2.1 Why Would Female Militants’ Presence Affect Public Support? Processing Information on Militants’ Gender

Militants’ gender can sway public opinion and impact decisions on sponsoring rebel groups. Foreign policy scholarship maintains that the information available to foreign leaders and the public is often disproportionate, with the latter having less access (Baum and Potter, Reference Baum and Potter2008). When making political judgments, people rely on the heuristic cues available to them to compensate for limited information (Popkin, Reference Popkin1994). Especially in conflict situations where uncertainty – about the goals, characteristics, strategies, and actions of actors – is pervasive, people use heuristics to infer the goals and natures of the warring parties.

The gender of militants is an important cue for audiences, to help them form beliefs about the conflict about which they lack context. All social identities – such as racial, ethnic, and gender – are associated with a set of norms that prescribe the expected behavior for members of that category. These norms and expectations influence behavior because they affect individuals’ preferences. The perceived social identity of rebel members thereby works as a shortcut, informing people about the characteristics of the conflict environment, such as how they behave, what they are fighting for, and what they are up against.

That said, women constitute a smaller fraction of combatants in contemporary rebellions. While their noncombatant roles are more common, they participate as fighters in about one-third of armed movements, with 15 percent of rebel groups having a dedicated women’s wing with frontline units exclusively for female fighters (Matfess and Loken, Reference Loken2024; Wood and Thomas, Reference Wood and Thomas2017). Despite their relatively limited presence, how can female combatants convey so much about the entire group’s character, shaping opinions on backing the group?

Social psychology literature suggests that individual traits can serve as powerful cognitive shortcuts for identifying the whole group to which those individuals belong, a phenomenon known as “entitativity” (Campbell, Reference Campbell1958). Entitativity refers to the perception of a collection of people as bonded together in a meaningful unit (Crawford, Sherman, and Hamilton, Reference Crawford, Sherman and Hamilton2002). This perception of group unity prompts observers to use the available information to infer dispositional qualities in the target group, form a cohesive impression of the group, assume consistency across situations, and attempt to resolve any inconsistencies in the information about the group (Hamilton and Sherman, Reference Hamilton and Sherman1996). For instance, spectators at a soccer game, though not an organized group, are often seen as having high entitativity; people attribute traits like aggressiveness to all fans based on the actions of a few, viewing them as a unified – and aggressive – group (Campbell, Reference Campbell1958). Entitativity is a crucial precursor to stereotyping, especially for outgroups with so-called essential identities that are perceived as fixed, like gender or race. This is particularly the case when only a few individuals are visible in a field (e.g., women in the military), where they stand out as “tokens” (Agadullina and Lovakov Reference Agadullina and Lovakov2018; Crawford et al., Reference Crawford, Sherman and Hamilton2002).

In conflict settings, the entitativity mechanism leads to stereotyping that extends characteristics associated with women to the entire rebel organization. The involvement of women as perpetrators of violence disrupts the conventional view of women as inherently peaceful. This creates cognitive inconsistency. Observers may resolve this inconsistency by projecting stereotypical female traits onto the entire rebel group, simplifying complex conflict dynamics into familiar gendered beliefs.

This stereotype projection can take two forms: Observers may attribute traditionally “feminine” traits, such as compassion or nurturance, to the group, assuming that women’s presence makes the group more peaceful. Alternatively, women’s violent roles may clash with gender norms, leading to perceptions of female combatants as deviants and their groups as morally corrupt. How the group is ultimately perceived – favorably or unfavorably – depends on (i) the pervasiveness of gender roles, (ii) the media framing, and (iii) the rebels’ international outreach activities.Footnote ² The pervasiveness of embedded gender norms suggests that people are likely to assign their preexisting gendered assumptions to the rebel group upon seeing women as part of it, even absent information – that is, with a lack of media framing or strategic rebel propaganda. That said, media framing and rebels’ intentional gendered strategies solidify these perceptions. I will now unpack each of these factors.

First, the pervasiveness of gender roles is one way that the process of stereotype projection to an entire group operates. The link between women, peace, and morality is so deeply ingrained that it underpins national identity and shapes notions of citizenship (Peterson, Reference Peterson1992). Scholars posit that concepts of state and citizenship cannot be understood without considering gender; women are envisaged as moral authorities, entrusted with guarding and passing national identity to generations (Yuval-Davis, Reference Yuval-Davis1997). People readily rely on these embedded norms even more in contexts with ambiguity about actors’ responsibilities. Ambiguity encourages cognitive distortion aligned with traditional gender norms (Heilman and Haynes, Reference Heilman and Haynes2005), reinforcing perceptions that women are more peaceful and moral than men. This reinforces the “peaceful woman” stereotype, even when women engage in violence, and shapes beliefs about the rebel group’s behavior (as more peaceful), helping people solve the inconsistency in seeing women as part of a violent group.

Second, media framing amplifies these perceptions by leveraging gendered frames. The mainstream media, along with elites, plays a crucial role in shaping public attitudes toward foreign policy (Entman, Reference Entman2003). The media reports news through specific frames that cue the receiver to contextualize events, using selective issues, words, and photographs to influence how the story is perceived. Despite the diversity of contemporary media outlets, framing usually aligns with entrenched audience predispositions, as the media seeks to satisfy public demand as a strategic actor (Baum and Potter, Reference Baum and Potter2008). Framing a news item is most potent when it is culturally congruent with schemes that most members of society habitually employ (Entman, Reference Entman2003), and emotionally charged news influences foreign policy attitudes more than sole information (Gadarian, Reference Gadarian2010).

While the demand for foreign policy–related news is typically low, especially in the US, several factors capture public interest, such as casualty levels and elite discord (Baum and Potter, Reference Baum and Potter2008). The presence of female fighters is one such case that prompts public attention. They garner more sensational and emotional media coverage than men, which increases their propaganda value for rebels (Nacos, Reference Nacos2005). For instance, in recent conflicts in Syria and Iraq, women have been “hypervisible,” becoming central figures in media coverage (Sjoberg, Reference Sjoberg2018).

Despite pursuing different objectives, mainstream media reporting and rebel outreach strategies often rely on similar frameworks to engage the public; both reinforce and are shaped by existing gendered expectations. In conflict, when women participate as combatants, the Western media in particular portrays them as mothers, victims, or feminists in romanticized narratives that evoke emotions aligned with socially resonant gendered themes (Sjoberg and Gentry, Reference Sjoberg and Gentry2007; Toivanen and Baser, Reference Toivanen and Baser2016). This framing corroborates the gendered framing by rebel groups, emphasizing women’s bravery, motherhood, and peacefulness (Rajan, Reference Rajan2011). This depiction reinforces the “peaceful women” stereotype even when women engage in violence; it downplays the severity of their violent acts, portraying them as victims of harsh conditions (more so than males) or as advocates for gender equality (detailed in Section 3).Footnote ³

Third, rebel groups are usually aware of the propaganda value of female militants, and strategically deploy them for international support. Typically, groups recruit women not just for international outreach but for other advantages as well, including expanding the labor force, building ties with locals, and, as they can more easily bypass security checks, aiding covert operations such as espionage, recruitment, and weapons transport. However, groups are also aware of women’s outreach advantages and make strategic choices to attract external support.

For example, the LTTE leadership was keenly aware of the impact that armed women could have on foreign perceptions of the movement’s objectives and actively promoted them through various media (Wood, Reference Wood2019). Its international branches produced and distributed documentaries, published books and magazines highlighting women’s roles, and arranged media interviews with female militants (Brun, Reference Brun2005). United Nations reports revealed that Nepal’s Maoists inflated female militants’ numbers to almost double for international propaganda purposes (Ortega, Reference Ortega2010). Gerakan Aceh Merdeka (GAM) in Indonesia coordinated with Western media to promote its limited number of female combatants to foreigners (Manekin and Wood, Reference Manekin and Wood2020). The leader of the Kurdistan Workers’ Party (PKK – Partiya Karkerên Kurdistanê, in Kurdish), Öcalan, also recognized the role of women in mobilizing international support. From its early years, the PKK cultivated ties with global women’s organizations, securing the women’s military unit’s (YAJK) participation in the 1995 UN World Conference on Women (Başer, Reference Başer2022). Öcalan frequently credited women with shaping global perceptions, emphasizing in the PKK’s official bulletin that they elevated the group’s appeal and image in Europe and the Middle East (PKK, 2001: 7). He specifically highlights the Kurdish women’s resistance against ISIS as a turning point, stating that their courage significantly amplified international attention for PKK (Başer, Reference Başer2022).

These examples illustrate how the media, rebel strategies, and public attitudes interact. Rebels leverage gendered outreach to attract support, while audiences, influenced by ingrained gender biases, interpret female militants in ways that align with preexisting perceptions. Since media portrayals and rebel messaging are rarely neutral, their interaction likely amplifies the effect of female combatants far more than controlled experimental settings can capture.

