Armed Violence and Patriarchal Values: A Survey of Young Men in Thailand and Their Military Experiences

What is the relationship between armed violence and patriarchal values? This question is addressed with the help of a survey of young men in the conflict-affected southern provinces of Thailand. In Study 1 we find that men with more patriarchal values are more prone to volunteer for paramilitary service. Study 2 uses a natural experiment made possible by the conscription lottery in Thailand to compare survey responses of men who were involuntarily enlisted to do Military Conscription Service (treatment group) with the responses of men who participated in the lottery but were not enlisted (control group). We find no difference between the treatment and control groups in patriarchal values. We conclude that patriarchal values drive voluntary participation in armed conflict, whereas military service as a conscript in a conflict zone does not cause patriarchal values.

W hat is the relationship between armed violence and patriarchal values? This question is addressed with the help of a survey of young men in the conflict-affected southern provinces of Thailand. In Study 1 we find that men with more patriarchal values are more prone to volunteer for paramilitary service. Study 2 uses a natural experiment made possible by the conscription lottery in Thailand to compare survey responses of men who were involuntarily enlisted to do Military Conscription Service (treatment group) with the responses of men who participated in the lottery but were not enlisted (control group). We find no difference between the treatment and control groups in patriarchal values. We conclude that patriarchal values drive voluntary participation in armed conflict, whereas military service as a conscript in a conflict zone does not cause patriarchal values.
W ho decides to take up arms in a situation of violent conflict? This question has mostly been answered in rather general terms, focusing on drivers of conflict that pertain to many more individuals than the relatively few who voluntarily participate in armed violence. However, recent studies have suggested a novel explanation that has the potential to narrow down more precisely who chooses to take up arms in a situation of violent conflict: individuals with patriarchal values. In a series of research briefs, Johnson and colleagues provide evidence that sexist ideas about masculinities and gender inequality were associated with support for violent extremism in Bangladesh, Indonesia, the Philippines, and Libya (Johnston and True 2019;. Bjarnegård, Brounéus, and Melander (2017; find that patriarchal values are associated with actual participation in political violence. Studies of political activists in Thailand demonstrate that masculine honor ideology-that is, the combination of patriarchal values and ideals of masculine toughnessexplains voluntary partaking in political violence. This article moves this research forward in two important ways: (1) it studies voluntary participation in organized violence in the form of long-term commitment to government paramilitary forces in a situation of active armed conflict, and (2) it also investigates the possibility of reverse causality running from participation in organized violence to patriarchal values by analyzing the conscription lottery in Thailand as a natural experiment.
The article consists of two separate but related studies. Study 1 finds that men with more patriarchal values are more prone to volunteer for paramilitary service. Study 2 finds no difference in patriarchal values between a treatment group, consisting of conscription lottery participants who had to do Military Conscription Service, and the control group, consisting of men who participated in the lottery but did not have to serve. We conclude that patriarchal values also drive voluntary participation in armed conflict for government paramilitary forces but that military service as a conscript in a conflict zone does not shape patriarchal values. These findings suggest that causality runs from patriarchal values to participation in organized violence rather than from participation in organized violence to patriarchal values.

UNRAVELING THE CONNECTION BETWEEN ARMED VIOLENCE AND PATRIARCHAL VALUES
During the past three decades, the connection between sex and gender equality on the one hand and war and violence on the other has received increasing attention. The relationship has been studied from both a research and policy perspective at the state and individual levels (for a recent review, see Cohen and Karim 2022). Although many influential contributions have focused on macro-level correlations indicating that more gender equal societies are more peaceful (e.g., Caprioli 2000;Caprioli and Boyer 2001;Dahlum and Wig 2020;Melander 2005), the literature focusing on the individual level has received less scholarly attention despite a long-standing engagement in the question of how gender equality influences violent attitudes and behavior.
