Two of the century’s most famous American campaign advertisements address gender stereotypes and security concerns. Hillary Clinton’s 2008 “3 a.m. Phone Call” highlighted her foreign policy experience relative to the primary opponent Barack Obama. The ad showed a mother looking in on her sleeping children while a male narrator described Clinton as, “someone who already knows the world’s leaders, knows the military, someone tested and ready to lead in a dangerous world.” A decade later, Texas congressional candidate M. J. Hegar, “an Air Force combat veteran and a mom,” aired the viral “Doors” ad, focusing on her combat experience and decorations for valor. Both ads were praised for their effectiveness at advancing the candidates’ security bona fides (though both candidates lost).
That the two ads focused on security should not be surprising. Women often work to demonstrate expertise across numerous issues when running for office (Atkinson and Windett Reference Atkinson and Windett2019; Bauer Reference Bauer2020)—and women are often especially disadvantaged in the masculine policy space of security (Holman, Merolla, and Zechmeister Reference Holman, Merolla and Zechmeister2022; Schramm and Stark Reference Schramm and Stark2020; Schwartz and Blair Reference Schwartz and Blair2020). Could signaling military bona fides open a way for women to enter the national discourse in a space that has long excluded them (Holman, Merolla, and Zechmeister Reference Holman, Merolla and Zechmeister2022; Schramm and Stark Reference Schramm and Stark2020)?
Even as women are underrepresented in American politics relative to the U.S. population, veterans are overrepresented.Footnote 1 The six women veterans (joining 90 men) in the 117th Congress is a near-historic high (Bialik Reference Bialik2017), and this number is even more remarkable given that the first woman veteran was only elected in 1998 (Harberkorn Reference Harberkorn2018; Mitchell Reference Mitchell2018). As Seth Lynn, an advocate for veteran candidates, argues, “Women candidates, fair or unfair, are often questioned about whether they’re tough enough for the job, whereas male candidates aren’t.” If that woman is a veteran, according to Lynn, the question “never even comes up” (Mitchell Reference Mitchell2018).
Bridging a large body of International Relations (IR) theory with research on the role of gender in political communication, we consider the intersection of gender and veteran status. We find that highlighting one’s security bona fides—either through hawkish rhetoric or by highlighting veteran status—can help women and men receive higher leadership trait evaluations. These qualifications, however, are less effective when the goal is to persuade people that a policy is a matter of national security—a process that requires authority (Wæver Reference Wæver2011). Put another way, bona fides may help politicians, particularly women, be perceived as leaders but will not help them be perceived as authorities.
Our results come from two experiments conducted on national samples of American adults. The second study is our main focus, but we present the first study as it informs our argument.Footnote 2 These studies contribute to research programs in IR, gender and politics, and political communication, as well as reinforce the value of political psychology methods, used extensively in IR’s “behavioral revolution” (Hafner-Burton et al. Reference Hafner-Burton, Haggard, Lake and Victor2017; Kertzer and Tingley Reference Kertzer and Tingley2018), for more critical theoretical approaches to international politics (Baele and Thomson Reference Baele and Thomson2017; Jackson Reference Jackson2011, 105–6).
WAR, THE STATE, AND MAN
We begin by observing that the construction of security, defense, and war as the state’s primary function has powerful political consequences to this day (Hobbes Reference Hobbes1982; Schmitt Reference Schmitt2007; Tickner Reference Tickner1992; Tilly Reference Tilly1992). The provision of security for its citizens is traditionally regarded as the state’s most profound duty, a principal source of its legitimacy, a central obligation for aspiring leaders, and a justification for extracting resources from the public. We couple this concept of the state as security provider with feminist and critical IR claims that state legitimacy and coercion are founded through socially constructed power relations built on both militarism and gender exclusion (Elshtain Reference Elshtain1995; Enloe Reference Enloe2000; Sjoberg Reference Sjoberg2013; Tickner Reference Tickner1992).
The Process of “Securitizing”
This special status of security allows actors to “securitize” issues (Buzan, Wæver, and de Wilde Reference Buzan, Wæver and de Wilde1998, 23–4) on behalf of the state, thereby gaining additional political resources and avoiding normal political constraints. Wæver (Reference Wæver2011, 469) describes the process succinctly as “a securitizing actor claiming an existential threat to a valued referent object in order to make the audience tolerate extraordinary measures that otherwise would not have been acceptable.” The referent object is most traditionally the state. An audience ultimately decides if the threat merits an extraordinary response. Between these two components lies the “securitizing actor,” who strategically presents to the audience a case that the referent object is endangered.
Although the theory rests on the idea of authority, this securitization research program, often known as the “Copenhagen School,” has paid comparatively little attention to how an actor achieves the “authorization” to speak about security (Balzacq Reference Balzacq2005; Léonard and Kaunert Reference Léonard, Kaunert and Balzacq2010; McDonald Reference McDonald2008). Much of the research assumes that common securitizing actors include “political leaders, bureaucracies, governments, lobbyists, and pressure groups” (Buzan, Wæver, and de Wilde Reference Buzan, Wæver and de Wilde1998, 40). According to Buzan, Wæver, and de Wilde (Buzan, Wæver, and de Wilde Reference Buzan, Wæver and de Wilde1998, 41), a successful securitization process depends less on “who performs the speech than of what logic shapes the action.” This idea that the “who” is less important than the “what,” however, has come into question by important theoretic approaches in IR.
Who Gets to Securitize?
Securitizing, at its essence, means persuading an individual that something is related to national security. To have the authority to do so, however, a speaker needs to demonstrate one’s bona fides at providing security. There are different means of signaling these bona fides. The obvious example is that politicians may present themselves as hawks.