However, not all women would give the same impression. A group’s ideology and whether it shares an affinity with one’s country, when such information is available, are important factors shaping perceptions. I expect the theorized relationship to be weak or absent in cases such as women in veils, women depicted as suicide bombers, and abducted women. The veil symbolizes Islam, and female militants linked to groups espousing extremist interpretations of Islam are usually vilified or viewed as mentally ill, as depicted in racialized portrayals in the Western media. Female suicide bombers killing civilians tend to be evaluated negatively compared to women, for instance, engaging in guerrilla warfare. They are often portrayed as mad, or monsters, drastically deviating from feminine nurturing norms (Gentry and Sjoberg, Reference Gentry and Sjoberg2015). These portrayals vilify rather than humanize them, dampening their appeal to audiences. Also, the dynamics discussed here apply more to women who join armed groups voluntarily, as abducted women are less likely to gain support by signaling peace, nonviolence, or gender progressiveness. Uganda’s Lord’s Resistance Army (LRA) gained notoriety for abducting thousands, forcing them into child soldiering or sexual slavery, which undermined support. Yet, without such information, people would likely assume voluntary participation.Footnote ⁴

Women in noncombatant roles are less likely to attract international attention or support because women serving as cooks, cleaners, nurses, or messengers, while essential to the functioning of the group, do not challenge traditional gender norms in the same way as combatants. They are less likely to be framed as markers of progressivism, empowerment, or humanitarian concern than the combatant women who visibly break norms. Noncombatant women are also less likely to raise negative perceptions about military capability, as they do not engage in fighting. Hence, they may evoke weaker emotional or ideological reactions – whether about equality or concern about military power. Overall, while their roles are crucial, they remain largely out of the international spotlight and are less likely to change public opinion. While these are plausible, further research is needed to explore the impact of these various forms of women’s participation in conflict on international audiences. For instance, some noncombat roles, such as nurses, can still trigger humanitarian emotions, especially if they are framed well by advocacy campaigns and rebel lobbyists. This Element, however, focuses on women in combatant, rather than noncombatant, roles, where their involvement in violence sharply contradicts traditional gender norms.

Another question is to whom the rebels are signaling and whether their strategies vary by audience. Rebels can leverage different reputation-building strategies to appeal to different constituencies (Akcinaroglu and Tokdemir, Reference Akcinaroglu and Tokdemir2018). For instance, service provision strengthens legitimacy among locals, while compliance with international law boosts international standing. However, they can seek legitimacy from both, making the efforts interconnected. Rebels seeking local authority also engage with humanitarian actors, as international backing weakens state narratives (Jo et al., Reference Hyeran, Joowon and Barrett2025). In this context, the gender norms of their audience would shape how rebels market female combatants, as they would tailor narratives to align with their targets’ values. For example, the PKK emphasized its gender-progressivism in engaging with Western democracies, while embodying traditional norms in its local communications, portraying women as primarily mothers and symbols of innocence to raise sympathy and reinforce ceasefire credibility (Başer, Reference Başer2022). Szekely (Reference Szekely2020) contends that, in the Syrian Civil War, rebel groups and states deployed female combatants as a cost-effective means of signaling alignment with the US and differentiating themselves from ISIS. They also manipulated the visibility of women, emphasizing or minimizing their presence to appeal to different audiences. As such, we would expect rebels seeking support from liberal democracies to be more likely to convey egalitarian messaging than those courting actors like Saudi Arabia, where such narratives hold less appeal.

Another question is whether there’s a gap between rebels’ marketing strategies and the experiences of female militants. Organizations can embrace gender equality as both a normative commitment and a strategic tool (Başer, Reference Başer2022). Even without genuine commitment, the presence of women fighters can project an egalitarian image, but this risks backlash if seen as inauthentic. Scholars disagree on whether armed groups can truly cultivate gender emancipation. Sixta (Reference Sixta2008) views female militants as First Wave feminists, arguing that they resist triple oppression – Western, societal, and organizational. Others maintain that even groups promoting gender equality often remain patriarchal, abandon reforms post-conflict, and expose women to new vulnerabilities (Mazurana et al., Reference Mazurana, McKay, Carlson and Kasper2002). Recent research shows mixed outcomes: group-level studies find that women can advocate for designing and implementing inclusive peace deals (Brannon and Thomas, Reference Brannon and Thomasforthcoming; Thomas, Reference Thomas2024), while individual-level accounts, like in FARC (Revolutionary Armed Forces of Colombia), highlight how women’s newfound greater agency remains constrained by collective goals and patriarchal (Barrios Sabogal, Reference Sabogal and Camila2021). This agency is also shaped by intersectional factors like class, ethnicity, and education, as in the Nepal Maoists (Giri, Reference Giri2023). This Element doesn’t aim to judge whether rebels’ gendered messaging is genuine or if female fighters are actually empowered, but it argues that associations between women and values like equality, peace, and moderation, amplified by media and rebel narratives, can shape support for their groups.

Overall, insights from social psychology, public opinion on foreign policy, and the media indicate that information about militants’ gender should systematically influence public opinion. It follows that the gender of rebel membership matters for conflict dynamics, as the presence of women makes a difference in how rebel groups are perceived. In the next section, I outline pathways through which women and their rebel groups can shape perceptions and influence foreign support, walking through examples from media framing and rebel outreach strategies.

2.2 How Does Female Militants’ Presence Shape Foreign Audiences’ Opinions? Mechanisms of Support for Female Combatants

After establishing why messages of women combatants would shape public perceptions differently than male insurgents, this section explains how, or through which mechanisms, they affect foreign audience’s opinions. Public support for foreign conflict assistance is influenced by both normative and instrumental concerns (Holsti, Reference Holsti2004). Normatively, individuals are more likely to support interventions framed around humanitarian values – such as protecting civilians or aiding groups perceived as morally just or those linked to their ideologies and core values (Kertzer and Zeitzoff, Reference Kertzer and Zeitzoff2017; Kreps and Maxey, Reference Kreps and Maxey2018). On the other hand, instrumental support is often contingent on expected national benefits or gains on the ground (Gelpi, Feaver, and Reifler, Reference Christopher, Feaver and Reifler2009).

The traits associated with women can potentially impact foreign public support for the insurgency through both normative and instrumental concerns, which can pull the support in either direction. Normatively, groups with female combatants increase public support as they are considered to exercise moderation, pursue just goals, and/or support gender equality. Yet, instrumental concerns can either increase or decrease support. Specifically, citizens’ concerns about their country’s international reputation can increase support, while doubts about practical gains may decrease it, as women are often viewed as unfit for combat.

Building on scholarship on foreign conflict assistance and gendered conflict dynamics, I outline mechanisms through which female combatants can shape public support for foreign armed movements in three groups: (1) humanitarian concerns, (2) ideologies and values, and (3) instrumental concerns. I divide the normative concerns into the first two categories, as doing so clarifies distinct pathways through which gendered assumptions can shape support – namely through ethical obligations surrounding conflict dynamics versus alignment with the group’s values. Table 1 outlines the underlying factors driving foreign public support for conflict assistance, the gendered assumptions amplifying these factors, and the expected support trends for groups with female combatants based on these drivers. To the best of my knowledge, no studies have systematically tested these perceptions about female fighters or how these perceptions increase or decrease public support for their government’s backing of the group.

Table 1Mechanisms of support driven by the gender of the insurgents

Table 1Long description

The table lists mechanisms across humanitarian concerns, ideologies and values, and instrumental concerns. For each category, it highlights core public support drivers, such as concern for civilians, morality, repression, democracy, gender equality, strategic interests, and national reputation, and then shows how women fighters modify these beliefs. Each row concludes with the expected effect on public support, indicating when gendered perceptions increase or decrease support for insurgent groups. Women’s presence can make groups seem less violent, more moral or democratic, and more reputation-enhancing, hence increase support, but can also raise doubt about military strength, which may decrease support.

3.1 Case Selection

The experiment is conducted over two samples: citizens of the US and of Tunisia. The US is the country engaging in foreign conflicts the most. It has a long history of providing various forms of assistance to rebel groups globally, ranging from involvement in Nicaragua, Cuba, Angola, and Iran during the Cold War, to interventions following the “global war on terror” policy in Libya and Afghanistan, to more recent engagements in Syria. These actions often sparked debates within domestic politics, where the opposition and the public have questioned the appropriateness and the consequences of supporting or terminating aid to various groups. Hence, understanding US attitudes on support for foreign conflicts is crucial in its own right and can also offer insights into the perspectives of other democratic audiences across Western nations.

Tunisia offers a valuable case with which to test the generalizability of the findings. On the one hand, like the US, Tunisia has been recognized as a democracy, hence we would expect public opinion to shape policymaking, making it a relevant comparison. The first free and fair elections were held in 2011 after Arab Spring, and a new constitution adopted in 2014 solidified the democratic framework. Tunisia was the only Arab nation classified as “free” in the 2020 Freedom House Index and was often referred to as “a democratic success story.” However, since 2021 Tunisia has been downgraded to “partially free” by Freedom House and reclassified from a “liberal democracy” to an “electoral autocracy” in V-Dem, as President Said curbed the powers of the parliament and the judiciary. This democratic backsliding raises questions about the extent to which public opinion shapes policymaking. Nonetheless, hybrid regimes like Tunisia, where democratic and authoritarian elements coexist, provide an opportunity to examine how public opinion functions within constrained democratic settings. Analyzing Tunisia helps me assess whether the findings apply beyond liberal democracies and to discuss public opinion’s role under conditions of increasing executive dominance.

In both Tunisia and the US, women constitute 16–17 percent of their state’s militaries (US Department of Defense, 2022; US Embassy in Tunisia, 2021), suggesting a similarity in people’s familiarity with women’s involvement in more masculine domains. On the other hand, Tunisia differs from the US in important ways, notably regarding military power and historical involvement in foreign conflicts, the strength of traditional gender norms, exposure to conflict, and the portrayal of female fighters in mainstream news media. These differences may lead to different attitudes toward gender-diverse groups and conflict assistance in general, representing potential challenges for the mechanisms underlined in Section 3. These make the Tunisian case informative about the extent to which the theorized mechanisms travel across contexts.