This literature has demonstrated that individuals who are more positive toward gender equality also express more peaceful attitudes in terms of foreign policy solutions, attitudes to neighboring countries, and tolerance for religious minorities. Importantly, this individual-level evidence demonstrates that we should not primarily understand this pattern as sex differences but that the relationship holds for women as well as for men Brooks and Valentino 2011;Cook and Wilcox 1991;Conover 1988;Tessler, Nachtwey, and Grant 1999;Tessler and Warriner 1997). Recent research has also demonstrated that gender inequality, measured as patriarchal values, has an influence not just on attitudes to violence but also on an individual's decision to partake in political violence (Bjarnegård, Brounéus, and Melander 2017;. In this line of research, gender equality has been defined and measured in many different ways, ranging from more formal country-level measures (such as the representation of women in parliaments or female suffrage) to measures of women's well-being and status (including fertility rates or women's share of the labor force) and even as individual attributes (such as attitudes toward gendered divisions of labor in the household). As the field has developed, so has the sophistication with which different facets of the concept of gender equality are discerned, disentangled, and assessed (Ellerby 2017;Forsberg and Olsson 2021). Increasingly, for explaining violent behavior, the evidence points to the importance of values, operating at both the societal and the individual levels. Focusing on gender aspects in value systems implies tapping into deep-rooted ideas about the appropriate roles of men and women. Such gendered value systems are intimately connected to the ascribed biological sex of an individual, but they are more flexible and not reduced to a binary. Among groups of men and women, ideas about gender can range from gender equal to patriarchal, favoring a traditional division of roles where men are valued higher than women, particularly in the public realm. Patriarchal values are thus seen as part of a gender order that upholds a fundamental binary between men and women while simultaneously privileging one of the binaries (men) over the other (women). The system has some universal relevance as part of a "global gender order," albeit with significant local variations and applications (Bjarnegård, Brounéus, and Melander 2017;Connell 1995;Ford and Lyons 2012). A move away from patriarchal values toward a more gender-equal value system entails an increasing blurring of boundaries upholding the binary between men and women and a reluctance to privilege one over the other. A focus on patriarchal values and violence brings about a change in emphasis in this line of research: from the assumed peacefulness of women to the potential propensity for violence associated with patriarchal values.
The association between patriarchal values and propensity for violence is increasingly well documented and tends to focus the process of "othering." Othering is a process of differentiation between "us" and "them" and is inherent in the gender order that constitutes the foundation for patriarchal values but is also considered necessary as part of the psychological rationalization involved when any one individual physically harms another (Bjarnegård, Brounéus, and Melander 2017;Hudson et al. 2009;2012). Violence in general-and organized and military violence is no exception-is overwhelmingly carried out by men (Bjarnegård et al. 2015;Goldstein 2003). Although this suggests that the carrying out of violent acts tends to be associated with the male sex and ideals of manliness, it does not follow that all men are equally violent in a biologically deterministic way. Likewise, recent research has reminded us that women are not predestined to be peaceful but that they can be perpetrators of wartime violence, also against other women (Cohen 2013b;Sjoberg 2016). Rather, the understanding of a connection between violence and gender needs to be coupled with an understanding of violence as something that only some men (and even fewer women) aspire to (Bjarnegård, Brounéus, and Melander 2017). It is thus important to continue the investigation of how the men who participate in violence differ from those who do not, and why. Moreover, determining the causal direction between patriarchal values and violence remains a major challenge. The established association could be due either to equality-oriented societies and individuals being less prone to support violent solutions or to societies and individuals exposed to conflict and violence becoming less gender equal as a consequence of this exposure.
Valorization of violence at the individual level is generally attributed to socialization processes. The question we ask here is whether these socialization processes primarily precede or are shaped by military experiences. Some parts of the literature look to childhood experiences and events rather than to military experience to explain early life socialization into individual propensity for violence. Research has demonstrated that violence is transmitted both intergenerationally (Whitfield et al. 2003) and through cultural norms affecting how children are treated (Lansford and Dodge 2008). More specifically, Bjarnegård, Brounéus, and Melander (2019) show that boys who were either beaten or who saw their mother being beaten are subsequently more likely to adhere to patriarchal values as well as to ideals of masculine toughness, and they are ultimately more likely to choose to participate in acts of political violence as adult men. Some North American survey studies suggest that the fact that individuals with a civilian and military background differ in characteristics and values is that young men who have already developed a certain set of pro-military attitudes tend to self-select into joining the armed forces (Bachman et al. 2000;Bachman, Sigelman, and Diamond 1987;Dorman 1976).
However, other accounts connect valorization of violence directly to experiences in the military, whether deeply traumatic war experiences or formalized military training. In this literature, militarized socialization is often thought to foster conformity and obedience in order to accept a combat role (Levy and Sasson-Levy 2008). Studies investigating militarized socialization have demonstrated that the experience of serving in the military per se may affect individuals. For instance, Lindo and Stoecker (2014) find that American men who did military service during the Vietnam War were significantly more likely to be incarcerated subsequently for violent crime. Another study, focusing on Canada, suggests that military training increases social dominance orientation scores among participants (Nicol, Charbonneau, and Boies 2007). On the other hand, Sundberg (2016) studied Swedish soldiers in an International Security Assistance Force contingent before and after a tour of deployment in Afghanistan and concluded that the soldiers' values remained largely stable despite the challenging environment. Importantly, he did find that combat exposure had some minor effects on the soldiers' values, indicating that the severity of experiences should be taken into account. This is in line with Guimond (1995) who demonstrates that both the intensity and stage of the military socialization process matter.