Another approach for signaling these bona fides is highlighting prior experience with the military. Indeed, the U.S. military’s immense public prestige relative to other institutions is a potent political tool when it comes to policy persuasion (Krebs, Ralston, and Rapport Reference Krebs, Ralston and Rapport2021). Golby, Feaver, and Dropp (Reference Golby, Feaver and Dropp2018) find that “senior military officers” can nudge respondents’ opinions on foreign military operations (both pro and con). Jost and Kertzer (Reference Jost and Kertzer2023, 15) find that the American “public is significantly more likely to defer to advisers with combat and (especially) military experience, even on non-military issues” (see also Motta, Ralston, and Spindel Reference Motta, Ralston and Spindel2021). Similarly, voters look favorably on veteran candidates over nonveterans (Hardy et al. Reference Hardy, Coker, Funk and Warner2019; Richardson Reference Richardson2022).
Can Women Securitize?
Hansen (Reference Hansen2000) highlights the Copenhagen School’s myopia on gender, observing that women’s voices in general are silenced due to their frequent exclusion from political power. Considerable evidence exists that leadership is generally associated with masculinity (Bauer Reference Bauer2020; Schneider and Bos Reference Schneider and Bos2019) and that women are less likely to be assessed as “strong leaders” (Croco and Gartner Reference Croco and Gartner2014; Schneider and Bos Reference Schneider and Bos2014). Issues like the military, counterterrorism, and security remain tied to men in politics (Dolan Reference Dolan2004; Holman, Merolla, and Zechmeister Reference Holman, Merolla and Zechmeister2011; Reference Holman, Merolla and Zechmeister2022; Rosenwasser and Dean Reference Rosenwasser and Dean1989). Lawless (Reference Lawless2004, 479), for example, argues that an “atmosphere of war” interacts with voters’ perception about men’s competence at “legislating around issues of national security and military crises,” and Holman, Merolla, and Zechmeister (Reference Holman, Merolla and Zechmeister2022, 251) find that during a time of threat, “the public looks for masculine traits.” This work suggests that women politicians may not benefit when questions of national security arise.
Knowing this, women may anticipate that they will receive less support (Butler and Preece Reference Butler and Preece2016) and thus preemptively work to overcome any political handicap (Anzia and Berry Reference Anzia and Berry2011; Bauer Reference Bauer2020). This may mean deliberately sending masculine signals (Bauer Reference Bauer2017). One such signal is military bona fides (Hansen Reference Hansen2000; Huddy and Terkildsen Reference Huddy and Terkildsen1993; Meeks Reference Meeks2012). Swers (Reference Swers2007), for example, finds that women more frequently sponsor defense-related U.S. Senate bills. At the executive level, research suggests that women may have an incentive to send international military signals as a way of addressing questions of leadership (Schwartz and Blair Reference Schwartz and Blair2020). Perhaps most directly, Thomsen (Reference Thomsen2011) argues that women deliberately take on hawkish positions to counteract such public perceptions.
Sending these types of military cues may not be purely advantageous for women. Research points to a version of the “double bind”: women must send masculine signals to succeed, but in doing so may seem unlikeable to voters (Bauer Reference Bauer2017; Swers Reference Swers2007). Hawkish signals specifically lead women to be perceived as more capable in this area (Thomsen Reference Thomsen2011), but Bauer (Reference Bauer2017) finds that women who engage in counter-stereotypic behavior do face more backlash from the opposing party.
But while leadership is a trait that is militarized and gendered, security as an issue is itself gendered as well. If, as Anderson-Nilsson and Clayton (Reference Anderson-Nilsson and Clayton2021, 820) argue, people are most persuasive in “contexts that fit well with salient aspects of their identity,” a military bona fides cue may not be enough to overcome the idea that security and foreign policy are generally perceived as masculine (Holman, Merolla, and Zechmeister Reference Holman, Merolla and Zechmeister2022). In other words, people may have a positive association with veterans that transcends gender, but that positivity may not translate to policy persuasion in a traditionally masculine issue area.
LEADERSHIP VERSUS AUTHORITY
To consider the role of gender in securitizing, we begin with two ideas: leadership, a trait evaluation, and authority, the ability to persuade on policy. Being perceived as someone who has leadership qualities can make one a more attractive political candidate, while being an authority can raise the salience of a political issue and mobilize the public to act upon it, “mak[ing] the audience tolerate extraordinary measures that otherwise would not have been acceptable” (Wæver Reference Wæver2011, 469). Demonstrating security bona fides may help women (and men) signal leadership capability. Yet existing scholarship is less clear whether the same bona fide signals can overcome gender gaps in authority. Footnote 3
Gender, Security, and Leadership
People often evaluate politicians across a variety of traits, chief among them leadership (Ditonto Reference Ditonto2017; Schneider and Bos Reference Schneider and Bos2019). Indeed, one of the barriers women face in politics is that “there is a high level of incongruence between being female and being a leader” (Bauer Reference Bauer2020, 4). Women politicians are likely to gain more support when they send leadership cues (Bauer Reference Bauer2015; Reference Bauer2017; Huddy and Terkildsen Reference Huddy and Terkildsen1993). This suggests that women who send signals about their security bona fides will be evaluated as superior leaders and candidates relative to women who do not.
Sending signals about military bona fides can happen in different ways. Previous research, for example, has suggested that women may make hawkish statements to send clear security and leadership signals (e.g., Swers Reference Swers2007; Thomsen Reference Thomsen2011). Discussing prior military experience—for example, veteran status—would be another, likely more direct path to signaling military bona fides (Hardy et al. Reference Hardy, Coker, Funk and Warner2019).