Particularly, the US has a more formally educated and high-income population where gender-progressive informal and formal institutions are more accepted, while the Tunisian population is generally more conservative. I expect that the strength of traditional gender stereotypes would be greater in Tunisia, providing a good case to test whether women’s involvement in combat creates backlashes. Further, audiences experiencing conflict in their own home country may be more hesitant to support sending conflict assistance abroad as they are more familiar with the devastating consequences of such aid. Tunisia experienced more internal conflict (such as al-Qaeda in the Islamic Maghreb [AQIM] and the Islamic State–linked Jund al-Khilafah-Tunisia [JAK-T]). The US, on the other hand, while facing terrorist attacks (9/11) and intervening in Afghanistan and Iraq, has not faced persistent internal insurgencies involving organized rebel groups within its borders in recent decades.

Moreover, the salience of female fighters in the mainstream media differs significantly for the US and Tunisia. The US supported the YPG (People’s Defense Units – Yekîneyên Parastina Gel, in Kurdish) against ISIS in Syria and Iraq. Numerous news stories featuring the Kurdish female fighters of the YPJ – an all-female brigade of the YPG – circulated in the US, depicting them as feminist heroines and the carriers of Western secular ideals against the radical Islamist agenda of ISIS, praising their bravery and military prowess (Yesiltas, Reference Yesiltas2022). For instance, a New York Times article title read “Women Fight ISIS and Sexism in Kurdish Regions” (Flanagin, Reference Flanagin2014). This gender dichotomy between ‘empowered YPJ’ versus ‘subjugated ISIS women’ dominated Western coverage of the Syrian war (Sjoberg, Reference Sjoberg2018).

On the other hand, in the Tunisian media, female fighters have remained largely invisible; when they are mentioned, it is typically in a negative context. A keyword search involving phrases such as “female fighters, female insurgent, female terrorist, female jihadist, rebels, insurgents, jihadist” from the two most read newspapers in Tunisia – Al Chourouk and La Presse – between 2013 and 2023 yielded forty-six news articles, only thirteen of which focus specifically on female fighters. Most of those forty-six articles focus on the so-called sexual jihad (appearing thirty-eight times), which refers to noncombatant women joining ISIS to provide sexual labor to ISIS men; women are depicted in a very negative frame, condemning their radicalization. Only one article offers a positive portrayal, depicting an Iraqi woman as a symbol of resistance who kills an ISIS leader in revenge after being subjected to “sexual jihad” (Al Chourouk, 2015). Overall, given all these differences, Tunisia provides a useful case for addressing these potential differences and testing arguments’ generalization beyond Western states.

3.2 Survey Experiment Design

I conducted online survey experiments with citizens. Various international audiences can be relevant for rebel groups aiming to conduct public outreach, build reputable images, and attract foreign support, such as international media and journalists, NGOs, international organizations, foreign governments (including politicians, bureaucrats, diplomats), diaspora communities and advocacy groups (including anti-war organizations, peace movements, human rights groups), academic institutions and think tanks, crowdfunding platforms, public figures, and citizens. Due to the breadth of the appropriate sample and for practicality, I conducted survey experiments with citizens. The literature on foreign conflict assistance typically focuses on citizens’ opinions as they influence government policies, media narratives, and international support for these groups (Kertzer and Zeitzoff, Reference Kertzer and Zeitzoff2017).

I recruited 840 US participants via AmazonTurk and 927 Tunisian participants via TGM Research in March–June 2024. The Tunisian sample compares well with census data on key dimensions, such as gender, education, and age. The US sample is slightly tilted toward males and more Democrat-leaning, as with most online surveys, but aligns well with census data on key factors (Table A1). Experimental research often uses AmazonTurk samples, and various studies have agreed on their reliability (Clifford, Jewell, and Waggoner, Reference Scott, Jewell and Waggoner2015).

The experiments involved the same questions in both samples. At the beginning, participants completed a standard battery of demographic questions, followed by a brief questionnaire measuring their gender-equality attitudes, opinions on whether the US/Tunisia should take an active role in world affairs, and news consumption habits. These factors are relevant to the research questions as gender attitudes are expected to influence support for gender-diverse movements, and those who believe that their governments should be active internationally are likely to have stronger opinions on foreign interventions. News consumption measures their interest in foreign policy and exposure to female fighters in the mainstream media.

The respondents were then asked to consider a hypothetical scenario that their country could face in the future, one that is general and not about a real country or group in the news today. Respondents were randomly assigned to either a “female combatant” or a “male combatant” treatment (the control group). Those in the “female combatant treatment” saw news items featuring female fighters, while the control group saw identical scenarios with male fighters (Figure 1). Each treatment included images and a background story for the insurgency, designed as news items in a newspaper, helping respondents visualize the conflict as a situation likely to happen. The photos were chosen for their similarity: combatants holding guns upward among other combatants, likely in military training.Footnote ⁵ Everything in the treatments was the same except for the photo of the insurgents (see Appendix for the survey instrument). The experiment reflects a low-information environment, which typically characterizes the public’s understanding of conflicts abroad. The design deliberately excludes the usual positive portrayals of female combatants in media and rebel narratives, discussed in Sections 2 and 3, to isolate female fighters’ impact on support, hence presents a hard case to find a treatment effect.

Figure 1 Treatments.Footnote ⁶

The news item includes information about a fictitious rebel group fighting against a similarly fictitious US/Tunisian adversary state, asking for support from the US/Tunisia. Brutger et al. (Reference Brutger, Kertzer, Renshon, Tingle and Weiss2023) show that using hypothetical scenarios with fictional actors does not change results compared to real scenarios, supporting the reliability of this approach. The vignette for both treatments is as follows:

Participants were asked to answer a series of questions capturing their attitudes toward the potential US/Tunisian sponsorship of the armed group (outcome measures), and their perceptions of the armed group. Questions in the latter category measure the characteristics attributed to insurgent groups upon seeing the gender of the combatants. How people perceive groups based on the gender of the cadre is interesting and important in its own right, but I am also interested in seeing whether and to what extent these beliefs about women insurgents influence support for conflict assistance, hence I refer to them as mediators in the rest of the study. Chaudoin, Gaines, and Livny (Reference Stephen, Gaines and Livny2021) show that the order of mediators and outcome measures can be consequential for the mediation analysis results. Hence, I randomized the sequence in which the mediator and outcome variables appeared to ensure reliable results. To prevent the order of questions from influencing responses, I separately randomized the sequence of outcome measures and the sequence of mediators within their respective groups as well.

There are four outcome variables to capture the variation in different support types: (1) verbal endorsement, (2) economic support, (3) military intervention, (4) accepting refugees. Examining these distinct support types offers insight into how the presence of female combatants might influence different forms of aid. Verbal endorsement, being the least costly, is likely most sensitive to their inclusion, while costly options like military intervention may be less affected. Refugee acceptance further reveals attitudes toward the broader conflict environment, as it involves humanitarian considerations beyond direct support for the insurgents. Also, in refugee contexts, vulnerability is often portrayed through a gendered lens, with images of women and children symbolizing helplessness and tragedy, usually biasing public opinion against male refugees (Kreft and Agerberg, Reference Kreft and Agerberg2024). Female combatant treatment challenges this “feminization of vulnerability,” likely juxtaposing women’s agency in violence with common portrayals of victimhood, making it noteworthy to assess their influence on refugee acceptance attitudes. These outcomes are assessed by asking respondents to indicate how strongly they agree or disagree with the respective statements: The US/Tunisia should (1) verbally endorse the Moravian insurgent group, (2) provide economic support to the Moravian insurgent group, (3) militarily intervene on behalf of the Moravian insurgent group, (4) accept 500 refugees from the conflict hotspot. Each outcome is modeled as a separate dependent variable and analyzed using ordinary least squares (OLS) regression for US and Tunisian respondents independently.

Moreover, respondents were asked to evaluate the rebel group, through questions such as “In your opinion, to what extent does the Moravian insurgent group use violence against civilians; embrace democratic/gender-equal attitudes; [follow a] tactic of armed struggle [that] is morally justified; [and to what extent would] supporting the Moravian insurgency … be good for the security interests of the US/Tunisia?” (see Appendix for the survey instrument). All questions are measured on a five-point scale, from 1 = strongly disagree to 5 = strongly agree. Again, each question is modeled separately, with US and Tunisian respondents analyzed independently, using OLS regression.

3.3 Results

Do female combatants increase foreign support for sponsoring their groups? Figures 2 and 3 present the main results for the US and Tunisian samples, respectively. I present results without controls in the main text and those with controls (such as age, gender, education, US activity, ideology, news following, and gender attitudes) in the Appendix. Overall, results show that seeing female combatants increases people’s support for foreign insurgencies compared to seeing male combatants.

Figure 2 Effects of combatant gender on foreign public support: US.

Note: The figure represents average support levels by treatments based on Table A2 (Models 1, 3, 5, 7 without controls) in the US sample. Lines represent 95 percent confidence intervals. Dependent variables are measured on a five-point scale, with higher numbers indicating greater approval. Average treatment effects (ATEs) are statistically significant for verbal endorsement (p < 0.01), refugee acceptance (p < 0.1) [and economic support when control variables are included (p < 0.1, Model 4)], but not for military intervention.