Within feminist security studies, the concept of militarized masculinities has evolved to capture the particular interconnection between militarized violence and masculine ideals (Eichler 2014;Wibben 2016). Because this is the concept that most elaborately connects the two, particularly in a militarized setting, we will here use militarized masculinities as a basis for exploring and unraveling the interconnected concepts of patriarchal values and violence. We look to this literature for guidance on how to distinguish between the valorization of violence on the one hand and the male gender role on the other. Which one is described as affecting the other? Such a scrutiny of the narrower concept of militarized masculinities should also provide valuable clues to the broader puzzle about whether experiences of violence affect one's views of gender equality or, rather, patriarchal values affect one's propensity to participate in violence.
The concept of militarized masculinities takes the notion that a soldier is defined as masculine and even remains a "key symbol of masculinity" as a point of departure (Eichler 2014, 81). Militarized masculinities are commonly said to include the idea that this type of idealized manhood is characterized by "prowess and ability to withstand physical hardships, aggressive heterosexuality and homophobia, celebration of homosociability in the team, deployment of controlled aggression, and completion of assigned tasks with minimal complaint" (Woodward 2003, 44). Although encompassing a quite large set of characteristics, militarized masculinities have two defining features pertaining to our discussion of patriarchal values and violence: their connection to a masculine ideal and the valorization of violence. Militarized masculinities constitute a form of public endorsement or even celebration of the value of traditional military culture, including the use of violence (Hopton 2003), and violence is constructed as the valorized domain of men. To reach a militarized masculinities ideal, a real man is supposed to have warrior qualities, and with these qualities come certain privileges. Women are assigned the role of the soft and nursing contrast category, and men who fail to "man up" are shamed in terms that imply the lowly status of women (e.g., Connell 1995;Enloe 2014;Yuval-Davis 1997).
According to Eichler (2014, 81), militarized masculinity, at its most basic level, "refers to the assertion that traits that are stereotypically associated with masculinity can be acquired and proven through military service or action, and combat in particular." This quote is interesting for us, because it actually points in two different directions, suggesting that traits associated with masculinities can be acquired and proven in a military setting. The first suggestion, that traits are acquired, points in the direction of military socialization-that is, that the experience of going through military service or being in combat will provide men with certain traits. The second suggestion, that traits are proven, instead implies that these are traits developed through socialization processes taking place earlier in life and that certain individuals already possess and are able to demonstrate once in a soldier role or in a combat situation. This ambiguity is present in much literature on militarized masculinities, which tends to view the military and war as spaces-or stages-that simultaneously enable "boys to become men" and leave room for the enactment of masculinities through state-or group-sanctioned violence. This ambiguity certainly suggests important mutual reinforcement, but it also raises questions about the causal direction in the relationship between patriarchal values and violence. In more concrete terms, it points to questions about whether the ideal of militarized masculinity is primarily achieved through self-selection or military socialization. First, we will take a closer look at the suggestion that individuals are socialized to self-select into militarized violence and that military organizations design their recruitment to attract a certain type of individual.

SELF-SELECTION AND RECRUITMENT INTO MILITARY ORGANIZATIONS
Research about self-selection into violence has interpreted military and conflict settings as places where individuals who are already socialized into valorizing both masculinities and violence can act it out or put it to use for instrumental purposes. This perspective builds on the observation that war and violence do not just "happen." States, organizations and, to some extent, individuals self-select into war and to the perpetration of organized violence. Studies of the Sierra Leone civil war (Humphreys and Weinstein 2008) and the Rwandan genocide (McDoom 2013;Verwimp 2005) have found that young men are highly overrepresented among those who self-select into violent participation. Qualitative studies have delved deeper into the reasons for this, seeking to understand the participants and their choices. For example, a study of recruitment to the Viet Cong guerillas in the context of the Vietnam war in the early 1960s listed a number of reasons, Armed Violence and Patriarchal Values: A Survey of Young Men in Thailand and Their Military Experiences including discontent and safety concerns, and specifically highlights that for many young recruits "the desire to win glory, or perhaps just the respect of their community" was the main reason for joining (Donnell 1967, xii). In a similar vein, Wood (2001) argues that to individuals who joined the FMLN insurgents in El Salvador in the 1970s, fighting represented the reassertion of their personal dignity, leading to a sense of pride. Although there is a variety of reasons for joining the military (Duncanson 2013, 151), there are some common denominators that recruiting military organizations draw on. Identification with the masculine soldier is often one of the main motivators for joining the military (Woodward 2003).
