(0.19) | (0.19) | ||||
Constant | 3.99*** | 3.77*** | 3.80*** | 3.48*** | 3.38*** |
(0.14) | (0.15) | (0.17) | (0.22) | (0.27) | |
Additional controls? | No | No | No | No | Yes |
N | 1,049 | 1,049 | 1,049 | 1,049 | 1,049 |
R2 | 0.0160 | 0.0273 | 0.0780 | 0.0780 | 0.1603 |
Ref Group: . | Priors only . | Priors and ideology . | Priors and Cameron's ideology . | Priors, Cameron approval, and obligation/consent . | Full model (abbreviated) . |
---|---|---|---|---|---|
Prior: Desperate . | (1) . | (2) . | (3) . | (4) . | (5) . |
Prior: pay-and-benefits | 0.53** | 0.48** | 0.48** | 0.47** | 0.42** |
(0.15) | (0.15) | (0.15) | (0.15) | (0.14) | |
Prior: patriot | 0.63*** | 0.54** | 0.54** | 0.51** | 0.28+ |
(0.16) | (0.16) | (0.16) | (0.15) | (0.15) | |
Prior: citizen | 0.40* | 0.36* | 0.38* | 0.33+ | 0.21 |
(0.18) | (0.18) | (0.18) | (0.18) | (0.17) | |
Political ideology | 0.55** | 0.54** | 0.22 | ||
(0.16) | (0.16) | (0.18) | |||
Cameron conservative | −0.06 | −0.16 | |||
(0.13) | (0.12) | ||||
Cameron liberal | −0.27+ | −0.37* | |||
(0.15) | (0.14) | ||||
Cameron moderate | 0.06 | −0.01 | |||
(0.11) | (0.10) | ||||
Cameron approves of mission | 1.09*** | 0.94*** | |||
(0.16) | (0.16) | ||||
Consent | −0.79*** | −0.76*** | |||
(0.16) | (0.16) | ||||
Obligation | 0.55** | 0.25 | |||
(0.19) | (0.19) | ||||
Constant | 3.99*** | 3.77*** | 3.80*** | 3.48*** | 3.38*** |
(0.14) | (0.15) | (0.17) | (0.22) | (0.27) | |
Additional controls? | No | No | No | No | Yes |
N | 1,049 | 1,049 | 1,049 | 1,049 | 1,049 |
R2 | 0.0160 | 0.0273 | 0.0780 | 0.0780 | 0.1603 |
Notes: Standard errors in parentheses. +p < .10, *p < .05, **p < .01, ***p < .0001.
Model 5 includes controls for: right-wing authoritarianism (β = 0.33, p = .2), social dominance orientation (β = −1.39, p < .0001), blind patriotism (β = 0.83, p < .0001), military thermometer (β = 0.69, p < .0001), hawkishness (β = 0.26, p = .146), which render political ideology insignificant in this model. Model 5 also includes controls for age, education, income, household service, self-service, sex, and race. For the full table, see table A2.2 in the online appendix. For identical models with ordered logistic regressions, see table A2.1 in the online appendix.
Other ancillary findings in table 2 are also consistent with the logic of consent. Respondents who saw Cameron as politically liberal—and thus as personally disinclined to use force—were significantly more opposed to the mission than those in the reference category, who saw Cameron as holding no particular political views (β = −0.273, p = .068).14 In addition, respondents who endorsed consent as a normative principle (“consent”) were significantly less supportive of the prospective mission—even when controlling for belief in Cameron's approval and a battery of other variables.
Equally important, the findings in table 2 provide no support for the logic of obligation. Contrary to Hypothesis 1a, respondents who endorsed intrinsic service narratives were not less likely to favor the mission. Just the opposite: those beliefs’ coefficients were systematically positive when desperate was the base category, and the patriotic belief's coefficient retained its positive sign (but lost significance) when pay-and-benefits was the base category (table A2.9 in the online appendix). Further, respondents who adhered to the good-citizen narrative were not significantly less favorable toward the military mission than were respondents who favored the patriot account, contrary to Hypothesis 1c (F = 0.37, p = .55). Moreover, and strikingly at odds with the logic of obligation, respondents who said that one should exercise special caution with soldiers because they make sacrifices on others’ behalf (“obligation”) were more likely to support the prospective mission, though not at conventional levels of significance in a model with full controls (table 2, comparing Models 4 and 5).