Figure 2Long description

Lines represent 95% confidence intervals. Dependent variables are measured on a 5-point scale, with higher numbers indicating greater approval. Average treatment effects (A T E) are statistically significant for verbal endorsement (p<0.01), refugee acceptance (p<0.1) [and economic support when control variables are included (p<0.1, Model 4)], but not for military intervention.

Figure 3 Effects of combatant gender on foreign public support: Tunisia.

Note: The figure represents average Tunisian support levels by treatments based on Table A3 (Models 1, 3, 5, 7 without controls). Lines represent 95 percent confidence intervals. Dependent variables are measured on a five-point scale, with higher numbers indicating greater approval. Average treatment effects are not statistically significant at conventional significance levels.

Figure 3Long description

Lines represent 95% confidence intervals. Dependent variables are measured on a 5-point scale, with higher numbers indicating greater approval. A T Es are not statistically significant at conventional significance levels.

For the US sample, results indicate that the presence of female combatants significantly increases verbal endorsement of the conflict, with an average treatment effect (ATE) of 0.24 (and 0.28 with control variables), which is statistically significant at the 0.01 level (Table A2, Appendix). Similarly, for economic support, the presence of female combatants shows a positive but weaker impact, with a statistically significant effect only when control variables are included (0.13 at p < 0.05). Likewise, the presence of female combatants has a positive and significant effect on accepting refugees from the conflict region (0.15, p < 0.1). That said, there is no significant effect on support for military intervention on behalf of the insurgency, with coefficients close to zero. These results suggest that the presence of female combatants substantively and significantly increases public support for the rebel groups, particularly in areas of verbal endorsement, economic support, and refugee acceptance for American citizens.

The effect sizes are relatively large. The support for verbally endorsing a foreign rebel group increases by 11–13 percent when women insurgents are visibly present compared to the support levels for male combatants.Footnote ⁷ To put it into perspective, this effect is four to five times greater than the effect that political ideology has in supporting US sponsorship (see Model 2 in Table A2 in the Appendix). Also, this effect size is slightly higher than the public’s reluctance to support military strikes against democratic countries compared to autocracies in Tomz and Weeks’ (Reference Tomz and Weeks2013) seminal article (often considered a key empirical foundation of the democratic peace thesis).

The Tunisian sample’s results are parallel to those of the US sample. Overall, seeing female combatants increases support for the rebel group in all four support domains; however, the change in support levels is much smaller, and none of the support indicators are statistically significant (Figure 3). Verbal endorsement, economic sponsorship, support for military intervention, and accepting refugees all show slight increases with ATEs of 0.08, 0.09, 0.09, and 0.03, respectively, but these are substantively small and not significant at traditionally accepted levels (Table A3 in the Appendix). The average agreeableness on any type of support for a conflict abroad is notably lower than that of the US sample, for both male and female combatants. For instance, favoring verbal endorsement for female combatants in the US sample is approximately 30 percent higher than in the Tunisian sample.

This difference between US and Tunisian samples can be related to the factors outlined in Section 3.1, namely low exposure to female combatants in the news items and less propensity to view getting involved in foreign conflict as legitimate, as reflected in their historical foreign policy trends. While support for women combatants is not statistically significant compared to support for male combatants in Tunisia, there is no backlash toward them either.

3.3.1 Heterogeneous Treatment Effects: How Do Respondent Gender and Gender Attitudes Affect Support?

Support for armed groups increases among both men and women when female combatants are involved, with similar response magnitudes. In the US, male respondents’ support for verbal endorsement rises from 3.28 for male combatants to 3.48 for female combatants, while female respondents’ support increases from 2.90 to 3.16 (Figure 4). Although female respondents show significantly lower baseline support levels across all support types, this gendered effect disappears when we control for other factors such as age, education, ideology, foreign policy attitudes, and gender attitudes (Table A4). Also, women demonstrate a substantive increase in endorsement for female combatants, bringing their support levels closer to those of male respondents (with a 0.2 point increase among male respondents, and 0.26 among female respondents). The nonsignificant interaction terms confirm that both genders react similarly to the presence of female combatants across types of support, indicating no substantial gender-based difference in supporting female combatants.

Figure 4 Heterogeneous treatment effects by respondent gender: US.

Note: The figure represents ATEs on different support types across respondent genders in the US sample, based on Table A4 (Models 1, 3, 5, 7 without controls). Lines represent 95 percent confidence intervals. Dependent variables are measured on a five-point scale, with higher values indicating greater approval. Interactions between respondent gender and the treatment are not statistically significant at conventional levels.

Figure 4Long description

Lines represent 95% confidence intervals. Dependent variables are measured on a 5-point scale, with higher values indicating greater approval. Interactions between respondent gender and the treatment are not statistically significant at conventional levels.

Both genders generally react similarly to the presence of female combatants in terms of economic support and acceptance of refugees. However, the presence of female combatants slightly decreases support for military intervention among American male respondents (by −0.08) and slightly increases support among female respondents (by 0.14), albeit this is not statistically significant. Nevertheless, the decrease among men may suggest a potential aversion to military intervention when female combatants are involved. Otherwise, the consistent positive direction suggests a general trend toward higher overall support of both female and male respondents when female combatants are involved.

The results are interesting because the change in support levels upon seeing female combatants is slightly higher among female respondents (except for refugee acceptance) than male respondents. These results contribute to the scholarship finding a consistent gender gap in support of the use of force: In line with the literature, women’s baseline support levels for the insurgency are lower, yet this gap disappears when controls are included. Also, contrary to this scholarship, when women are the combatants, female support for conflict assistance rises in a way that is parallel to how male support levels rise. Social identity theory may explain this; people favor in-groups over out-groups. This deserves attention as gender is not typically considered among the social identity categories shaping conflict support. Future research can explore whether this reflects gender solidarity or holds across different contexts.

The Tunisian sample shows similar results regarding female respondents’ receptiveness to supporting female combatants (Figure 5). The effect of seeing female versus male combatants does not vary significantly between male and female respondents. In other words, both male and female respondents react similarly to the presence of female combatants across the support types, with no evidence of a gendered difference in response to female combatants, except for refugee support, where women show lower support overall. Notably, unlike in the US, Tunisian women do not show significantly lower baseline support for sponsoring groups abroad, challenging the expectations of the women–peace hypothesis.

Figure 5 Heterogeneous treatment effects by respondent gender: Tunisia.

Note: The figure represents ATEs on different support types across respondent genders in the Tunisian sample, based on Table A5 (Models 1, 3, 5, 7 without controls). Lines represent 95 percent confidence intervals. Dependent variables are measured on a five-point scale, with higher values indicating greater approval. Interactions between respondent gender and the treatment are not statistically significant at conventional levels.

Figure 5Long description

Lines represent 95% confidence intervals. Dependent variables are measured on a 5-point scale, with higher values indicating greater approval. Interactions between respondent gender and the treatment are not statistically significant at conventional levels.

Additionally, the presence of female combatants leads to a significant increase in Tunisian males’ support for military intervention (0.17, p < 0.1, with an even stronger effect when control variables are considered, see Table A5). This finding is intriguing as it contrasts with the American men’s decreased support for female combatants. It also counters the expectations of a conservative backlash, which might suggest instead that norms around protecting women may drive increased willingness among men to support military intervention when female combatants are involved.

How do individual attitudes toward gender equality impact support for female combatants? Do conservative audiences disapprove of female combatants abandoning traditional gender roles, leading them to withhold support? Figures 6 and 7 display Americans’ and Tunisians’ support levels by their average gender-egalitarian attitudes, respectively. I measure gender-egalitarian attitudes by asking respondents to what extent they agree with the statement “In general, men are more fit than women to be political leaders.” I measure gender-egalitarian attitudes prior to the treatments: 1 denotes the most gender equal and 5 the most gender unequal attitudes.

Figure 6 Heterogenous treatment effects by respondent gender attitudes: US.

Note: The figure represents average support levels on different support types across respondent gender attitudes in the US sample, based on Table A6 (Models 1, 3, 5, 7 without controls). Lines represent 95 percent confidence intervals. Dependent variables are measured on a five-point scale, with higher values indicating greater approval. Interactions between respondent gender attitudes and the treatment are not statistically significant at conventional levels.

Figure 6Long description

Lines represent 95% confidence intervals. Dependent variables are measured on a 5-point scale, with higher values indicating greater approval. Interactions between respondent gender attitudes and the treatment are not statistically significant at conventional levels.

Figure 7 Heterogenous treatment effects by respondent gender attitudes: Tunisia.

Note: The figure represents average support levels on different support types across respondent gender attitudes in the Tunisian sample, based on Table A7 (Models 1, 3, 5, 7 without controls). Lines represent 95 percent confidence intervals. Dependent variables are measured on a five-point scale, with higher values indicating greater approval. Interactions between respondent gender attitudes and the treatment are statistically significant for economic sponsorship (p < 0.05) and military intervention (p < 0.01).

Figure 7Long description

Lines represent 95% confidence intervals. Dependent variables are measured on a 5-point scale, with higher values indicating greater approval. Interactions between respondent gender attitudes and the treatment are statistically significant for economic sponsorship (p<0.05) and military intervention (p<0.01).