Many military organizations depend on attracting volunteers and motivating conscripts, and the idea of transforming young boys into real men is a powerful tool for this (Whitworth 2004). As states and militaries strategically seek to recruit soldiers, they speak to the type of individual who is already likely to be attracted to and fired up by the ideal masculinities that prevail in the armed forces (Eichler 2014). Whom the state seeks to attract depends on both whom they think they can convince and on ideological ideas about what kind of military they ultimately want to build (Strand 2019). From the point of view of the state, idealizing militarized masculinities may be important because they facilitate the recruitment of soldiers who are already more willing to put themselves at risk (Kovitz 2003). When the military constructs itself as masculine, it attracts a certain type of individual. Even if these individuals think that the military will, in some way, change them, they actively seek and desire such change. For many the process of becoming a soldier thus begins long before enlisting, and it is deeply connected with the aspiration to become a "real" man (Woodward 2003). This is in line with Eichler's (2014) succinct summary of what militarized masculinity entails-namely, "the idea that real men are soldiers and real soldiers are men" (90). The view of militarized masculinity as an ideal that motivates individuals' decision to self-select into the military implies that embracing the ideals of militarized masculinity begins before becoming a soldier.
The military has long been a site for the production of a hegemonic form of masculinity that affects society, men, and boys at large (Nagel 2019). In Hearn's (2011, 36) words, "men and militarism are so obviously coupled, that it is hard to know where to start." From this perspective, the construction of militarized masculinities starts long before enlistment and takes place not just in the military but also in society at large. If the origin of militarized masculinities primarily lies outside military organizations, any change to this ideal first and foremost has to be about demilitarizing masculinities rather than about demasculinizing the military (Eichler 2014).

THE POSSIBILITY OF REVERSE CAUSALITY: MILITARY SOCIALIZATION
Socialization is the process through which actors adopt the norms and rules of a given community. Although self-selection into the military suggests that the relevant socialization takes place in early life, military socialization implies that experiences specifically related to the military organization shape individuals' values. The specific militarized and violent environments in which soldiers operate are here suggested to shape and reshape their interests as well as their identities (Checkel 2017). The traditional military has been described as a school of nationhood and the men it fosters as the ideal and privileged citizens (Sheehan 2014). A rich literature investigates such fostering of militarized masculinities among those who serve in the military or among groups of men living in a militarized or conflict-ridden setting. If the primary role of the military is "the formation and mobilization of a body of individuals capable of engaging in military activities" then this requires that ideas and values conducive to this goal are efficiently instilled and impressed in future soldiers (Woodward 2003, 43).
Historically, the military has been "exclusively and aggressively male" and has emphasized manliness defined in terms of physical strength, courage, and violence (Sheehan 2014, 19). Being a male-dominant organization with its function based in violence, the military has encouraged the ideal of militarized masculinities as a means of inducing men to become warriors (Blake 1970). Indeed, military service is often portrayed as transforming "boys" into "men" (Nagel 2019;Yuval-Davis 1997). According to Brown's (2012, 41) study of military recruitment, the U.S. military offers character development and personal transformation, making reference to traditional warrior traits like strength and courage. Ehrenreich (1997) notes that "The difference between an ordinary man or boy and a reliable killer, as any drill sergeant could attest, is profound. A transformation is required" (10).
Such transformation requires specific methods, rites, or practices designed to create soldiers who are willing to kill and die for a cause (Baaz and Stern 2009;Connell 1995;Goldstein 2003). This process, ranging from social pressure and ritualized ceremonies to drills and boot camps, has been described as having the purpose of breaking down the individuality of the soldiers and replacing it with identification with the military institution (Whitworth 2004). The concept of socialization, too, implies more than behavioral adaptation to an environment; it is used to describe a change in how individuals view themselves (Checkel 2017). In this way, the production of militarized masculinities can be very sinister, encouraging violent masculinities that have been linked to the acceptance and rationalization of practices such as sexual violence and rape, as well as homophobia. Military institutions and armed forces typically contribute to discourses emphasizing the role of men as fighters (Baaz and Stern 2009). Baaz and Stern (2009), in their study of wartime rape, explicitly situate the military as a venue where men learn to be violent and masculine, not as a stage where a certain type of masculinities is simply acted out. Because this construction of masculinities in the military rarely resonates with the individual's self-perception, it has to be clearly communicated, strictly delineated, and constantly enforced by institutional practice and individual manifestations. Cohen looks at wartime rape as a method of socialization at the group level, used to create unit cohesion (Cohen 2013a).
However, to what extent military experiences socialize men into militarized masculinities and which types of practices, methods, or experiences that trigger such socialization have not been explored in depth. We know even less about the conditions under which men change in a misogynistic direction as a result of military or combat experiences.

MILITARY MASCULINITIES IN VOLUNTEER AND CONSCRIPTION SETTINGS
It is not always possible to determine whether early life socialization causes individuals to self-select into violent settings later in life or whether they are socialized into a certain logic and behavior as a result of being exposed to militarized violence. Some men are drafted into wars they would never willingly take part in, others see it as their duty and are happy to serve the nation, and for others economic incentives may play a role in their choice to take up arms. Some countries use conscription that leaves young men (and sometimes women) with no choice but to do military service, whereas other countries rely on volunteers.