Respondents’ beliefs about military service also had substantively significant effects on mission favorability. Because all variables in table 2 have been normalized to fall between 0 and 1, their relative substantive effects are readily interpretable. Political ideology is often thought to be highly predictive of attitudes toward the use of force. Yet, in models in which the ideology variable was significant (e.g., table 2, Models 2 and 3), the size of its coefficient (β = 0.54, p < .01) was roughly the same as those associated with respondents’ beliefs in the pay-and-benefits (β = 0.48, p < .01) and patriotic accounts (β = 0.54, p < .01). People's disposition to hawkishness—believing that military force is often useful and necessary in addressing international problems—is obviously related to their views of military missions. Yet, as the coefficient plot in figure 1 shows, respondents’ service narratives were approximately as predictive of their support for the prospective mission. Belief in the normative principle of soldier consent (β = −0.76, p < .0001) was around as predictive of opposition to the mission as blind patriotism (β = 0.83, p < .0001) and warmth of feeling toward the military (“military therm”) (β = 0.69, p < .01) were predictive of support for the mission. Revealingly, the belief that Cameron himself approved of the mission was, in its substantive impact, on a par with these variables (β = 0.94, p < .0001). Finally, per figure 1, beliefs about service mattered more than demographic factors often thought important to public opinion on the use of force, notably race (nonwhite: β = −0.13, p = .17) and sex (female; β = −0.23: p = .006).
The substantive significance of these service narratives is borne out also in respondents’ predicted probabilities of favoring, to any degree, the prospective mission (figure 2). Respondents who believe that people join the military chiefly out of patriotism were likely to favor the mission with a probability of 78.6 percent. Those who believe that soldiers are moved mainly by the pay and benefits received in exchange for service were somewhat less supportive of the mission (72.1 percent)—in line with Hypothesis 2a. Strikingly, respondents who believe that people enlist because they have no other way to escape their desperate circumstances favored the mission with a predicted probability of only 54.2 percent—in strong support of Hypothesis 2b. The pattern, in other words, is consistent with consent theory, and at odds with the logic of obligation. Additionally, we find that the post-treatment questions gauging respondents’ agreement with consent and obligation as normative principles have substantive effects as well. The predicted probability of favoring the mission fell 12 percent (from 72.2 percent to 60.2 percent) when moving from strong disagreement to strong agreement with the principle of consent. In contrast, the predicted probability of favoring the mission increased 20 percent (from 54.9 percent to 74.8 percent) when moving from strong disagreement to strong agreement with the normative logic of obligation.15
Experimental Results
The observational results are much more consistent with the logic of consent than the logic of obligation. But they are necessarily merely suggestive of consent theory: it remains possible that some untheorized factor drives respondents’ support for the mission, which in turn leads them to perceive the fictional soldier Cameron as approving of the mission. If this were true, Cameron's perceived consent would be more the product of mission support than its cause. For that reason, we also conducted a survey experiment.
As our primary concern with respect to the experiment is causal identification, we are not very interested in the magnitude of the primes’ effects. Moreover, for two reasons, we expect the primes’ effects to be modest in size. First, we have constructed a “hard test” by portraying a mission that respondents across the political spectrum should generally support: intervening on behalf of Martesia would, according to the treatment vignette, “further America's strategic interests” (as conservatives and realists would wish) and “strengthen international law” (as liberals would like). It would also “uphold the treaty with Martesia,” and alliance commitments are known to increase the American public's willingness to support military intervention (Tomz and Weeks 2021). However, such a hard test is advantageous, because the logics of consent and obligation can shape the decision calculus only when respondents might support the mission. Second, because we seek to control for respondents’ prior beliefs about military service and therefore ask about these beliefs pre-treatment, respondents may be anchoring themselves to those existing beliefs. As a result, it may be harder to prime them effectively.
The experiment reflects realistic accounts of soldiering and is thus externally valid. While it is true that US political leaders rarely, if ever, refer to US soldiers as anything other than exemplars of patriotism and good citizenship (Krebs 2009), news articles—like that in the experiment—are replete with more realistic portraits of soldiers, often based on interviews with soldiers themselves. In these accounts, soldiers and veterans acknowledge the diversity of motives that underlie enlistment and thereby complicate the simplistic narratives favored in political discourse.16 Moreover, as noted earlier, all four service narratives resonated with respondents, when asked about their beliefs pre-treatment. Finally, it is especially realistic to hear an extrinsic account from soldiers themselves—as in our experiment. Respondents with military experience, especially since the installation of the AVF, were more likely to identify an extrinsic motivation as service members’ primary reason for enlistment (Krebs and Ralston 2020).17
The experiment provides strong support for the logic of consent. Notably, in line with Hypothesis 2a, respondents who received the two intrinsic primes were generally more favorable toward the mission than respondents who received the two extrinsic primes. Relative to those exposed to the desperate prime (the reference group), respondents who received the patriot and good citizen treatments were both more likely to favor the mission (table 3, Models 3 and 4). However, in contrast to the observational results, there was not a statistically significant difference between respondents assigned to the desperate and pay-and-benefits treatments—though, consistent with Hypothesis 2b, the sign of the latter variable's coefficient was positive. When respondents exposed to the intrinsic treatments were combined into a single group, and those who received the extrinsic primes were also, the former group was, as expected, more favorable toward the prospective military mission (table 3, Models 1 and 2).