The results indicate that while gender attitudes do not lead Americans to evaluate female combatants differently than male combatants, individuals with stronger gender-unequal attitudes show significantly increased support for overall insurgency (for both female and male combatants) for all support types except refugee acceptance. This effect is statistically significant (p < 0.01), substantively large (0.2 and 0.4-point increase for verbal and for economic and military support, respectively), and robust, even accounting for factors like age, gender, education, political ideology, foreign policy attitudes, and news consumption patterns.

The Tunisian results differ from those of the US (Figure 7). Gender attitudes are not statistically related to Tunisians’ support levels for the overall insurgency. However, individuals with more unequal attitudes tend to show higher economic and military support for groups with female combatants (unlike the US). While the presence of female combatants decreases economic and military support among those with highest gender-egalitarian attitudes, support levels increase among those with stronger traditional gender perspectives by 0.15–0.16 points (Models 3, 5, 6 in Table A7).

This result defies expectations of a conservative backlash; individuals with the most gender-unequal attitudes – those who generally oppose women’s public roles – show greater support for women who step away from traditional gender roles by joining a rebel group. On the other hand, those who embrace more gender-progressive attitudes, those who have normalized women’s public participation, show little change in support when women are fighting in a conflict. This finding highlights a previously overlooked interaction between gender attitudes and the gender of perpetrators in conflict, discussed further in the conclusion, Section 5.

3.3.2 How Do Foreign Audiences Perceive Armed Groups with Female Combatants?

The results so far indicate that the gender composition of armed groups significantly shapes the level of support that foreign publics deem appropriate. Then, do foreign publics apply different standards to groups with female combatants? Do they believe that women fighters’ presence is a sign of a moral or feminist cause, or do foreign audiences focus on the potential strategic benefits instead, such as enhancing their country’s reputation? How does the inclusion of female fighters influence the foreign publics’ perceptions of these groups? Here, I examine the direct effects of the mechanisms outlined in Section 2.2, focusing on how the presence of female combatants shapes perceptions of rebel groups. Specifically, I explore three dimensions: their humanitarian character, assessed through perceptions of (1) the group’s use of civilian violence, (2) its moral justification of its armed tactics, and (3) the level of repression it faces; the group’s ideologies and values, evaluated through (4) its commitment to democratic principles and (5) gender equality; and instrumental concerns for the sponsoring state, including (6) the group’s military capacity and (7) its ability to enhance the state’s security and (8) reputation.

The findings, drawn from both American and Tunisian samples, indicate that the presence of women insurgents generally has a largely positive impact on rebel groups’ image. Contrary to expectations, there is no significant indication that the group is perceived as weaker when women are involved. Instead, when people see women combatants, they assume that the insurgency is more committed to democracy and gender-egalitarian values – this is the most profound impact of female combatants in both countries. They also change foreign audiences’ opinions on the organization’s armed tactics: When women are present, armed groups’ tactics are considered more moral (in both countries) and less likely to be viewed as attacking civilians (especially in the US sample).

I begin by detailing the US results. Figure 8 shows that the perception that the group is egalitarian increases by 0.61 points when women are involved. This increase is more than half a point on a five-point scale, statistically significant and robust to model variations, suggesting that the presence of women combatants is strongly associated with viewing armed groups as promoting gender equality. Second, the presence of female combatants significantly increases the perception that the armed organization embraces democratic principles. The perception that the organization is democratic increases by 0.19 points (and 0.23 points with control variables), suggesting that individuals view groups with female combatants as more inclusive and representative of democratic ideals (Table A8 in the Appendix).

Figure 8 How are groups with female combatants perceived in the US?

Note: The figure represents ATEs of opinions toward rebel group characteristics and sponsor interests in the US sample, based on Table A8 (Models 1, 3, 5, 7, 9, 11, 13, 15 without controls). Lines represent 95 percent confidence intervals. Dependent variables are measured on a five-point scale, with higher values indicating higher evaluations.

Figure 8Long description

Lines represent 95% confidence intervals. Dependent variables are measured on a 5-point scale, with higher values indicating higher evaluations.

Female combatants also make a difference regarding the humanitarian appeal of these groups. When women are involved, people are less likely to think that the organization attacks civilians. Female combatants are associated with a statistically significant reduction in perceived civilian violence by around 0.15 points (p < 0.01). This implies a modest but notable shift in public perception toward viewing these groups as less violent toward civilians. Female combatants enhance the image of the armed groups as inclusive, progressive, and less violent.

Similarly, female combatants enhance the perception that the organization is morally just by 0.13 points, which is significant at p < 0.1 only when control variables are involved. Further, the visible presence of female combatants can lead people to think that supporting the organization is good for the reputation of the sponsoring state. Compared to males, female combatants lead to a minor positive change, 0.13 points (significant at p < 0.1 only when control variables are included, Table A8). Although the change is small, it suggests that the involvement of female combatants may enhance perceptions of the strategic reputation of the US, potentially making its involvement in armed conflicts appear more favorable.

While female combatants are viewed as militarily weaker than male combatants, this effect is minor and not statistically significant. When asked if supporting the group helps US security interests, there is a minor positive effect that is not statistically significant. Similarly, there is no evidence for the prevalence of the thinking that the conditions must be so severe that even women are fighting. The presence of female combatants does not seem to affect how people view the repression levels in the conflict region.

Overall, the results show that the presence of female combatants significantly enhances the image of the armed groups, especially in terms of reinforcing values of egalitarianism, democracy, as well as restraint in violence against civilians. The visible presence of female combatants leads to modest increases in perceptions that the organization pursues morally just strategies and that supporting the insurgency could enhance the US’ reputation.

The results from Tunisia are very similar to those of the US (Figure 9). Tunisians also view the values and tactics of rebel groups more favorably when women are involved, suggesting external validity and the robustness of the results across different audiences with varying levels of gender norms and conflict exposure.

Figure 9 How are groups with female combatants perceived in Tunisia?

Note: The figure represents ATEs of opinions toward rebel group characteristics and sponsor interests in the Tunisian sample, based on Table A9 (Models 1, 3, 5, 7, 9, 11, 13, 15 without controls). Lines represent 95 per cent confidence intervals. Dependent variables are measured on a five-point scale, with higher values indicating higher evaluations.

Figure 9Long description

Lines represent 95% confidence intervals. Dependent variables are measured on a 5-point scale, with higher values indicating higher evaluations.

Similar to the US sample, Tunisians think that groups with female combatants are more democratic and gender-equal, by 0.17 points (p < 0.05) and by 0.24 points (p < 0.01), respectively. Moreover, upon seeing women combatant photos, Tunisians are significantly more likely to think that the organization is pursuing a more moral fight, by 0.14 points compared to those who saw male fighter photos (p < 0.1). The results are statistically significant both with and without control variables.

Other dimensions are not statistically significant but are noteworthy as they defy expectations in not showing negative perceptions toward women insurgents. There is no sign of backlash for sponsoring supporting female fighters or underestimating the military strength of female combatants; on the contrary, groups with female combatants are evaluated in a positive light on all grounds in Tunisia as well (more so than in the US, for instance, in terms of military capacity). Also, female combatants give the impression that the group attacks civilians less than all-male groups (0.11). Taken together, the results suggest that female combatants improve the inclusivity and the humanitarian image of armed groups, regardless of their actual deeds, which attests to their symbolic power in marketing the rebellion to international audiences and patrons.

3.3.3 Causal Mediation Analysis

In this last step of the experimental analysis, I examine whether the shifts in beliefs identified in Section 3.3.2 are important mechanisms through which the treatments affect preferences for lending support to foreign armed groups. I use causal mediation analysis to illustrate whether these beliefs associated with female combatants mediate the relationship between the experimental treatment and support for sponsoring their armed groups. I use the approach described by Imai et al. (Reference Imai, Keele, Tingley and Yamamoto2011) to estimate the treatment effect on each potential mediator and the effect of the mediator on support for intervention, where the estimates are then used to compute the average causal mediation effect. This framework helps differentiate between alternative mechanisms to calculate how much of the treatment effect travels through each mediating variable. The “total effect” comprises two components: the “indirect effect,” which refers to the effect traveling through the mediator variable, and the “direct effect,” referring to the remaining effect of all the other possible channels (Imai et al., Reference Imai, Keele, Tingley and Yamamoto2011). For each mediator, Monte Carlo simulations are used to estimate the effect of the treatment mediated by the proposed mediator variables, allowing for flexible modeling at every stage of estimation, as opposed to, for instance, classical structural equation estimation.

Figure 10 presents the results of the causal mediation analysis for the US, which are based on separate models for each mediator variable. The results show that potential mediators are correlated with the outcome support variables in the expected direction. An exception is the US security interests: I have hypothesized that female combatants would decline support for the insurgency as people would think that siding with females in a conflict may hurt their strategic interests; however, female combatants, again, are thought to improve the strategic interests of the sponsoring state. Another exception is the perception of civilian violence, discussed later. Overall, there are no meaningful differences across the outcome variables; across all four support categories – verbal endorsement, economic support, military intervention, and refugee acceptance – three mediators consistently stand out as statistically significant: gender equality, democracy, and civilian violence.

Figure 10 Causal mediation analysis: US.

Note: The figure represents the effect of the treatment on support levels mediated through each mechanism in the US sample. Mediation analysis was conducted individually on each mediator using Hicks and Tingley (Reference Hicks and Tingley2011). Lines represent 95 percent confidence intervals. Models are estimated by OLS regression.