Different types of military environments are also likely to attract different types of persons as well as spur different kinds of socialization processes. For example, paramilitary forces or guerilla organizations may not require formal military training before joining but will nevertheless offer less formalized instructions for new recruits. Military service can be very far from any combat experience, but in countries with ongoing conflicts even freshly inducted conscripts may experience armed violence and threats to their life. Also, serving in paramilitaries can be a very dangerous and violent experience, to which the conflict in the Deep South of Thailand and many other conflicts testify.
This study focuses on Thailand, with its system of unusually large voluntary paramilitary and state militia forces operating alongside The Royal Thai Armed Forces (RTARF), which uses conscription by lottery for military service. Military service is typically about two years. National conscription enforced by law has been a significant source of the RTARF's manpower since the early-twentieth century (Chambers 2013;Saowakon 2017). The conscription emphasizes state power as well as the duty of all male citizens of Thailand to the nation and the King and is seen as a way to maintain societal discipline and control (Junko 2002;Winichakul 1997). The military training for conscripts has a reputation of violently enforcing discipline through punishments and social and psychological pressure (Sripokangkul et al. 2019). All men reaching the age of 21 1 are subject to reporting to the annual conscription draft in April.
Moreover, many men-and a few women-in Thailand volunteer to various paramilitary forces. The paramilitaries receive varying amounts of training ranging from days to months. Also the pay varies from none to a decent salary, with the possibility for salary increase with service years and promotions and financial rewards associated with service in unstable border areas. The Deep South-the southernmost provinces of Thailand (Pattani, Yala, Narathiwat)-has been characterized as a conflict zone since 2004 due to an ongoing insurgency. The conflict puts security forces against armed separatists from the Malay Muslims in the region. Bombings and shootings are commonplace. The separatist militants primarily target the various state security agencies (Jitpiromsri 2019). This combination of different military organizations active in a conflict zone enables the study of self-selection into paramilitaries as well as of socialization through military service.
Militarized masculinities are attained through a mix of ascribed and performed identities. The ascribed legal identity of being a man is an important prerequisite for doing military service in many countries in the world, including Thailand. Ideals of masculinity differ depending on context, and Thailand is known for widespread gender fluidity and a relatively high tolerance for and visibility of transpersons, kathoeys. 2 This does not mean that there is full acceptance or integration of gender fluidity in all of society. Thailand ascribes legal gender at birth, and there is no legal option to change gender or go beyond the binary categories of man/woman. Military organizations are also conservative in this respect and are still largely reserved for those who ascribe to a male identity, identify as men, and are legally recognized as male. Kathoeys are usually exempted from military service, but the process required to get exemption on the conscription draft day often involves humiliation and harassment (Dhanakoses 2019). Although masculine ideals in general differ between contexts, militarized masculinities seem to have important commonalities, emphasizing similar hegemonic masculine ideals across a wide variety of contexts. In nearby Singapore, for instance, the National Service has been described as providing "an opportunity to embody the hegemonic masculine ideal by handling weapons, developing physical strength, wearing a uniform and demonstrating loyalty to fellow soldiers and the nation" (Lyons and Ford 2012, 154).
For Study 1, the study of self-selection into paramilitaries, we analyze two types of paramilitary organizations: Community Militias, operating under the Ministry of Interior, and the Rangers, under the Ministry of Defense. Volunteers for service in the paramilitaries have sometimes already done some sort of military conscription service, and these militia 1 Exemptions apply to those who have finalized three years of territorial defense student training, Buddhist monks above a certain rank and those with certain medical conditions. 2 Kathoey is an informal loose term in Thai language that refers to transgender individuals and cross-dressers. Some perceive it as derogatory, especially when used by nontransgender people. We use the word here as it is used in information materials about conscription prepared by the Thai Transgender Alliance for Human Rights (Coconuts Bangkok 2016). volunteers are thus likely to be somewhat older than of conscription age.
The Community Militias form the largest armed organization in the South of Thailand in terms of numbers. We use community militias as an umbrella term for a number of different types of organizations. 3 Operating under the Ministry of Interior, rather than the Ministry of Defense, they can be described as armed civilians and they are supposed to function more like community security guards than soldiers. The government has expanded the uniformed part of the community militias, called Or Sor (the official name is the Territorial Defense Volunteer Corps) to around 7,200 persons. This is part of a large-scale military mobilization and militarization of the entire region. The total number of people serving in different community militias is estimated to be about 95,000. They are deployed in all villages of the three southernmost provinces. Volunteers to the community militias are recruited to serve in their own village, providing protection to government infrastructure and government officials (teachers, village headmen, subdistrict leaders) that are often targeted by the insurgents. While Or Sor are fully employed and paid a monthly salary, other village or community militias are not. Community militias are generally men only, even though there are no formal obstacles for women to join (International Crisis Group 2007).