. | (1) . | (2) . | (3) . | (4) . | (5) . | (6) . | (7) . | (8) . | (9) . | (10) . |
---|---|---|---|---|---|---|---|---|---|---|
Intrinsic primes | 0.19* | 0.17* | −0.02 | 0.02 | 0.19** | −0.03 | 0.01 | |||
(0.07) | (0.07) | (0.08) | (0.07) | (0.07) | (0.08) | (0.07) | ||||
Patriot prime | 0.25* | 0.20* | −0.01 | |||||||
(0.11) | (0.10) | (0.11) | ||||||||
Citizen prime | 0.18+ | 0.21* | 0.00 | |||||||
(0.11) | (0.10) | (0.11) | ||||||||
Pay/benefit prime | 0.05 | 0.07 | 0.03 | |||||||
(0.10) | (0.10) | (0.10) | ||||||||
Cameron approval | 1.19*** | 1.18*** | 0.89*** | 1.27*** | 0.97*** | |||||
(0.14) | (0.14) | (0.13) | (0.14) | (0.13) | ||||||
Consent | −0.75*** | −1.06*** | −0.80*** | |||||||
(0.12) | (0.13) | (0.13) | ||||||||
Obligation | 0.52** | 0.24 | ||||||||
(0.16) | (0.15) | |||||||||
Prior: pay/benefits | 0.49*** | 0.49*** | 0.46** | 0.41** | ||||||
(0.12) | (0.12) | (0.12) | (0.12) | |||||||
Prior: patriot | 0.45** | 0.45** | 0.43** | 0.39** | ||||||
(0.13) | (0.13) | (0.13) | (0.13) | |||||||
Prior: citizen | 0.42** | 0.42** | 0.40** | 0.38** | ||||||
(0.14) | (0.14) | (0.14) | (0.14) | |||||||
Additional controls? | No | Yes | No | Yes | No | No | Yes | No | No | Yes |
N | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 |
R2 | 0.005 | 0.183 | 0.005 | 0.184 | 0.051 | 0.051 | 0.209 | 0.030 | 0.093 | 0.231 |
. | (1) . | (2) . | (3) . | (4) . | (5) . | (6) . | (7) . | (8) . | (9) . | (10) . |
---|---|---|---|---|---|---|---|---|---|---|
Intrinsic primes | 0.19* | 0.17* | −0.02 | 0.02 | 0.19** | −0.03 | 0.01 | |||
(0.07) | (0.07) | (0.08) | (0.07) | (0.07) | (0.08) | (0.07) | ||||
Patriot prime | 0.25* | 0.20* | −0.01 | |||||||
(0.11) | (0.10) | (0.11) | ||||||||
Citizen prime | 0.18+ | 0.21* | 0.00 | |||||||
(0.11) | (0.10) | (0.11) | ||||||||
Pay/benefit prime | 0.05 | 0.07 | 0.03 | |||||||
(0.10) | (0.10) | (0.10) | ||||||||
Cameron approval | 1.19*** | 1.18*** | 0.89*** | 1.27*** | 0.97*** | |||||
(0.14) | (0.14) | (0.13) | (0.14) | (0.13) | ||||||
Consent | −0.75*** | −1.06*** | −0.80*** | |||||||
(0.12) | (0.13) | (0.13) | ||||||||
Obligation | 0.52** | 0.24 | ||||||||
(0.16) | (0.15) | |||||||||
Prior: pay/benefits | 0.49*** | 0.49*** | 0.46** | 0.41** | ||||||
(0.12) | (0.12) | (0.12) | (0.12) | |||||||
Prior: patriot | 0.45** | 0.45** | 0.43** | 0.39** | ||||||
(0.13) | (0.13) | (0.13) | (0.13) | |||||||
Prior: citizen | 0.42** | 0.42** | 0.40** | 0.38** | ||||||
(0.14) | (0.14) | (0.14) | (0.14) | |||||||
Additional controls? | No | Yes | No | Yes | No | No | Yes | No | No | Yes |
N | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 |
R2 | 0.005 | 0.183 | 0.005 | 0.184 | 0.051 | 0.051 | 0.209 | 0.030 | 0.093 | 0.231 |
Notes: Standard errors in parentheses. +p < .10, *p < .05, **p < .01, ***p < .0001.