Certainly, the link between veteran experience and gender is unlikely to be straightforward. Indeed, as Thomas and Hunter (Reference Thomas and Hunter2019, ix) write, “the very word ‘veteran’ calls to mind the image of a man, especially combat veteran.” Research suggests that women veterans may be especially “invisible” within American society and that people perceive that women veterans’ military experiences differ from those of men (Best, Hunter, and Thomas Reference Best, Hunter and Thomas2021). It is possible, then, that while veteran status is likely to increase leadership evaluations for both men and women (Hardy et al. Reference Hardy, Coker, Funk and Warner2019), the ultimate leadership evaluation may still be higher for men who are veterans. Indeed, this is the expectation we preregistered prior to fielding Study 2: being a veteran would generally increase leadership perceptions, but men who are veterans will be perceived as stronger leaders than women veterans.
Gender, Security, and Authority
Leadership, however, is one outcome; another is policy persuasion (authority). Research suggests that messages are more persuasive when the source is credible (Druckman Reference Druckman2004; Ryan Reference Ryan2013).Footnote 4 A source is viewed as credible when people believe that the source is both capable of and intends to carry out some behavior (Madsen Reference Madsen2019). Could military bona fides overcome these issues by signaling greater capability? Prior to fielding Study 2, we preregistered this idea: having bona fides would help a politician be persuasive on security—that is, have authority.
Yet persuasion also depends on other characteristics of the source (Chaiken and Maheswaran Reference Chaiken and Maheswaran1994; Druckman Reference Druckman2022). A source’s credibility, for example, may also depend on context, be it a policy area or times of heightened international threat. Given the association between security and masculinity, then, women may be disadvantaged in the security context (Schramm and Stark Reference Schramm and Stark2020). The gendered idea of the security state may be so pervasive that even when women manage, in Disraeli’s words, to climb to the top of the greasy pole, “international politics is such a thoroughly masculinized sphere of activity that women’s voices are considered inauthentic” (Tickner Reference Tickner1992, 4). Therefore, while military bona fides increase the ability to persuade, our preregistered expectation was that this again should lead to more success for men than women. Our initial expectations, then, were parallel for both leadership and authority.
EMPIRICAL STRATEGY
The intersection of gender and securitization suggests two outcomes. One is the role of gender and military bona fides in determining perceived leadership. The second is in determining authority—the ability to persuade on a security policy. We track the connection between these different factors using experiments (Druckman et al. Reference Druckman, Green, Kuklinski, Lupia, Druckman, Green, Kuklinski and Lupia2011; Hyde Reference Hyde2015).
We rely on trait evaluations to consider the extent to which a politician is perceived as a “leader.” This is a desired trait in elected officials which, while deliberately vague, contains the potential for masculine and militarized biases (e.g., Holman, Merolla, and Zechmeister Reference Holman, Merolla and Zechmeister2022). Second, we consider measures of the candidate’s ability to shift policy preferences—that is, “authority.” We note that our goal is not to track whether a person will vote for a particular candidate. Rather, our goal is to consider whether a candidate can persuade people through a security statement. We test these ideas using two national samples of American adults.
Our first, preliminary study considered the intersection of gender and a securitizing statement. Participants in this study were randomly assigned to either a man or a woman candidate and saw either a hawkish or a neutral statement from the candidate. In this study, the hawkish statement serves as both a signal of military bona fides and the securitizing statement. The goal of this first study is to track the effect of the statement on our two key outcomes: leadership (measured through an evaluation of the candidate as a leader) and authority (measured as the extent to which the candidate persuaded the participant).
Among the first study’s limitations, the most important is that a single statement cues both bona fides and securitization. Therefore, we turn to a second, main study, which differs in several important ways. First, it tests a series of preregistered hypotheses. Second, it separately randomizes the military bona fides cue and the securitizing statement. Rather than relying on a hawkish statement to signal bona fides, we use a cue that is less partisan and that research suggests to be quite powerful: previous military experience (Golby, Feaver, and Dropp Reference Golby, Feaver and Dropp2018). Finally, given that people are more likely to follow a co-partisan’s cues (Berinsky Reference Berinsky2007) and that partisanship profoundly and simultaneously affects assessments of gender and security (Holman, Merolla, and Zechmeister Reference Holman, Merolla and Zechmeister2011), we address partisanship more directly in the second study.
Both studies share the same basic approach to balance between internal and external validity. We present the participants with a fictitious first-time candidate for the U.S. Congress, and gender is cued through candidate images, names, and pronouns.Footnote 5 In the United States, both constitutionally and in the popular perception, the President is by far the actor most likely to engage in securitizing. Other executive officials such as Secretary of Defense, and perhaps more visible legislators such as the Speaker of the House, may also engage in this behavior to some extent. However, presenting study participants a “generic” president or defense secretary poses several problems, not the least of which is that there never has been a woman in either position. Further, one’s attitude to well-known figures such as Nancy Pelosi or Mitch McConnell, much less Donald Trump or Joe Biden, is likely to be heavily pretreated (Druckman and Leeper Reference Druckman and Leeper2012). Therefore, to track the role of gender in this process, we rely on a design that allows us the most control with minimal pretreatment effects.