The Rangers (Thahan Phran), is a different paramilitary group. It is a light infantry organized by the Royal Thai Army and the Royal Thai Marine Corps, operating under the Ministry of Defense. The Ranger force was founded in 1978 to fight Communist insurgents by driving them from their mountain strongholds in northeast Thailand. They were reorganized and downsized during 2000-2001, only to be massively expanded in the South after 2004 as a consequence of the escalation of insurgent violence. There are now over 20,000 rangers. They often operate in conjunction with the Border Patrol Police (BPP), but they are trained to engage in combat, whereas the BPP is primarily a law enforcement agency. The Rangers are generally not held in high regard by civilians because they have been accused of many human rights violations (Ball 2007, 163-78;Peace Survey Network 2020).
To be a ranger you have to be between 18 and 29 years old, male or female, and a citizen of Thailand. Women still constitute a small minority of rangers. Rangers need to reside close to the deployment site but are usually not deployed to their own villages. They have rudimentary training from 45 days up to six months, emphasizing weapon handling of assault rifles and grenade launchers. There are both full-time and part-time rangers with limited remuneration, an average estimate is 13,500 baht (about USD 440) a month. They serve four years at a time and are often deployed to areas with high levels of violence. Whereas regular troops in the Deep South (where about 60% are young conscripts) are more likely to stay on base or patrol in large groups or in vehicles, rangers instead go on small foot patrols in remote areas and on small roads. They patrol checkpoints but also sometimes set up camps in forest areas.
For Study 2, the study of socialization through military service, we study the Royal Thai Armed Forces (RTARF), which uses conscription by lottery. In what follows, we briefly explain the aspects of this quite complex system that are most relevant for our study, whereas we provide a more in-depth explanation in the supplementary materials. Only men are eligible for conscription into military service. Women may join the RTARF in noncombat roles. Today, there are approximately 357,000 active-duty troops.
Under the 1954 Military Service Act all male citizens of Thailand are registered for military duty at age 18. They will remain in this category until they enter the military reserves or are exempted from their military obligations. Approximately a fourth of each cohort will enter the military reserves by going through Territorial Defense Student Training while in secondary school. 4 Those who do not will remain registered for military duty at age 21 and will then have to report to the conscription draft unless they are given deferral due to ongoing higher studies. No more deferrals are possible beyond age 26.
At the conscription draft, senior monks and those with physical or mental disabilities are permanently exempted from military obligations. This is a negligible share of each cohort. The remainder is given the choice to waive the lottery and opt for doing military conscription service. If the number of men waiving the lottery is insufficient to fill the recruitment quota set by the military each year, a lottery will be held for the remaining group. In the lottery those picking a red card must do military conscription service. The rest, who picked black cards, are transferred to the reserves and do not have to do military conscription service. In any given year, there are many more black cards than red cards. If there are enough persons waiving the lottery to fill the recruitment quota at a draft center, all remaining draftees in that district will be transferred to the reserves. Those in the reserves remain eligible to be called up for military duty in case of a national security need until they reach the age of 46, but this has never been done.
Reasons for waiving the lottery vary and depend among other things on one's educational status. For example, those with high school or undergraduate education who waive the lottery get a shorter period of service. Some may be interested in a military career or attracted by the pay. The great majority of men take 3 These include the village defense volunteers (Chor Ror Bor), the village protection force (Or Ror Bor), the civil protection volunteers (Or Phor Por Ror), and the territorial defense volunteer corps (Or Sor). Although there are some differences in training and remuneration, they were deemed to be similar enough to be collapsed into one question in the survey. Numerous casualties testify that serving in the militias in the Deep South is dangerous and members must be willing to carry and use firearms. Data collected by Deep South Watch shows that 221 Rangers and 437 members of the community militias were killed in the conflict during the period 2004-2020. 4 These individuals are not part of our sample. part in the lottery because they try to avoid having to do military conscription service. Many claim to be afraid of having to do military service (Quinley 2019), and this is particularly true in the southern part of Thailand where the armed conflict rages and where conscripts may serve inside a conflict zone. Reports of severe abuse of conscripts occur now and then, which makes the public image of military conscription likely to be negative (Sripokangkul et al. 2019). For additional information about the military system in Thailand, see Figure A9 in the supplementary materials.

DESIGN AND METHOD
A survey was carried out in the South of Thailand, using the conscription lottery as an experimental component. As a first stage, 300 villages were randomly selected and served as clusters. Out of the 300 clusters, 100 were in the conflict-affected Deep South. Information was collected from village headmen about men between 21 and 39 years old who had either participated in the conscription lottery for Military Conscription Service or waived lottery participation and gone through Military Conscription Service. Snowball chains of up to 10 individuals were also used to identify respondents. For each of the 300 clusters, sampling continued until five men had been interviewed from each of three categories: (1) men who participated in the conscription lottery and had to do Military Conscription Service, (2) men who participated in the lottery and were not selected for Military Conscription Service, and (3) men who had waived participation in the lottery and did Military Conscription Service. Thus, the data consist of 4,500 respondentsthat is, 1,500 from each of the three categories (out of which 500 from each category live in the Deep South), see Table 1. This design was intended to boost statistical power to detect associations, but it does not allow descriptive generalizations about the distribution of variables in the population as a whole. The sample is not representative because men who waived the conscription lottery and those who participated in the lottery and had to serve are oversampled relative to the population at large. The respondents were between 21 and 39 years old in 2018, and 86% of them were between 24 and 34. Out of the men who were enlisted through the conscription lottery, 70% served in the army, 28% in the navy, and 2% in the air force.