Models with controls (Models 2, 4, 7, and 10) include: right-wing authoritarianism, social dominance orientation, blind patriotism, military thermometer, hawkishness, ideology, age, education, income, household service, self-service, sex, and race. Among variables commonly associated with support for military missions, RWA, blind patriotism, military thermometer, and hawkishness were consistently predictive. All variables were normalized to fall between 0 and 1. For the full table, see Table A2.10 in the online appendix. For ordered logistic regression results, see table A2.3 in the online appendix.
. | (1) . | (2) . | (3) . | (4) . | (5) . | (6) . | (7) . | (8) . | (9) . | (10) . |
---|---|---|---|---|---|---|---|---|---|---|
Intrinsic primes | 0.19* | 0.17* | −0.02 | 0.02 | 0.19** | −0.03 | 0.01 | |||
(0.07) | (0.07) | (0.08) | (0.07) | (0.07) | (0.08) | (0.07) | ||||
Patriot prime | 0.25* | 0.20* | −0.01 | |||||||
(0.11) | (0.10) | (0.11) | ||||||||
Citizen prime | 0.18+ | 0.21* | 0.00 | |||||||
(0.11) | (0.10) | (0.11) | ||||||||
Pay/benefit prime | 0.05 | 0.07 | 0.03 | |||||||
(0.10) | (0.10) | (0.10) | ||||||||
Cameron approval | 1.19*** | 1.18*** | 0.89*** | 1.27*** | 0.97*** | |||||
(0.14) | (0.14) | (0.13) | (0.14) | (0.13) | ||||||
Consent | −0.75*** | −1.06*** | −0.80*** | |||||||
(0.12) | (0.13) | (0.13) | ||||||||
Obligation | 0.52** | 0.24 | ||||||||
(0.16) | (0.15) | |||||||||
Prior: pay/benefits | 0.49*** | 0.49*** | 0.46** | 0.41** | ||||||
(0.12) | (0.12) | (0.12) | (0.12) | |||||||
Prior: patriot | 0.45** | 0.45** | 0.43** | 0.39** | ||||||
(0.13) | (0.13) | (0.13) | (0.13) | |||||||
Prior: citizen | 0.42** | 0.42** | 0.40** | 0.38** | ||||||
(0.14) | (0.14) | (0.14) | (0.14) | |||||||
Additional controls? | No | Yes | No | Yes | No | No | Yes | No | No | Yes |
N | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 |
R2 | 0.005 | 0.183 | 0.005 | 0.184 | 0.051 | 0.051 | 0.209 | 0.030 | 0.093 | 0.231 |
. | (1) . | (2) . | (3) . | (4) . | (5) . | (6) . | (7) . | (8) . | (9) . | (10) . |
---|---|---|---|---|---|---|---|---|---|---|
Intrinsic primes | 0.19* | 0.17* | −0.02 | 0.02 | 0.19** | −0.03 | 0.01 | |||
(0.07) | (0.07) | (0.08) | (0.07) | (0.07) | (0.08) | (0.07) | ||||
Patriot prime | 0.25* | 0.20* | −0.01 | |||||||
(0.11) | (0.10) | (0.11) | ||||||||
Citizen prime | 0.18+ | 0.21* | 0.00 | |||||||
(0.11) | (0.10) | (0.11) | ||||||||
Pay/benefit prime | 0.05 | 0.07 | 0.03 | |||||||
(0.10) | (0.10) | (0.10) | ||||||||
Cameron approval | 1.19*** | 1.18*** | 0.89*** | 1.27*** | 0.97*** | |||||
(0.14) | (0.14) | (0.13) | (0.14) | (0.13) | ||||||
Consent | −0.75*** | −1.06*** | −0.80*** | |||||||
(0.12) | (0.13) | (0.13) | ||||||||
Obligation | 0.52** | 0.24 | ||||||||
(0.16) | (0.15) | |||||||||
Prior: pay/benefits | 0.49*** | 0.49*** | 0.46** | 0.41** | ||||||
(0.12) | (0.12) | (0.12) | (0.12) | |||||||
Prior: patriot | 0.45** | 0.45** | 0.43** | 0.39** | ||||||
(0.13) | (0.13) | (0.13) | (0.13) | |||||||
Prior: citizen | 0.42** | 0.42** | 0.40** | 0.38** | ||||||
(0.14) | (0.14) | (0.14) | (0.14) | |||||||
Additional controls? | No | Yes | No | Yes | No | No | Yes | No | No | Yes |
N | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 | 1402 |
R2 | 0.005 | 0.183 | 0.005 | 0.184 | 0.051 | 0.051 | 0.209 | 0.030 | 0.093 | 0.231 |
Notes: Standard errors in parentheses. +p < .10, *p < .05, **p < .01, ***p < .0001.