Study 1: Preliminary Results
Our first study randomized participants along three factors: (1) candidate gender, (2) “hawkish” or neutral statement, and (3) candidate party, producing a total of eight randomly assigned treatments (Figure 1). We focus on the relationship between the gender of the candidate and the statement. Specifically, we assess whether making the statement affects people’s perceptions of the candidates’ leadership characteristics and whether people will follow the policy in the statement. We have no a priori expectations about the role of partisanship but include partisanship as a factor to ensure external validity (i.e., people rarely encounter candidates for national office without also encountering their party) and to avoid confounds as people often assume that women are Democrats (Dolan Reference Dolan2005).
Figure 1. Experimental Tasks, Study 1
For this study, conducted via the Internet, participants saw a treatment as an article on a screen. Our participants (
$ N=1,\hskip-0.15em 029 $
) were members of a national adult panel provided by Survey Sampling International (SSI; since re-branded as Dynata); SSI samples approximate the U.S. population.Footnote 6
Participants were given a sample news article about a generic candidate for Congress laying out their views on foreign policy. Within the text, depending on the treatment, the candidate is described as a “rising star” in the Democratic (Republican) Party, presented in the text and accompanying photo as a white man (woman), and gives a Defense (Neutral) statement.Footnote 7 In this study, the treatment includes both a statement of threat and of a need for security; we therefore interpret the results in light of both.Footnote 8
We first consider whether assessments of politicians’ leadership change when they invoke defense. To consider leadership, we asked participants to rate how well they believe the politician in the treatment fits the description “strong leader” on a seven-point scale, where a score of 1 means that the term does not describe the politician well at all, to 7, which means the term describes them very well. Table 1 presents the results by partisanship/gender of the politician, as well as by gender only (treating partisanship as a nuisance factor given randomization). We see that men are rated more positively than women, but mentioning threat and defense increases women’s ratings—but has no effect for men.Footnote 9
Table 1. Change in Believing Candidate Is a “Strong Leader,” Study 1
Note: Each comparison is a two-tailed t-test. Outcome measure: mean belief that a politician is a strong leader, on a 1–7 scale.
$ N=1,\hskip-0.10em 009 $
, but more participants were randomized to the conditions with women.
We also asked “What is the most important issue of the day?” Participants were asked to select just one from a set of issues, and we present these results graphically in Figure 2.Footnote 10
Figure 2. Defense Importance by Candidate Treatment, Study 1
Note: OLS coefficients shown in SI D.1. Republican women: difference = 0.037 (
$ p=0.31 $
); Democratic women: difference=
$ -0.003 $
(
$ p=0.94 $
). Women overall (both parties): difference = 0.010 (
$ p=0.70 $
). Republican men: difference = 0.151 (
$ p=0.02 $
); Democratic men: difference = 0.093 (
$ p=0.08 $
). Men overall (both parties): difference = 0.12 (
$ p=0.004 $
).
As a check, we find that there are no differences in the percentage of participants who list defense as an important issue in the neutral conditions. Participants who are randomly assigned to the neutral treatment report that defense is important at equal rates regardless of the gender of the candidate (
$ p=0.69 $
).Footnote 11 We next consider how participants in the defense treatments differed from those in the neutral treatments.
First, we see that women have little influence on perceptions of defense as an important issue. When the candidate is a Democratic woman, the treatment has no effect on whether participants find defense to be an important issue. In contrast, where the politician is a man, mentions of threat/defense increase the belief that defense is an important issue (Figure 2).Footnote 12
Study 1 suggests that when women send a bona fide signal—through invoking threat/defense—they are perceived as stronger leaders. On the other hand, women are less likely to be able to “securitize” with that very same statement.
Study 2: Experimental Design
Our first, preliminary, study demonstrates a leadership benefit to military bona fides, but does not suggest a parallel one for authority. Yet it is possible that this difference is due to the study’s considerable limitations. Thus, we turn to a second study, this article’s focal experiment. While Study 1 combined the securitization statement and defense bona fides as a single treatment, Study 2 (
$ N=1,\hskip-0.15em 665 $
, recruited through Research Now Survey Sampling International (RNSSI) sample statistics in SI E.2) distinguishes a candidate’s securitization statement and the bona fide cue as separate treatments.Footnote 13 We use veteran status as it has been suggested as a powerful cue about military bona fides for political candidates (McDermott and Panagopoulos Reference McDermott and Panagopoulos2015; Richardson Reference Richardson2022).
A second difference between the preliminary and focal studies is how we address political party. Since partisanship remains a crucial factor in participants’ perception of candidates, but balancing the power and parsimony of our study, we hold party constant by using a primary election context and sorting participants by their party identification at the outset. Therefore, participants who identify as Republicans (strong, weak, and leaning) see a Republican primary, while those who identify as Democrats are assigned to a Democratic primary. Those who identify as “pure” Independents, and people who did not give a party, were randomly assigned to either a Republican or Democratic primary.Footnote 14
We do this for several reasons. First, defense has generally been considered a Republican-owned issue (Petrocik Reference Petrocik1996). As a result, if we ignored party entirely, a participant might assume that a politician speaking about defense was a Republican. This assumption would then muddy the comparison between the defense and neutral conditions. We avoid this potential for bias by deliberately clarifying partisanship in all conditions. Second, not only does partisanship sometimes overwhelm gender effects (Bauer Reference Bauer2018; King and Matland Reference King and Matland2003) but people often assume that a woman candidate is a Democrat (Dolan Reference Dolan2005). Finally, asking participants to rate only in-party candidates offers a more externally valid approach. Over the last decade, Americans have become highly unlikely to follow cues from the opposing party (Druckman et al. Reference Druckman, Klar, Krupnikov, Levendusky and Ryan2021). As a result, the authority to set the agenda—of the type we theorize here—likely happens within party.