Figure 10Long description

Mediation analysis was conducted individually on each mediator using Hicks and Tingley (2011). Lines represent 95% confidence intervals. Models are estimated by O L S regression.

The perception that the group with female combatants embraces gender equality accounts for the largest mediated effect of support among other mediators, across all four support categories. Seeing women as combatants increases support for verbal, economic, and military backing of the insurgency by around 0.37, and increases accepting refugees from the conflict zone by 0.3 points through the mediated effect of the insurgency’s perceived gender equality (p < 0.01).Footnote ⁸ Similarly, seeing women combatants increases supporting the insurgency verbally, economically, and militarily by around 0.13 points through the mediated effect of the insurgency’s perceived embracing of democratic ideals,Footnote ⁹ and accepting refugees by 0.09 points (p < 0.01). These results are noteworthy because female combatants on their own did not directly influence people’s support for military intervention (as reported in Figure 8), but when people perceive that these groups advocate for women’s rights and democratic values, they are in favor of intervening in the conflict even militarily on behalf of these groups.

The indirect effects of civilian violence are negative in the US sample, differing from the expectations. While the presence of female combatants leads to the perception that the rebel group uses less civilian violence (as in Figure 8), the mediated effect through this perception is negative, as people express lower support for the group.Footnote ¹⁰ This might be related to associating a reduction in civilian violence with weakness or ineffectiveness of the group (compared to perhaps more aggressive all-male groups), making them less inclined to support, despite the overall positive moral signal of reduced civilian violence. Indeed, American citizens think that the group with female combatants is militarily weaker compared to the male combatants by −0.11 (albeit not reaching statistical significance: Figures 8 and 10, and Tables A8 and A10 in the Appendix). Importantly, the negative mediated effect of civilian violence is driven by male respondents; female respondents’ support levels are not affected by perceptions of civilian violence (see Figure A2 in the Appendix). Other potential mechanisms, such as an increased concern for government repression, rebels’ military capacity, or concerns for sponsors’ reputation or security interests, are not statistically significant mediators in the US sample.

The results from the Tunisian sample closely mirror those of the US sample (Figure 11). Tunisians who believe that groups with female combatants are more gender equal and democratic are more likely to favor getting involved in the conflict and supporting the insurgency. The effect of female combatants, mediated through perceptions of the insurgency’s gender equality and democracy, increases support by approximately 0.07 and 0.06, respectively (p < 0.05). Notably, this effect is statistically significant across all support categories, including military intervention, which received low support initially in the absence of the mediated effect of the gender-equal and democratic perception of gender-diverse groups. This means that, while the presence of female combatants alone does not directly lead to favoring military intervention, perceiving them as proponents of democracy and egalitarian values does. Tunisians’ support for gender-diverse armed groups is also influenced by the perception that these groups are fighting a morally just war, a factor that does not apply to male combatants. When seen as fighting for a moral cause, female combatants increase support for the Tunisian government’s verbal endorsement, economic assistance, military intervention, and accepting refugees by around 0.05 (p < 0.1), highlighting the mediating role of moral justification in shaping support for gender-diverse armed groups in Tunisia. That said, the effect sizes in the Tunisian sample are lower than in the US sample (approximately 75 percent for perceived gender equality and 50 percent for perceived democracy).

Figure 11 Causal mediation analysis: Tunisia.

Note: The figure represents the effect of the treatment on support levels mediated through each mechanism in the Tunisian sample. Mediation analysis was conducted individually on each mediator using Hicks and Tingley (Reference Hicks and Tingley2011). Lines represent 95 percent confidence intervals. Models are estimated by OLS regression.

Figure 11Long description

Mediation analysis was conducted individually on each mediator using Hicks and Tingley (2011). Lines represent 95% confidence intervals. Models are estimated by O L S regression.

Other hypothesized mechanisms, such as the strength of the insurgency, government repression, or considerations about the reputation or interests of the sponsoring state, do not seem to mediate the supporting gender-diverse rebel groups; when these perceptions are mediators, the mediated treatment effect is very small, and statistically insignificant in both samples.

The findings are important as they indicate that Tunisians place a high value on gender equality and democracy, to the extent that they are willing to support military intervention on behalf of armed groups that include female combatants. They again challenge the assumption that women stepping out of traditional roles would deter more conservative audiences from offering support. On the contrary, the inclusion of women in combat roles can enhance perceptions of an insurgency’s legitimacy and moral standing, even in more traditional societies.

I further explore whether there is a variation in how male versus female respondents evaluate rebel groups in the US and Tunisian samples (Figures A2.5.1 and A2.5.2 in the Appendix). The mediated effects of the treatment on support levels, run separately for male and female respondents, show that men’s and women’s responses are broadly similar. However, American women’s support for insurgencies with female fighters is more strongly influenced by perceptions that the group’s tactics are morally justified, and that it is democratic and gender-equal; these factors have about twice and one-and-a-half times more influence on women’s support than American men’s. In contrast, gender differences are less pronounced among Tunisians. Both Tunisian men and women value the group’s perceived gender equality about the same when deciding their level of support. However, Tunisian women are more responsive to the belief that groups with female fighters face greater repression, which increases their support, compared to Tunisian men. Apart from these minor differences, the mechanisms driving support for rebel groups are largely consistent across genders in both samples.

In sum, evidence from survey experiments shows that foreign publics are more likely to favor supporting rebel groups when female fighters are visibly present. American citizens, in particular, are more likely to support their government’s decision to verbally endorse, send economic aid to, and accept refugees from the conflict region if the organization includes women militants. This support is largely driven by the perception that the rebel group is more democratic and gender-egalitarian for both Americans and Tunisians. The findings further suggest that, contrary to the assumption in gender in conflict research, people do not think that the repression by the adversary must be severe if women are fighting. Also, there is no evidence that the presence of women signals the weakness of the armed group. Instead, groups with women militants are perceived to use less violence against civilians (especially in the Tunisian sample) and to engage more in moral conduct (particularly in the US sample). The findings align with the research underlining the importance of normative and humanitarian reasons driving support for the use of force abroad (Kreps and Maxey, Reference Kreps and Maxey2018), as opposed to the instrumental gains for the sponsoring state.

While overall support for gender-diverse groups is similar between male and female respondents in both samples, American women show a greater increase in support for the insurgency when they see women involved compared to American men. Similarly, Tunisian women demonstrate a larger rise in support for verbal endorsement than Tunisian men when female combatants are present. These results are important because they prompt a reconsideration of the women–peace hypothesis, which suggests a positive linear link between being women and violence-averse attitudes. Studies suggest that women are less likely to support war or the use of force abroad compared to men; however, they have not tested how women react to conflict when women are the perpetrators of violence (Eichenberg, Reference Eichenberg2003; Wood and Ramirez, Reference Wood and Ramirez2018). In this Element, while women generally show lower baseline levels of support for conflict abroad than men, their support increases more than men’s when they see women insurgents involved in the conflict.

Additionally, the results are largely consistent across both Tunisian and American citizens, defying the expectations of a backlash against female combatants in more conservative societies. In effect, Tunisians with stronger gender-unequal attitudes show higher support for groups with female combatants, indicating that female combatants may enhance the perceived legitimacy of the conflict even more so for more conservative audiences.

These findings suggest that higher support for gender-diverse insurgencies is deeply rooted in gender stereotypes that persist even when confronted with contradictory information that women are perpetrators of violence. Rather than acknowledging women’s capacity for violence, respondents maintain the belief that women are inherently peaceful: They view groups with female combatants as less violent, more morally just, and aligned with inclusive ideologies.Footnote ¹¹ This tendency to cling to preexisting gendered expectations, even when faced with disconfirming evidence, highlights how deeply entrenched these societal beliefs about women’s peacefulness are.

Finally, the treatment design strictly isolates female combatants’ distinct impact on support, stripping away the common positive frames of media portraying them as gender-equality or peace activists, mothers, or physically attractive, discussed extensively in Sections 2 and 3. Male combatants, on the other hand, are typically depicted through technical issues regarding combat effectiveness and conflict dynamics. The survey experiments show that, even in the absence of rebel propaganda or media framing, audiences project their own beliefs onto female rebels and perceive them positively compared to males. This suggests that in real-world settings, female combatants may have an even greater influence on public opinion.

4.1 Public Opinion, Foreign Policy, and Conflict Support

Does public opinion shape foreign policy decisions on external conflict support, and to what extent? First, the public involves many non-state actors: Not only the general citizens but also foreign policy professionals, think-tank experts, academics, business leaders, the media, international organization staff, religious leaders, government officials, military personnel, lobbyists, diaspora communities, and activists can influence the decision-making calculus of leaders. These include, for example, the leadership of the International Committee of the Red Cross (ICRC), which urged the Irish Free State’s Foreign Minister to follow the Geneva Conventions in 1923 (Sivakumaran, Reference Sivakumaran2012), or the Israeli lobby, which plays a decisive role in US foreign policy, or diasporas such as Norway’s Somali community, which directed the Norwegian response to Ethiopia’s intervention in Somalia, with then-Secretary of State Johansen stating, “It is important to have a dialogue with the diaspora [since] Somali voters can determine who gets power in Oslo” (Tellander and Horst, Reference Tellander and Horst2019: 144).