Models with controls (Models 2, 4, 7, and 10) include: right-wing authoritarianism, social dominance orientation, blind patriotism, military thermometer, hawkishness, ideology, age, education, income, household service, self-service, sex, and race. Among variables commonly associated with support for military missions, RWA, blind patriotism, military thermometer, and hawkishness were consistently predictive. All variables were normalized to fall between 0 and 1. For the full table, see Table A2.10 in the online appendix. For ordered logistic regression results, see table A2.3 in the online appendix.
While the experimental results support consent theory, the treatments’ effect sizes were, as expected, substantively small (around 0.2 on a 1–7 scale), especially when compared to other variables commonly associated with support for the use of force. For instance, the key psychological constructs right-wing authoritarianism (β = 0.78, p < .01) and blind patriotism (β = 0.90, p < .0001) respectively had an impact on mission favorability about 4.5 and 5.25 times larger than the impact of the intrinsic experimental treatments relative to the extrinsic experimental treatments (β = 0.17, p < .05). The coefficient for hawkishness (β = 0.55, p < .01) was about three times larger than that associated with the intrinsic experimental treatments.
The inclusion of pre-treatment controls had no impact on the experimental results (table 3, Models 2 and 4). Consistent with the consent framework, respondents’ prior beliefs that soldiers generally serve out of desperation (the base category) were consistently significantly and negatively associated with favoring the mission—as reflected in the positive, significant coefficients of the other prior belief variables. Moreover, agreement with consent as a normative principle was, as in the observational study, associated with less support for the mission (table 3, Models 8–10). In sum, when it comes to mission favorability, the experimental findings provide significant support for Hypothesis 2a and some support for Hypothesis 2b—both derived from the logic of consent.
If consent theory helps explain respondents’ support for the mission, questions measuring perceptions of the fictional soldier's view of the prospective mission should mediate between treatment and outcome. As figure 3 demonstrates, respondents exposed to the patriot prime were most likely to think Cameron approved; those who received the citizen prime inferred the same but less strongly; those assigned to the employee vignette were less likely still; and those who read the desperate prime were least likely, by far, to think Cameron approved. In a multivariate analysis, the treatments were predictive of whether respondents thought Cameron himself approved, and the inclusion of full controls had no impact on the treatments’ significance.18 Importantly, respondents’ prior beliefs in service narratives also had no impact on their view of Cameron's approval (table A2.4 in the online appendix). These results constitute strongly suggestive ancillary evidence in support of consent theory.
Moreover, the belief that Cameron personally supported the mission was positively associated with respondents’ favorability toward the mission, even in models that included pre-treatment beliefs about military service. In a model with only the treatments and the respondent's belief that Cameron approves of the mission, moving from the belief that it is “very unlikely” that Cameron approves of the mission to the belief that it is “very likely” that he approves (“Cameron approval”) was associated with a 1.185-point increase in support for the mission, roughly equal to one standard deviation (σ = 1.39) on the mission support variable (table 3, Model 6). When controls were added, the Cameron approval variable's coefficient remained statistically and substantively significant (β = 0.894; p < .001; table 3, Model 7). Crucially, including this variable also rendered nearly all previously significant treatment effects insignificant (table 3, Models 5 and 6). These results suggest that the experimental treatments’ effects run through respondents’ beliefs about whether Cameron approves of the mission and thus presumably consents to his deployment. Moreover, it was specifically respondents’ perceptions of Cameron's approval, rather than their general beliefs about the need for soldiers’ consent, which powerfully mediated the treatment effect. Including respondents’ post-treatment views of consent as a normative principle had virtually no impact on the “intrinsic” coefficient (table 3, Model 8). Finally, a model of respondents’ predicted probabilities of favoring, to any degree, the prospective mission is represented in figure 4. Respondents who thought it “very unlikely” that Cameron approves of the mission were likely to support the intervention with a probability of 53.8 percent. In contrast, those who believed it “very likely” that Cameron approves were far more favorably inclined, with a predicted probability of 82 percent.19
A more formal analysis, based on Tingley et al.’s counterfactual approach to measuring mediation (Imai, Keele, and Tingley 2010; Tingley et al. 2014), bolsters this conclusion. This approach estimates the average causal mediation effect (ACME) as the indirect effect of the mediating variable on the outcome of interest, holding all else constant. The mediation analysis finds that respondents’ beliefs about Cameron's approval of the mission substantially mediate the experimental treatments’ direct effects. When this mediating variable is included, the estimated direct effect of the intrinsic treatment is essentially negated: from increasing respondents’ support for the mission by 0.169 points (controlling for individual covariates, table 3, Model 2), it falls to 0.016 points, which is not statistically distinguishable from 0 (figure 5). For the intrinsic treatments, the ACME of these beliefs is 0.153—meaning that they mediate just over 90 percent of the treatments’ direct effects on mission favorability.20