Study 2 randomized participants along three factors: (1) gender, (2) combat veteran status (e.g., military bona fides), and (3) securitization speech act. The gender cue in this study is identical to that of the first. We cued combat veteran status within the treatment by telling a randomly selected group of participants that “[Candidate name] is a combat veteran.”Footnote 15
Our final factor randomizes whether the participant received a statement from the candidate. In cases where the participant is randomly assigned to receive a statement, the candidate’s securitization statement draws attention to a need to “increase our security” and proposing a policy. We cued international competition by mentioning three countries widely viewed as adversaries by most Americans: China, Russia, and Iran. In Study 1, the defense treatment had every candidate make an explicit hawkish statement to signal bona fides. Here, our goal was to have a statement that was not hawkish as much as focused on connecting a policy idea to national security—that is, securitizing. Moreover, to further measure the ability of security discourse to shift “normal” political preferences, we deliberately made the statement run counter to the relative preference most partisans place on defense and infrastructure spending. Thus, for Democrats (and randomly selected Independents), the securitization treatment consisted of the following statement:
The candidate released the following statement on the federal budget: “shifting resources away from infrastructure and towards defense will grow the economy by creating 120,000 additional jobs and increase our security by allowing us to better compete internationally against China, Russia, and Iran.”
This represents a clear securitization move—a budgetary decision framed through national security. The Republican treatment is an even tougher test; it uses the language of security to get respondents to change a core Republican belief and shift resources away from defense. Republican participants (and randomly assigned Independents) received the following treatment (italics added here to emphasize the key partisan difference):
The candidate released the following statement on the federal budget: “shifting resources away from defense and towards infrastructure will grow the economy by creating 120,000 additional jobs and increase our security by allowing us to better compete internationally against China, Russia, and Iran.”
Clearly, while both statements have a securitization element, there is a substantive difference in policy goals. Therefore, to ensure that this does not produce additional confounds—for example, if one statement seems more unexpected given the source (e.g., Baum and Groeling Reference Baum and Groeling2009) or if response to this unexpectedness depends on source gender (e.g., Vraga Reference Vraga2017)—we conducted additional post hoc checks (SI E.4).Footnote 16 We do not observe evidence of this type of confound. We present the full set of experimental tasks in Figure 3.Footnote 17
Figure 3. Experimental Tasks, Study 2
Study 2: Preregistered Hypotheses
Building on our first study, the design of Study 2 leads us to a set of preregistered expectations.Footnote 18 We noted these expectations earlier in the manuscript, but specify the hypotheses here as well.Footnote 19
Hypothesis 1 (leadership): Candidates (across both genders) who are combat veterans are perceived as more likely to be strong leaders relative to candidates who are not combat veterans.
Next, we follow our theoretic arguments, which suggest that while being a veteran is generally beneficial for all candidates, the advantage may differ based on gender:
Hypothesis 2 (leadership): Men who are combat veterans will be perceived as stronger leaders than women candidates who are combat veterans.
In other words, we expect that women will benefit from military bona fides (H1) relative to women who cannot signal bona fides, but we acknowledge the possibility that the benefit may not be as strong as it is for men (H2).
Our next two preregistered hypotheses address candidates’ ability to shape respondent policy preferences—their authority—via their statements. While H3 addresses the basic premise that military bona fides increase authority, H4 is the intersection of gender, military bona fides and authority.
Hypothesis 3 (authority): Candidates (across both genders) who are combat veterans will be more successful at changing policy preferences than those who are not combat veterans.
Hypothesis 4 (authority): Men who are combat veterans will be more successful at changing policy preferences than women candidates who are combat veterans.
STUDY 2: RESULTS
We first consider how participants evaluate politicians’ leadership, and whether these evaluations are affected by a politician’s gender and veteran status (preregistered H1 and H2). Second, we consider how gender and military bona fides shape authority with a securitizing statement (preregistered H3 and H4).
We note that our experimental design produces more potential comparisons than the key hypotheses we preregistered. Given the possibility of these other comparisons (e.g., we can test the way making a statement affects perceptions of leadership, though we have not preregistered any hypotheses or analyses to this end), we conduct additional exploratory analyses which consider all possible interactions of our experimental factors. We also include the means across all variables by condition in SI F.1. Given our specification of treatment effects as interactions, we present the results as marginal effects of a shift in one of our treatments.
Candidate Leadership
We begin by considering perceptions of the candidate’s leadership. As with Study 1, our measure is a rating of how well the politician in the treatment fits the description “strong leader” on a seven-point scale, where 1 means that the term strong leader does not describe the politician well at all, and 7 means that the term describes them very well. We also asked about favorability and qualification for office with results shown in SI E.5. In addition to these direct measures of leadership, we also ask participants two measures as checks: perceptions of knowledge on defense and on the economy. We use these to ensure that veteran status is leading people to focus on defense; results are shown in SI E.6. To give a sense of the baseline across the entire sample, while the male candidate has a higher mean leadership evaluation than the woman, the difference is not significant (0.07,
$ p=0.32 $
, two-tailed).
H1 predicted that being described as a combat veteran should be beneficial for both men and women.Footnote 20 As this hypothesis does not specify any particular role for our third experimental factor—the candidate’s statement—we follow previous research and treat it as a nuisance factor in our first set of analyses, a benefit of randomization (e.g., Druckman et al. Reference Druckman, Klar, Krupnikov and Ryan2022).Footnote 21
Figure 4a shows the marginal effect of the candidate being a combat veteran; Figure 4b presents the mean leadership values. The figure demonstrates that for both men and women, the effect of being a combat vet is always positive, though it is only significant for women. As SI E.5 shows, these effects are robust across other measures of candidate traits and to controlling for demographic pretreatment measures. Overall, the results in Figure 4 suggest that a candidate is better off being a combat veteran than not in terms of leadership perceptions regardless of gender.