The empirical investigation of this paper consists of two parts (see Table 1). First we analyze whether men with more patriarchal values are more prone to volunteer for paramilitary service in the conflict zone in the Deep South of Thailand. Study 1 uses two alternative dependent variables. The first is whether the respondent is active in any of the community militias, and the second is whether the respondent is active in the rangers. Study 1 is an observational study that relies on control variables to increase confidence in the interpretation that observed associations may be causal. It investigates the proposition that patriarchal values drive self-selection into militarized violence. It makes use of all three categories of participants but only includes those residing in the Deep South where the armed conflict takes place. Thus, the number of respondents analyzed in Study 1 is 1,500 (500 from each of three categories).
Study 2 deals with the possibility of reverse causality and tests for the effect of military socialization in the form of conscript service on patriarchal values. This part uses a natural experiment in the form of the Armed Violence and Patriarchal Values: A Survey of Young Men in Thailand and Their Military Experiences conscription lottery. The natural experiment analyzed in Study 2 compares the first category, which is the treatment group, with the second category, which is the control group. The third group, consisting of men who chose to waive participation in the lottery and did Military Conscription Service, is not used in the natural experiment (as they self-selected into service). Thus, the number of respondents analyzed in Study 2 is 3,000 (1,500 from each of two categories). Patriarchal Values is thus the independent variable in Study 1 and the dependent variable in Study 2. It is an index based on five questions about the respondent's attitudes toward gender equality that is intended to assess the extent to which the respondent upholds the binary between men and women and assigns a higher value to men. The questions concern to what extent the respondent agrees that men make better political leaders than women, that men make better business executives, that university education is more important for a boy than for a girl, that the husband should make important decisions in the family, and that a woman should tolerate violence in order for the family to hold together. The exact questions and the construction of the index are presented in the supplementary materials (Table A5).
In Study 1 we also use several additional variables as control variables and in robustness tests, among them variables indicating whether the respondent has any children, educational status, income, and marriage status. As Study 1 uses all three categories of respondents, we also use dummy variables indicating to which category the respondent belongs: whether he participated in the conscription lottery and had to serve, participated in the lottery but did not have to serve (the excluded baseline category), or waived lottery participation and did National Military Service.
To test for balance in the natural experiment in Study 2, we use seven questions that refer to conditions that were determined before the time of the conscription lottery: age, Muslim religion, Malay Muslim identity, whether the respondent's father served in the military, the extent to which the situation in the Southern Border Provinces was discussed in the respondent's home when growing up, how often the respondent saw his mother being beaten when growing up, and how often the respondent was beaten at home when growing up. 5 These variables are also used as additional controls or in robustness checks in Study 1 (supplementary materials Tables A7-A10). In the supplementary materials we provide descriptive statistics for all variables (Table A1) and by category of Military Conscription Service status (Tables A2-A4).