Respondents’ perceptions of Cameron's approval might flow from the treatments through subsidiary inferences. For instance, among respondents who received an intrinsic treatment, 34 percent thought Cameron was likely conservative; among respondents who received an extrinsic treatment, only 13.5 percent reached the same conclusion (χ2 = 82.14, p < .001). Believing Cameron to be conservative made respondents more supportive of the prospective mission, but including this variable did not substantially mediate the experimental treatment effects. However, the estimated main effect of Cameron's perceived conservatism was not significantly different from believing him moderate and was only marginally different from believing him liberal (F = 2.82, p = .094). Seeing Cameron as conservative increased respondents’ support for the mission only relative to perceiving Cameron as not having strong political views of any kind (the reference category).21 Similarly, among respondents who received an intrinsic treatment, 79 percent thought Cameron white; 71 percent of respondents assigned to the pay-and-benefits treatment inferred the same, as did just 50 percent of those exposed to the desperate service account.22 Believing Cameron “not white” made respondents less supportive of the prospective mission, but did not substantially reduce the treatment effect. Moreover, crucially, that variable (“C not white”) loses significance once Cameron's perceived approval of the mission is included in the analysis (table A2.5 in the online appendix). We infer that Cameron's perceived political affiliation and race are two of several factors affecting his perceived support for the mission, but relatively little of the treatments’ effects flow through these variables.
The survey experiment thus provides substantial evidence for the logic of consent. Exposing respondents to extrinsic soldier narratives was associated with less support for the military mission, and these effects appear to have been mediated powerfully by respondents’ perception of the fictional soldier's personal approval of the mission. If respondents’ support for the mission drove their perceptions of the fictional soldier's consent, there would not be systematic effects across treatments, since the mission was identical in all the soldier experiment vignettes. Moreover, if respondents’ support for the mission drove their perceptions of the fictional soldier's consent, we would also expect respondents’ perceptions to vary by mission type—yet there was no evidence of that in the three mission vignettes.23
Conclusion
This article's observational and experimental findings generally support the contention that conceptions of soldiering impact respondents’ decisions about the use of force. The results are more consistent with the logic of consent than with the logic of obligation. Americans are more comfortable sending soldiers into dangerous circumstances when they believe that the soldiers themselves would welcome their deployment. Consent, our theoretical framework suggests, lies at the conjuncture of motivation and constraint, and our observational and experimental results generally bear this out. When people believe that soldiers are chiefly patriots or good citizens—that is, intrinsically motivated—they are more supportive of military missions. When they think of soldiers as motivated by the benefits they receive in exchange for service—as employees—they are less supportive of military missions. When they think of soldiers as desperate—as both extrinsically motivated and as essentially coerced into service, not by government policy but by circumstances—they are least supportive of military operations.
Perhaps, however, Americans do not think much about narratives of soldiering when considering prospective military operations—except when prompted by scholars.24 We cannot completely discount this possibility, but it seems unlikely. First, even if many or most Americans would not invoke the logics of obligation or consent when explaining their attitudes toward particular uses of force, such logics may nevertheless shape their thinking. Tacit, “common sense” presuppositions have a deep effect on decision-making, and that impact is arguably greater because those presuppositions go unspoken and unquestioned.25 Second, if such service narratives were not salient to ordinary people, we would expect respondents’ beliefs to be weak and malleable. However, most respondents are “strong believers.” The survey measured the intensity of respondents’ beliefs about service motivation by asking them to estimate what percentage of people join the US military for each of the four supplied reasons. The majority of respondents (62.5 percent) assigned scores of 50 percent or more to one answer. Moreover, these strong believers were also consistent along the intrinsic/extrinsic dimension: strong believers in one extrinsic motivation were most likely to pick the other extrinsic motivation as the second-most common reason they believe service members join, and the same is true of strong believers in intrinsic motivations.26 Were such beliefs prompted merely by survey exposure, we would not expect to see so many strong believers or such consistency of thought.