Figure 4. Study 2, Candidate Leadership due to Veteran Status for Men and Women Candidates
Note: Left panel: marginal effects of veteran status by candidate gender; coefficient estimates (OLS) are in SI E.5. Positive effects mean that veteran status increases perceived candidate quality; negative mean a decrease; 95% confidence intervals and one-tailed p-values (following hypothesized directional predictions), seven-point scale. Right panel: mean leadership by gender and veteran status.
Figure 5. Study 2, Gender Differences in Perceived Candidate Leadership for Veteran and Nonveteran Candidates
Note: Marginal effects of gender by candidate veteran status; coefficient estimates (OLS) are in SI E.5. Positive effects mean that women are perceived more positively than men; negative mean that men are perceived more positively than women; 95% confidence intervals, seven-point scale.
We also hypothesized that men who are combat veterans will have higher leadership evaluations than women who are combat veterans. As the results in Figure 4 hint, we do not see evidence to this point—contrary to our preregistration. Indeed, there is almost no difference between leadership evaluations of men and women who are veterans.Footnote 22
In summary, we see two patterns. First, study participants give higher leadership ratings to veterans; they also rate veterans more positively on other trait dimensions. Moreover, the results on our check measures—SI E.6 —show shifts in perceptions of defense knowledge but null effects for perceptions of economic knowledge, suggesting that being described as a veteran does bring to mind defense and security. This reinforces the idea that veteran status signals military bona fides.
We find no evidence in support of our second preregistered hypothesis: there are no differences in participants’ ratings of men and women veterans. Women veterans are just as likely as their male counterparts to be viewed as “strong leaders.” In a research question posed in our preregistration, we asked about the possible gender differences in the effect of being described as a veteran. The boost from being a veteran is significant only for women, although the confidence intervals overlap.
Candidate Authority
Our next step is to consider whether military bona fides also influence authority. We therefore turn to our second set of analyses that track the effect of the securitization statement on people’s posttreatment issue preferences.Footnote 23
As our securitization statement either advocated for security through infrastructure or security through defense, we measure the extent to which a candidate is an authority by presenting the participant with a zero-sum choice between these two priorities (Caverley and Krupnikov Reference Caverley and Krupnikov2017). In our analyses, we code participants who select the option that matches the statement advocated by the candidate they were assigned to see in the treatment as 1, and those who do not as 0. An additional measure—the perceived importance of spending on defense versus infrastructure—was also included; we note the results in text, and present them in full in SI E.7.
We first test whether the statement is more effective when made by someone with military bona fides. Here, results in line with H3 would show that people are more likely to follow the statement when that candidate is a veteran. We then test whether there are gender differences in the extent to which military bona fides help with this authority. Here, we consider how the intersection of gender and veteran status influence candidates have the most authority, our preregistered H4.
Veteran Status and Authority
We begin by considering whether veteran candidates have greater authority when making statements about national security. Here, we focus on the marginal effect of making a statement on support for the statement’s position, differentiating between candidates who are veterans and those who are not. For now, we do not consider gender—we will address the intersection of veteran status and gender more explicitly later on. We see little evidence that people are more likely to follow the position of a candidate who is a veteran rather than one who is not—contrary to preregistered expectations.
If participants view a candidate as authorized to speak about security, then participants who read the candidate’s statement should be more likely to adopt the espoused position relative to those who were not exposed to the statement. A signal of authority would be a positive, statistically significant marginal effect of the candidate’s statement. As we show in the top panel of Figure 6, we do see that exposure to the statement has a positive marginal effect on support, but only when the candidate is not a veteran. While for nonveteran candidates the marginal effect of the statement is both positive and significant (
$ M=0.08,p=0.015 $
), for veterans the effect of the statement is null (
$ M=0.03,p=0.435). $
This is somewhat unexpected given previous research, as well as the fact that respondents regard veterans as more knowledgeable on defense issues (SI E.6).
Figure 6. Statement on Policy Support, Veteran Status, Study 2
Note: Left panel: marginal effects of making a statement on support for the position in the statement by veteran status. Positive marginal effects mean that exposure to the statement increases support for the position in the statement. Right panel: predicted probabilities of supporting the position in the statement. Both figures show 95% confidence intervals. Full coefficients from the OLS model which produced these estimates as well as a logit model showing similar results are in SI E.7.
A possible explanation for these results could be that participants in the veteran conditions are already primed by the candidate’s status as a combat veteran, thereby undermining the effectiveness of the statement. We address this potential alternative explanation by considering mean support for the policy in the statement by condition (Figure 6b). Some evidence of this alternative explanation would be that people assigned to veteran conditions generally differ regardless of exposure to the statement.
We see no evidence for this alternative explanation. First, when we compare participants not exposed to a statement, we see that exposure to the veteran treatment shows no significant differences relative to those seeing the nonveteran.Footnote 24 Indeed, the greatest differences emerge in response to the statement. While the statement has a strong effect on policy support when it is made by a nonveteran candidate, it has little effect when made by a veteran.Footnote 25
Gender and Authority
Our next step is an exploratory analysis (which was not part of our preregistration) considering the relationship between authority and gender. First, we note that in the conditions without a statement, there are no differences in policy preferences between participants who see a woman and those who see a man.Footnote 26 Relative to a condition with no statement, making a statement has no effect on participants’ policy preferences when the candidate is a woman (
$ M=0.003 $
,
$ p=0.941 $
), but when the candidate is a man, making a statement produces a significant increase (
$ M=0.11 $
,
$ p=0.002 $
) in support for the position in the statement (Figure 7a).