Does public opinion impact elite foreign policy decisions? Literature argues that audience attitudes influence state behavior by shaping leaders’ incentives and constraints (Holsti, Reference Holsti1992; Schultz, Reference Schultz2001; Tomz, Reference Tomz2007). Responsiveness to public opinion and the risk of domestic backlash for unfulfilled commitments – known as audience costs – underpin democratic accountability and democratic peace theory (Fearon, Reference Fearon1994; Tomz and Weeks, Reference Tomz and Weeks2013; Tomz, Weeks, and Yarhi-Milo, Reference Michael R. and Weeks2020). While post-WWII scholars viewed public opinion as incoherent and irrelevant, recent studies with sophisticated methods suggest a consensus that public opinion shapes leaders’ choices (Kertzer and Zeitzoff, Reference Kertzer and Zeitzoff2017). Examples abound: Waning support for the Vietnam War constrained further US interventions (Holsti, Reference Holsti1992). Canadian opposition to the Iraq War prevented troop deployment, despite elite preferences (Goldsmith and Horiuchi, Reference Goldsmith and Horiuchi2012). Intense public backlash after American casualties in Somalia led to the US withdrawal in 1993 (Logan, Reference Logan1996).

Recent research provides direct evidence, by conducting survey experiments with elites, that policymakers adjust their choices in response to public preferences. Israeli parliament members are more likely to support military strikes in Lebanon when public approval is high and believe that the disapproval of the public would entail heavy political costs (Tomz et al., Reference Michael, Weeks and Yarhi-Milo2020). Members of the UK’s House of Commons favor military patrols in the South China Sea more when the public shows greater support (Chu and Recchia, Reference Chu and Recchia2022). When the public favors increasing sanctions against Russia, US foreign policy practitioners are more likely to support stronger sanctions (Peez and Bethke, Reference Peez and Bethke2025). Even non-elected US military officers are more willing to engage in operations in Somalia when the public endorses intervention (Lin-Greenberg, Reference Lin-Greenberg2021).

A large body of work argues that public sentiment matters for foreign decisions, not only in democracies but in hybrid and authoritarian states, as their survival is also dependent on legitimacy in the eyes of the people (Li and Chen, Reference Xiaojun and Chen2021; Weeks, Reference Weeks2008; Weiss, Reference Weiss2014; Weis and Dafoe, Reference Weiss and Dafoe2019). China’s “Wolf Warrior” diplomacy reflects the government’s need to appease nationalist audiences, despite international ineffectiveness (Mattingly and Sundquist, Reference Mattingly and Sundquist2023). As the US pushed for allies in the Iraq invasion, Turkey’s parliament blocked US access due to overwhelming public opposition, despite the initial Erdoğan–Bush agreement, as the parliament “ultimately chose to listen to the voice of the public opinion, not to that of the government leadership” (Goldsmith and Horiuchi, Reference Goldsmith and Horiuchi2012: 567). In the 1960s, Tunisian President Bourguiba’s pragmatic stance toward Israel clashed with domestic sentiment, sparking mass protests during the 1967 Six-Day War, forcing him to avoid full normalization with Israel (Ghiles-Meilhac, Reference Ghiles-Meilhac2014). Pro-Iraq demonstrations in Tunisia during the Second Gulf War “had a considerable impact on the government’s crisis behavior, dissuading the government from associating with Israel” (Abadi, Reference Abadi2017: 519).

That said, audience costs in authoritarian regimes depend on certain conditions, such as regime type (personalist vs. single-party regimes), elite dynamics (military-backed leaders facing greater costs), and media control (Slantchev, Reference Slantchev2006; Svolik, Reference Svolik2012; Weeks, Reference Weeks2012). These suggest that while autocracies face some constraints, democratic leaders are more accountable to the public, the opposition, and civil society. Therefore, I focus on democracies, where public influence on leaders’ decisions is strongest.

With ample evidence of domestic actors’ influence on international affairs, the debate has shifted from whether it matters to how it shapes decisions and which arguments sway publics and elites. Elite factions try to establish a dominant narrative because broadening coalitions legitimizes conflict intervention and minimizes political costs. Given the high costs, reducing internal repercussions of conflict engagement abroad is deemed to be central to foreign policy decision-making (Krebs and Jackson, Reference Krebs and Jackson2007). Maxey (Reference Maxey2020) finds that humanitarian claims, sometimes more effective than security-based arguments, maximize domestic support for sponsorship decisions. Even in polarized contexts, framing interventions as humanitarian increases public approval (Hildebrandt et al., Reference Timothy, Hillebrecht, Holm and Pevehouse2013). Consequently, both rebel groups and foreign governments use humanitarian narratives through public diplomacy to restore legitimacy and preempt elite opposition, to reach not only policymakers but also the broader public through meetings, media campaigns, and opinion pieces (Mattiacci and Jones, Reference Mattiacci and Jones2020).

For instance, Kuwait hired the US PR firm Hill & Knowlton to secure US support against Iraq’s invasion, framing Kuwait as a progressive ally by highlighting women’s ability to drive in Kuwait. Surveys showed that this gendered narrative resonated with US audiences, mobilizing support (Mattiacci and Jones, Reference Mattiacci and Jones2020). Nepal’s Maoists successfully lobbied with India by framing their government as repressive, and held mass solidarity rallies in Delhi to pressure the Nepali government (Mishra, Reference Mishra2024). The Maoists’ inclusive approach toward female militants further bolstered their legitimacy (Giri, Reference Giri2023) (see also the SDF example at the beginning of Section 1). These examples show that rebels recognize the value of cultivating public support to catalyze foreign governments’ conflict assistance.

This does not mean that the presence of female fighters determines the direction of all foreign support, nor do I suggest that it is the sole or primary factor influencing the decision to provide support. States may sponsor rebel groups for a variety of reasons, with the main motive being the rebels’ ability to advance the sponsoring government’s objectives. However, sponsors can be driven by various ideological or domestic motivations (Gelpi et al., Reference Christopher, Feaver and Reifler2009; San-Akca, Reference San-Akca2016), and public attitudes create incentives for conflict involvement, as political costs arise when governments go against these preferences (Tomz and Weeks, Reference Tomz and Weeks2013).

To understand if the survey results reflect a larger trend and have practical consequences in foreign policy decisions to sponsor armed organizations, I analyze whether gender-diverse groups are more likely to attract support from democracies. Given that the public cares about the decency of governments’ actions (Kertzer et al., Reference Kertzer, Powers, Rathbun and Iyer2014), and policymakers are responsive to the public in democracies, I expect rebel groups with female fighters to be more likely to be supported by democracies.

4.2 Data and Research Design

To test the hypothesis, I use the expanded Non-State Actors in Conflict (NSA) dataset by Cunningham, Gleditsch, and Salehyan (Reference Cunningham, Gleditsch and Salehyan2009), which provides information on external state supporters of rebel groups. The NSA dataset covers 1989 to 2009 and contains all the conflict cases in the UCDP (Uppsala Conflict Data Program) Dyadic Dataset (Harbom, Melander, and Wallensteen, Reference Harbom, Melander and Wallensteen2008). The unit of analysis is the rebel organization–adversary state dyad. The support can be in the form of endorsement, non-military (i.e., economic) assistance, military assistance (i.e., weapons, equipment, training), or sending troops on the ground.Footnote ¹³

I capture whether rebels receive support from democratic countries using two measures. The first is democratic support, which identifies the regime type of foreign sponsors. This binary measure captures whether an armed organization ever received support from a democracy in its lifespan. This is operationalized by matching the foreign powers with their Polity scores, with states scoring 6 or higher on the combined Polity IV scale classified as democratic sponsors (Marshall, Gurr, and Jaggers, Reference Marshall, Gurr and Jaggers2017). To assess if female combatants attract a greater concentration of democratic sponsors, I employ a second measure, democratic proportion, operationalized as the proportion of democratic supporters to all foreign supporters. This follows the operationalization of Salehyan et al. (Reference Salehyan, Siroky and Wood2014), who examined the relationship between democratic sponsorship and rebel-inflicted civilian violence.

This measure does not distinguish between overt and covert support; hence, it captures both. This aligns with the theoretical argument. If democratic publics view the group more favorably due to female fighters’ perceived gender equality and democracy, overt support is more likely. However, if female combatants serve as a signal to leaders among other rebels in an environment of information asymmetry – operating through elite perception rather than public opinion – then covert support is also plausible. Since both support types are consistent with the argument, and given the unreliability of covert support data, I do not distinguish between them in the analysis.

To operationalize the independent variable, the presence of female combatants, I use the Women in Armed Rebellion Dataset (WARD) (Wood and Thomas, Reference Wood and Thomas2017), which covers information on female fighters’ presence for a cross-national sample of rebel organizations based on the UCDP Dyadic Dataset (Harbom et al., Reference Harbom, Melander and Wallensteen2008). Based on the UN definition, WARD defines female fighters as combatants “employed in frontline combat, female suicide bombers or assassins, and female auxiliaries or members of civil defense forces who receive military training, carry combat weapons, and could be called upon to participate in combat when necessary” (Wood and Thomas, Reference Wood and Thomas2017: 38). It captures whether a rebel organization recruits female fighters or not. This binary measure ensures focus on their presence, capturing whether women are involved in combat irrespective of their prevalence, which rebel groups often exaggerate to gain international attention (see Section 2.1). I include a second independent variable – female combatant prevalence – to measure whether the proportion of women matters for foreign support. This ordinal variable ranges from 0 to 3 (0 indicates no evidence of female fighters, 1 for less than 5 percent, 2 for 5–20 percent, and 3 for more than 20 percent women in the group from WARD). These measures exclude female members in noncombat positions, such as cooks, nurses, and couriers, which aligns with this Element’s theoretical focus on attitudes toward women who disrupt traditional gender norms. Female fighters attract more visibility and attention because they challenge these norms, while women in support roles are viewed as more typical.