These findings have troubling normative implications. The conventional wisdom is that the turn to voluntary recruitment in the United States has desensitized Americans to military casualties and rendered them more likely to endorse the use of military force. That prevailing logic has merit, insofar as the AVF touches fewer American families than did its predecessor. However, the findings reported here suggest that how soldiering is framed—the culture of military service—matters too. When US political leaders talk about soldiering, they often invoke a narrative of service centering on soldiers’ intrinsic motivations: patriotism, sense of duty, honor, self-sacrifice (Krebs 2009). This representation of soldiers may make Americans more comfortable with force. Americans may be doubly inclined to favor, and doubly less inclined to withhold their support from, military missions because (1) the US military is drawn from a fairly narrow social base and (2) Americans believe those soldiers to be model patriots and exemplary citizens. Recruitment and rhetoric work together in the contemporary United States to soften opposition to the use of force. If Americans seem militaristic, our research suggests, it is not just because the AVF concentrates the perceived costs of using force, but because of how Americans have been taught to think about the nature of military service. Were the nation's leaders to speak more honestly about military service, and were they to acknowledge that many soldiers are first and foremost professionals doing a (dangerous) job, perhaps the public would be less enthusiastic about the use of force. The prospects of that rhetorical shift seem dim. However, it is worth recalling that US politicians did not always valorize the nation's soldiers and officers as uniquely patriotic and as the best America has to offer. Even today, politicians in other countries—such as Germany and the United Kingdom—do not regularly conflate soldiering with civic virtue.
This study suggests several avenues for future research. First, it would be worthwhile to extend this study cross-nationally. Non-Americans likely have different beliefs about why soldiers enlist and about the basic purposes of the armed forces. Some countries still have conscription, and it is possible that their fellow citizens reflect differently on the nature of sacrifice when conscription is the norm. Whereas Americans appear to associate soldiers’ consent primarily with their motivation for service and only secondarily with the constraints upon their choices, it is possible that others would emphasize more heavily the constraint dimension. Second, this study could be usefully extended to different dependent variables, such as deference to the uniform. Are respondents more deferential to the military's priorities and preferences when primed to see soldiers as good citizens rather than as employees? Third, one can envision numerous contextual variables—whether the mission is prospective or ongoing, whether the mission is aimed at humanitarian objectives or strategic interests—also shaping these dynamics and warranting investigation. Finally, research might explore more directly whether intrinsic service rhetoric is deployed to bolster support for the use of force. For instance, are congressional proponents of particular military missions more likely to reference intrinsic service narratives, presumably to gin up the public for using force? Are opponents more likely either to omit any reference to soldiers themselves or to invoke directly the logic of consent—and thereby, on the margins, suppress public enthusiasm for war? The study of public opinion on military service and the use of force, we have argued, would profit from a cultural and discursive turn.
Supplementary Information
Supplementary information is available at the International Studies Quarterly data archive.
Ronald R. Krebs is Beverly and Richard Fink Professor in the Liberal Arts and Professor of Political Science at the University of Minnesota, and Editor-in-Chief of the journal Security Studies.
Robert Ralston is a Grand Strategy, Security, and Statecraft Fellow jointly appointed at the International Security Program, Belfer Center for Science and International Affairs, Harvard Kennedy School, and the Security Studies Program, MIT.
Aaron Rapport was, until his untimely passing in June 2019, a Lecturer in the Department of Politics and International Studies and a Fellow of Corpus Christi College at the University of Cambridge (UK).
Notes
Author's note: This article is dedicated to the memory of our late colleague and coauthor Aaron Rapport, formerly of the University of Cambridge, whose incisive contributions to this article are apparent on every page. Earlier versions of this article were presented at the Inter-University Seminar of Armed Forces and Society (2017), the annual conference of the American Political Science Association (2018), the Minnesota International Relations Colloquium, and the Upper Midwest International Relations Conference (2019). For helpful comments on earlier versions of the survey and the article, the authors thank the audiences at those presentations as well as Pedro Accorsi, David Blagden, James Burk, Bud Duvall, Tanisha Fazal, Peter Feaver, Chris Federico, Paul Goren, Lukas Herr, Kwanok Kim, Tony King, Andy Kydd, Yagil Levy, Sarah Kreps, Howie Lavine, Joanne Miller, Jon Pevehouse, and Jessica Weeks. For support of this research, the authors thank the University of Minnesota's grant-in-aid program. The data underlying this article are available on the ISQ Dataverse at https://dataverse.harvard.edu/dataverse/isq.
Footnotes
All four categories resonated with respondents. Respondents were presented with these four options, asked to select one as best representing why people primarily join the military, and then invited to offer another view “if none of the options in the preceding question well reflects your view as to people's primary reason for joining the military.” A very small number of respondents suggested alternative motivations, which fell largely into two categories: (1) family tradition of military service and (2) desire for adventure and travel. These two themes appeared in just 1.4 percent of responses.