Figure 7. Effect of Statement on Policy Support, Gender, Study 2
Note: Left panel: marginal effects of making a statement on support for the position in the statement by gender, treating veteran status as a nuisance factor. Positive marginal effects mean that exposure to the statement increases support for the position in the statement. Right panel: mean supporting the position in the statement. Both figures show 95% confidence intervals. Full coefficients from the OLS model which produced these estimates as well as a logit model showing similar results are in SI E.7.
We can consider these patterns in a different way. In Figure 7b, we present the probability of supporting the position in the statement by condition. The highest probability support is in the condition where a man has made the statement. This level of support is different from the level of support offered when a woman has made the statement at
$ p=0.039 $
. As we noted earlier, there are no gender differences in level of support when no statement is made.
Gender, Veteran Status, and Authority
To this point, we consider levels of authority by both candidate veteran status and candidate gender. Contrary to our preregistered hypothesis, we do not find that candidates who are veterans have more authority. We now turn to the intersection of veteran status and gender.
In our preregistered hypothesis, we stated the expectation that men who are veterans should have the highest authority relative to other types of candidates. In our final set of results, we see no evidence to this point.
We present our final set of results (Figure 8) as support for the position in the securitizing statement, by condition. Given that we have eight conditions, for ease of presentation, we depict the results for Men in Figure 8a and Women in Figure 8b. First, we see that (contrary to our hypothesis) candidates who are not combat veterans appear to have more authority. This is true for both men and women. Men who make securitizing statements are not significantly more likely to move people’s positions—and the results are not consistent.Footnote 27 Moreover, we do not see that men achieve significantly higher authority when making statements.Footnote 28 In sum, military bona fides—here in the form of veteran status—do not appear to increase authority for any politician.
Figure 8. Statement and Policy Support, Gender and Veteran Status, Study 2
Note: Figures show support for position in the statement, 95% confidence intervals. Full coefficients from the OLS model which produced these estimates are in SI E.7.
Using our additional spending measure we also do not see any evidence of the power of bona fides (SI Table E.15). Moreover, we also see that the statement has null effects regardless of candidate gender—another suggestion of limited findings in authority.
Additional Analyses and Alternative Explanations
In our preregistration, we consider the possibility that different types of participants may respond differently to our treatments. In the main text, we discuss two covariates: partisanship and gender.Footnote 29 We address partisanship because election patterns suggest Democratic participants treat a woman candidate as a more routine state of affairs than our Republican participants.Footnote 30 Moreover, people may perceive the securitization statement differently depending on the gender and partisanship of the candidate (e.g., Sanbonmatsu and Dolan Reference Sanbonmatsu and Dolan2009).Footnote 31 We consider participants’ gender as there is research to suggest people are more likely to support candidates who share their identity (Dolan Reference Dolan1998). We preregistered no a priori expectations and treat these analyses of subgroups as exploratory efforts.
We present the full exploratory results in SI F.3. In this section, we concentrate on party, because in order to pull respondents from their preferred position with the candidate’s statement, the treatments differed based on the respondents’ party identification (i.e., the candidates asked fellow Republicans to shift spending from defense to infrastructure in response to a threat to the United States). To do so, we consider an interaction between our three factors (Candidate Gender, Veteran Status, and Statement), but estimate separate models by party. We present the results in SI Table F.2, only reporting the substantive effects here.
Among Democratic participants, the effects are null. The effect of the statement for nonveteran women is
$ 0.098 $
(
$ p=0.167 $
); for the veteran women, the effect of the statement is
$ -0.047 $
(
$ p=0.511 $
). For nonveteran men, the change is
$ 0.107 $
(
$ p=0.123) $
, and for veteran men, the change is
$ 0.109 $
(
$ p=0.126 $
). Among Republican participants, regardless of veteran status, women candidates’ statements produce an effect that is not statistically significant. Both types of male candidates increase support for infrastructure over defense, but the effect is only significant for nonveterans (
$ 0.168 $
,
$ p=0.031 $
).Footnote 32
Summing up this exploratory section, even when we limited the analysis to the partisan identity, we assume to be most receptive to a given treatment (i.e., Democrats and women, Republicans and veterans), we see the same effects: veterans have no advantage in changing policy preferences in the name of security.
WHO MAKES 3 A.M. PHONE CALLS?
Before discussing the implications of the article, we want to reinforce the findings. In an experiment that separately manipulated military bona fides and the securitizing statement, we see that military bona fides—communicated through veteran status—did boost perceptions of women as leaders; there is an increase for men as well, though it does not reach significance. Contrary to our preregistered expectations, however, we see no difference in leadership ratings between men and women who are combat veterans. Moreover, contrary to our preregistered expectations, we do not see any evidence that being a veteran is especially helpful in increasing the authority to securitize. If women are disadvantaged when it comes to having authority in security spaces (Holman, Merolla, and Zechmeister Reference Holman, Merolla and Zechmeister2022), our results suggest that military bona fides may not be the ticket to obtaining that authority. We see the same pattern, however, for men as well.
Our preregistered expectations were parallel patterns in leadership and authority, yet it is important to note that these are two distinct outcomes. Indeed, it is likely that the policy persuasion at the heart of authority is a much higher barrier—after all, persuading a person to follow a politician’s policy position may be more challenging than changing their belief that a politician could be a good leader (e.g., Druckman Reference Druckman2022). Given the importance of authority in theories of securitization, further exploring this distinction between leadership and authority as an a priori expectation is key for understanding who can persuade in a security context.