I account for potential confounding factors as various conflict dynamics and rebel group characteristics can influence both state support for foreign armed groups and female recruitment. States are generally reluctant to engage in prolonged conflicts or those with high casualties due to concerns about managing their complexities (Aydin, Reference Aydin2010; Balch-Lindsay, Enterline, and Joyce, Reference Dylan, Enterline and Joyce2008; Regan, Reference Regan2002). These also impact female recruitment; rebels engaged in longer-lasting (Wood and Allemang, Reference Wood and Allemang2022) and high-casualty conflicts are more likely to recruit women to replenish their ranks (Thomas and Bond, Reference Thomas and Bond2015). I account for conflict duration, defined as the number of years since the conflict’s onset, and conflict intensity, measured by the number of battle-related deaths according to the UCDP Battle-Related Deaths Dataset (Allansson, Melander, and Themnér, Reference Allansson, Melander and Themnér2017).

Shifting the focus to rebel group characteristics, stronger groups can be better positioned for external support as they can challenge the target government effectively and ensure security for female participants (Salehyan et al., Reference Idean, Gleditsch and Cunningham2011; Thomas and Bond, Reference Thomas and Bond2015). Troop size is measured using the best estimate of average rebel troop size from the NSA dataset. To account for rebel capabilities relative to the state, following Manekin and Wood (Reference Manekin and Wood2020), I control for very weak rebels, indicating whether the aggregate military capabilities of the organization were “much weaker” than those of the state, from the NSA dataset (both rescaled by dividing by 1,000). Additionally, competition among rebel organizations can influence both external support and the inclusion of female fighters in recruitment strategies. As competition intensifies, groups are pressured to distinguish themselves, modify their ideologies (Tokdemir et al., Reference Tokdemir, Seden Akcinaroglu and Karakoc2020), cater more to sponsor demands, and address the heightened demand for fighters (Bob, Reference Bob2005; Walter, Reference Walter2015). Rebel competition is quantified by the number of groups fighting within the same conflict. Also, armed groups with transnational diasporas can better secure external support, as these communities can lobby their governments on their behalf (Salehyan et al., Reference Idean, Gleditsch and Cunningham2011). The mean of the transnational constituencies’ support is adopted from the NSA dataset. Further, groups that coerce civilians, including women, may project an image of radical violence rather than inclusivity. The hypothesized relationship linking female combatants to increased support may not work if the groups recruit women by force, as they would not be associated with desirable ideologies like democracy or gender equality (Braithwaite and Ruiz, Reference Braithwaite and Ruiz2018). This aspect is captured using a binary measure of forced recruitment from Cohen (Reference Cohen2013), reflecting the use of abduction, press-ganging, or other coercive recruitment strategies.

Finally, I control for two state-level characteristics. I account for the strategic importance of the civil war-experiencing country using the Composite Index of National Capability (CINC) score (Singer, Reference Singer1988). Stronger states can discourage others from funding their insurgents’ activities, or rebels operating against stronger states may be viewed as less likely to be successful (Salehyan et al., Reference Idean, Gleditsch and Cunningham2011). Target state power controls for this deterrent effect of being a powerful country, which I log to manage outliers. Target state regime type is another state-level variable controlling for the Polity score of the adversary state against which the rebels are fighting. Respecting the legitimacy of a democratically elected government has become a rather salient norm in the post–Cold War era, especially for democratic states. Alternatively, operating in a democratic country may affect the rebel group’s ability to establish connections with external actors (Manekin and Wood, Reference Manekin and Wood2020).

4.3 Analysis

I calculate the models using OLS regression, which is more robust for modeling misspecification and distribution assumptions (Angrist and Pischke, Reference Angrist and Pischke2008). The observational data used here is time-invariant because WARD is time-invariant; hence, there is a single observation per rebel group.

This can affect the interpretation of the relationship and warrants a note on endogeneity. My theory suggests that rebel groups with female combatants are more likely to receive support from democratic states, which view their inclusion as rendering the group a less adventurous and more legitimate endeavor, largely because public opinion favors women fighters. This awareness could drive democratic sponsors to influence rebel group composition strategically, having the rebel groups recruit women or at least advertise women after they support the insurgency. This dynamic is consistent with my argument because it indicates that leaders are aware of the legitimization that women bring to armed groups. Conversely, rebel groups recognizing the value of such support may strategically adjust their composition to include female combatants in response to this demand. This again aligns with the essence of my theory, and I expect both processes to be at work. Also, whether rebels strategically manipulate democratic values to gain support or genuinely adapt to them is an intriguing question with implications for stability and credibility, though this lies beyond the scope of this Element. Ultimately, whether democracies support female fighters to cater to public opinion, or rebel groups adjust their composition in response, or sponsors promote female inclusion, all pathways highlight the correlation between democratic sponsors and female combatants as indicators of legitimacy and commitment to democratic ideals within armed groups.

Another consideration is the possibility of spurious correlation: Do rebel groups with female fighters engage in causes that democracies already support? Is it the nature of the cause, rather than the presence of women, that attracts democratic backing? The analysis here does not directly test this possibility. However, historical trends suggest that female participation has been more common among Marxist-oriented leftist groups (Wood and Thomas, Reference Wood and Thomas2017), where gender equality is often central to their ideology. But during the Cold War, Western democracies actively opposed communist movements and backed their adversaries. The US, the most active supporter of rebel groups, primarily funded anti-Marxist forces. This suggests that the causes that groups with female militants fight for have not inherently aligned with causes backed by democracies, which can mitigate concerns about spurious correlation.

The results support the hypothesis. The presence of female combatants is positively associated with both the likelihood of receiving support from democratic states and the proportion of democratic supporters. According to Model 1 in Table 2, rebel groups with female combatants are 19 percent more likely to gain democratic support (p < 0.05). In Model 2, groups with female combatants have an 18 percent higher proportion of democratic supporters (p < 0.05). Model 3 shows that each one-unit increase in the female prevalence category (i.e., moving from 0 to 1, or from 1 to 2) is associated with an 11 percentage point increase in the probability that a rebel group receives support from a democratic sponsor. Model 4 suggests that a higher proportion of female combatants is correlated with attracting a higher proportion of democratic supporters. These findings show that when rebel groups include female combatants, democratic states are more likely to support them, compared to groups without female combatants. This aligns with the experimental evidence suggesting that female combatants signal a form of inclusivity in a way that attracts support from democratic regimes.

Table 2Female combatants and democratic support

Note: Robust standard errors clustered by conflict in parentheses ***p<0.01, ** p<0.05, * p<0.1.

Table 2Long description

The table presents four regression models predicting democratic support or the proportion of democratic support for rebel groups. Key predictors include female combatants and their prevalence. Other factors, such as conflict characteristics, rebel capacity, external support, and target-state attributes, are control variables. Female combatants and their prevalence show positive, statistically significant associations with democratic support across models. The table reports robust standard errors (clustered by conflict), observations, and R-squared values.

I check for the robustness of the models in several ways (see the Appendix). First, I estimate the same models via probit and ordered probit, respectively, instead of OLS (Models 1 and 2, Table A12). Second, I estimate models accounting for rebel group ideology, particularly leftist and Islamist ideologies (Models 3 and 4, Table A12), as these can influence the presence of female fighters and the likelihood of foreign support (Wood and Thomas, Reference Wood and Thomas2017). Leftist groups, which tend to recruit more women, at the same time may alienate foreign powers opposed to their revolutionary agendas. In contrast, Islamist groups are less likely to recruit women, and may influence foreign perceptions negatively, as media portrayals in the West are highly racialized, and female militants in rebel groups pertaining to radical extremist interpretations of Islam are often vilified, ridiculed, or shown as the embodiment of the epitome of jihadist evil, as seen in the coverage of ISIS (Kollárová, Reference Kollárová2015). These portrayals may diminish the chances of favorable foreign support.

Thid, I estimate the models excluding transnational support and very weak rebels (Models 5 and 6, Table A12). I expect transnational support to operate through similar hypothesized mechanisms; transnational constituencies, such as diaspora communities or INGOs, may also be influenced by gender stereotypes that attach humanitarian or normatively desirable values to female fighters. I also estimate models excluding very weak rebel groups to avoid potential collinearity.

The positive and significant relationship between female combatants and democratic support remains consistent across these model specifications, attesting to the robustness of these findings. The results provide a plausibility probe that favorable foreign public opinions toward female combatants can influence democratic leaders’ decisions to support gender-diverse groups. The observational data limits my ability to establish causality, hence the results are correlational rather than causal. Nevertheless, they align with the experimental results and confirm hypothesized expectations.

4.4 Discussion

Using survey experiments, I provide causal evidence that the visible presence of female combatants shapes perceptions of rebel groups as embodying gender equality, democracy, moral legitimacy, and reduced civilian violence, which enhances foreign support for these groups. Observational analysis further indicates that groups with female combatants are more likely to receive backing from democratic states, suggesting that favorable perceptions can translate into concrete advantages. These results raise two important questions: First, is there an objective difference between groups with and without female combatants regarding their adherence to morally desirable wartime strategies? Second, if women’s presence can benefit rebel groups, why do not all groups recruit women, or at least market themselves as having them?