Readers will note that Hypothesis 2b overlaps with Hypothesis 1b. With respect to extrinsically motivated soldiers, the logics of obligation and consent produce identical expectations.
For the Lucid platform, see https://luc.id/lucid-for-academics/.
See tables A1.1–A1.14 in the online appendix.
See tables A1.14–A1.16 in the online appendix. On Lucid's performance relative to national benchmarks, see Coppock and McClellan (2019).
The three vignettes without soldiering treatments vary the objective of the prospective mission: to defend a US ally, prevent a possible genocide, or fight terrorists. However, this experiment revealed that the impact of soldiering narratives does not vary significantly across missions. While the nature of the mission sometimes had a direct effect on respondents’ support—as one would expect based on past research—it did not significantly interact with our primary variable of interest. Therefore, in this analysis, we treat all respondents assigned to these vignettes as a single group. For results of the mission experiment, see table A2.6 in the online appendix.
For the vignettes and full survey, see the online appendix 2.
For full discussion, see Krebs and Ralston (2020).
We do not collapse the experimental treatments into high-constraint and low-constraint categories for three reasons. First, this would not be theoretically justified, as none of the hypotheses suggest that respondents receiving high-constraint treatments (citizen and desperate) will differ systematically from those receiving low-constraint treatments (patriot and pay-and-benefits). Second, including a collapsed constraint variable along with a collapsed motivation variable would produce collinearity. Third, it would not be empirically justified. Our analysis reveals that the constraint dimension's effects are not sufficiently uniform to warrant collapsing the vignettes. Respondents who received the two low-constraint treatments—patriot and pay-and-benefits—gave significantly different answers to the questions measuring the dependent variables, as did respondents who received the two high-constraint treatments—citizen and desperate. This is in stark contrast to the motivation dimension (see table 2, Models 1 and 2, and figure 1).
The treatments generally did not have significant effects on respondents’ endorsement of either consent or obligation. See table A2.11 in the online appendix.
For means of each treatment group and t-tests, see table A1.17 in the online appendix.
Earlier, in note 2, we observed that Hypotheses 1b and 2b overlap—that the logics of consent and obligation produce similar expectations within the category of the extrinsically motivated. However, Hypothesis 2b is more central to the logic of consent than Hypothesis 1b is to the logic of obligation. In addition, the observational findings are overall more consistent with the logic of consent than obligation.
Not surprisingly, therefore, when we combine the desperate and pay/benefits prior beliefs into one “extrinsic” category, this binary variable's coefficient is statistically insignificant under all specifications.
However, there was no significant difference between respondents who believed Cameron liberal and those who thought him conservative (F = 2.52, p = .113).
These predicted probabilities are based on a model that treats the mission favorability variable as binary (oppose/favor), rather than ordered; in other respects, it is identical to table 2, Model 4. Respondents who neither favored nor opposed the mission were dropped from this analysis (N = 313). All other variables were held at their means. See table A2.8, Model 4 in the online appendix.
See, among others, James Dao (2011), Adryan Corcione (2019), Marissa Cruz Lemar (2020), and Jeremy Stern (2020).
See table A1.19 in the online appendix.
See table A2.4 in the online appendix .
The predicted probabilities in figure 4 are based on table A2.8, Model 3 in the online appendix, which treats the mission favorability variable as binary (oppose/favor). Respondents who neither favored nor opposed the mission were dropped from this analysis (N = 451).
Section 3 in the online appendix 1 describes the sensitivity of the counterfactual mediation analysis to violations of key identifying assumptions, as well as an analysis of how the treatment variables moderate the mediation effect (cf. Muller, Judd, and Yzerbyt 2005).
See table A2.7 and figure A3.2 in the online appendix.
Multivariate findings are consistent with the descriptive data. In a multivariate regression, the intrinsic treatments were positively predictive of perceiving Cameron as white relative to the pay-and-benefits base category, and the desperate treatment was positively predictive of perceiving Cameron as not white. See table A1.18 in the online appendix.
As discussed above, these vignettes vary the mission while holding the soldier's portrait minimal. They constitute the basis for this article's observational analysis. However, if we treat these three vignettes as a “mission type” experiment, with Cameron approval as the dependent variable, there are no significant treatment effects. See table A2.6 in the online appendix.
Thanks to an anonymous reviewer for pressing us on this point.
On the power of the constructed “common sense,” see, among many others, Krebs (2015).
For further discussion of “strong believers,” see Krebs and Ralston (2020).