Our results also carry implications for future work on veteran status. Research suggests the possibility that veteran status is, in itself, gendered (Thomas and Hunter Reference Thomas and Hunter2019) and that veteran status is perceived differently for men and women (e.g., Best, Hunter, and Thomas Reference Best, Hunter and Thomas2021). In our study, veteran status did not seem to increase authority for women, yet the same was also true for men. This is an unexpected result given existing theories on the role of military bona fides and security in shaping public opinion. Yet while we focus on veterans, the experimental literature on security policy persuasion operationalizes military bona fides in other ways: “soldier,” “junior/senior officer,” “military general/retired general/retired military officer,” and “Chairman of the Joint Chiefs of Staff (CJCS) and the regional combatant commander” (Golby, Feaver, and Dropp Reference Golby, Feaver and Dropp2018; Jost and Kertzer Reference Jost and Kertzer2023; Motta, Ralston, and Spindel Reference Motta, Ralston and Spindel2021; Tomz, Weeks, and Yarhi-Milo Reference Tomz, Weeks and Yarhi-Milo2020, respectively). Our results, then, do conform with recent empirical work showing a more complicated, contingent relationship between veteran status and politics (Richardson Reference Richardson2022). Jointly, these patterns point to the possibility that there may be more nuance to both bona fides and authority—a fruitful consideration for future research.
In bringing together discussions of security, veteran status, and gender, this article synthesizes three scholarly approaches. First, in the American political context, we consider the constraints facing women who work to influence the political agenda. Engaging with feminist theories, we consider how masculinity and security are inherent to the public’s conception of the state and its leadership. For the Copenhagen School, the consideration of gender and veteran status in our studies speaks to the importance of focusing as much on the speaker and audience as on the speech act itself. Moreover, we suggest experiments as a useful means of addressing important aspects of securitization (and feminist) theory (Baele and Thomson Reference Baele and Thomson2017). To date, empirical work in this field has largely excluded such an approach (Baele and Jalea Reference Baele and Jalea2022; Balzacq Reference Balzacq and Balzacq2011), leading critics within the Copenhagen School to worry about its methodological “stagnation,” especially its tendency to focus on successful securitization outcomes (Potenz Reference Potenz2019).Footnote 33 An experimental approach may help to empirically disaggregate the securitization speech act from the policy itself, a central Copenhagen School insight.
It is also important to acknowledge the constraints of our findings. Experiments must always balance external and internal validity, and we acknowledge that security may not be the foremost factor in evaluating congressional candidates. The primary “securitizer” in the United States is the very recognizable (and historically male) occupant of the Oval Office. Perhaps when a woman achieves this position in the United States, the halo of being formally recognized as an authority figure may shift how we consider securitization. Still, as recent research outside the American context suggests, even when a woman holds an executive office, she may still be limited during times of security threats (e.g., Holman, Merolla, and Zechmeister Reference Holman, Merolla and Zechmeister2022). Nonetheless, we note that this article confines itself to the American case, and the effect we identify may not be universal given variations in threat perception, partisanship, and gender norms around the world.
Future research should also examine how the ideas of leadership and authority affect those who are already in office. The ideas we raise could have implications for political diversion “rally ‘round the flag” attempts (Holman, Merolla, and Zechmeister Reference Holman, Merolla and Zechmeister2022) as well as pursuits of conciliatory policies with international rivals (Mattes and Weeks Reference Mattes and Weeks2019). Many IR theories place tremendous weight on the importance to international politics of statements of resolve to domestic audiences (Kertzer Reference Kertzer2016; McManus Reference McManus2017)—what happens when the characteristics of the leader affect the extent to which the public is willing to follow these statements?
Finally, it is also possible that salience of statements on other international issues may be associated with gender in different directions than presented in this article. Women might be perceived as having more standing to securitize issues such as human trafficking, the environment, or global pandemics.Footnote 34 Hillary Clinton, for example, left her position as Secretary of State with both a reputation as a forceful advocate on international “women’s issues” as well as one of Obama’s more hawkish advisers. Perhaps women may be more effective than men at mobilizing extraordinary, militarized political responses to human trafficking, for example. Yet if this is the case, feminist theory suggests this to be part of the problem.
SUPPLEMENTARY MATERIALS
To view supplementary material for this article, please visit https://doi.org/10.1017/S0003055424000182.
DATA AVAILABILITY STATEMENT
Research documentation and data that support the findings of this study are openly available at the American Political Science Review Dataverse: https://doi.org/10.7910/DVN/OEHBYH.
ACKNOWLEDGEMENTS
The authors would like to thank Nichole Bauer, Lene Hansen, Michael Kenwick, Megan MacKenzie, Oriana Skyler Mastro, Michelle Murray, Sarah Sewell, and Rachel Stein for their essential comments.
FUNDING STATEMENT
This research was funded by a grant from the Kellogg Dispute Resolution Center.
CONFLICT OF INTEREST
The authors declare no ethical issues or conflicts of interest in this research.
ETHICAL STANDARDS
The authors declare the human subjects research in this article was reviewed and approved by the Massachusetts Institute of Technology Committee on the Use of Humans as Experimental Subjects and Stony Brook Office of Research Compliance. Certificates are available in the Dataverse. The authors affirm that this article adheres to the principles concerning research with human participants laid out in APSA’s Principles and Guidance on Human Subject Research (2020).
Comments
No Comments have been published for this article.