Study 1. An Observational Study of Self-Selecting into Paramilitary Service
Study 1 deals with the relationship between patriarchal values and voluntary participation in paramilitary activities in a zone of active violent conflict. This study is observational and seeks to determine whether the variation in our dependent variables in part is explained by individuals self-selecting into the paramilitary organizations. More precisely, in the theory section we derived the expectation that men with more patriarchal values are more likely to voluntarily participate in organized violence despite considerable risks. Figure 1 shows the results from logistic regressions testing this expectation with numerous control variables included to reduce the risk of omitted variable bias. The dots and diamonds represent the coefficients, and the horizontal lines show the 95% confidence intervals. 6 The sample in this part is restricted to respondents living in the Deep South, where the overwhelming majority of the violent incidents happen (Jitpiromsri, Waitoolkiat, and Chambers 2018). The first dependent variable is participation in community militias as was explained above. The second dependent variable is participation in the Royal Thai Rangers. Interestingly, Patriarchal Values has a strong significant effect in the expected direction for both dependent variables. As pointed out above, these analyses are observational and the sample also includes men who waived lottery participation and did Military Conscription Service. Men who waived lottery participation are significantly more likely to participate in the paramilitary activities later in life (Waived). This may be both because of self-selection and because the government strives to entice men with preexisting military training and experience to join. 7 The next control variable (NotBeatenHome) measures experiences of violence in childhood. We find that men who say that they rarely were beaten at home are significantly less likely to participate in paramilitary activities (this control variable ranges from "Very Often" to "Never" and has five steps). We interpret this as self-selection following learning: men who were more frequently beaten in childhood learn that violence is an appropriate way of dealing with conflict, which makes them more likely to join the paramilitaries. Moreover, the experience of being beaten in childhood may well contribute to higher patriarchal values in adulthood (e.g., Bjarnegård, Brounéus, and Melander 2019). We find that Malay Muslims (MalayMuslim) and Buddhists are equally prone to participating in the community defense volunteer groups but that Malay Muslims are significantly less likely to become Rangers. The Rangers are widely despised among Malay Muslims in the Deep South (International Crisis Group 2007; Peace Survey Network 2020), and this may explain why Malay Muslims are more hesitant to join. The next control variable reflects how often the respondent discussed the violent situation in the Southern Border Provinces with his family when growing up (TalkBefore18). For both dependent variables, men who discussed the situation more are more likely to participate in paramilitary activity. We think that this may reflect both self-selection because of interest stimulated by these discussions and higher rates of recruitment in areas within the Deep South that are particularly strongly affected by the conflict (where people are likely to talk more about these things). The next control variable is whether the respondent has one or more children (HaveChildren). Fathers are significantly more likely to join the Rangers but not the community militias. A possible reason is that fathers have greater expenses and thus may be more tempted by the higher pay offered in the Rangers. Militias generally receive no or small salaries, but some select paramilitaries receive full-time salary including allowances and other related benefits. Finally, we find that men with higher Education (Education) are more likely to be Rangers. This may well reflect that the government is particularly keen to recruit those with relatively more education (this may be less of an option for the community militias, which are made up of local villagers).
Although all the controls included in Figure 1 can be said to belong in the analyses in the sense that they are statistically significant in relation to one or both of the dependent variables as shown, it is important to note that Patriarchal Values remains highly significant, with strong effects in the expected direction if these controls are dropped one at a time (supplementary materials Tables A11-12). 8 In the supplementary materials we also show models with fewer independent variables (Tables A13-A16).
There are relatively few observations with missing values on one (or both) of the two dependent variables in Study 1, 5%-7% (supplementary materials Figures A3-A4). The variable of main explanatory interest in Study 1, Patriarchal Values, is missing for 25% of the observations (370 out of 1,500) from the Deep South (supplementary materials Figure A5). This is a relatively large percentage of missing values, so we take a closer look at the missingness in the supplementary materials (Figures A6-A7; Tables A17-A19) and perform alternative tests, including a pair of tests using multiple imputation to replace the missing values. In summary, missingness with respect to Patriarchal Values is not systematically related to our dependent variables in Study 1, and our alternative tests give results that are very similar to those from the baseline models. Thus, we conclude that our robustness checks reported in the supplementary materials suggest that missingness is unlikely to be a major threat to the validity of the results of Study 1.
The result that higher levels of Patriarchal Values are associated with a higher likelihood of participating in the community militias and the Rangers thus seems robust. Now we would like to get a sense of the strength of the effect in substantial terms. Translating the logistic regressions coefficients and comparing more intuitive odds ratios or probabilities only makes sense for a specific set of explanatory variables fixed at the same values and using the same sample (e.g., Norton and Dowd 2018). What is more, the baseline probability of participating in our models is in our case not very meaningful because our sample is nonrepresentative. For example, 28% of the respondents in the Deep South in our sample participated seldom or more often 6 Standard errors are clustered on village. The full regression tables and robustness tests can be found in the supplementary materials (Tables A6-A21). This includes models using two alternative coding schemes for the independent variable in Tables A17 and A20. 7 In Models 8 and 17 in the supplementary materials, a dummy for the category of men who participated in the lottery and had to serve (LotteryServed) is added but fails to attain significance. The third category, i.e., men who participated in the lottery and did not have to serve, is the excluded reference category. 8 In addition to these controls, we added several other controls one at a time. However, none of these additional control variables were significant for either dependent variable. The results for Patriarchal Values remained very similar. These are the additional controls: Sufficient Income, Age (also entered together with Age Squared), Married, Mother Beaten, Lottery Served, Father Served, Protest Before 18, Muslim (supplementary materials Tables A7-10).
in Ranger activities, which probably is a higher fraction than among all men in the same age group in the Deep South (as we have oversampled men who served in the military and therefore are more prone to being Rangers). The absolute probabilities of participating based on our data are consequently less interesting in themselves, and the more meaningful comparison is the relative difference between different individuals. Below we present two mosaic plots that show the relationship between patriarchal values (collapsed into four steps) and participation. These plots are equivalent to bivariate cross tabulations and give an intuitive sense of what our data look like. In the supplementary materials we also show predictive probabilities based on the regression models ( Figures A2-3).
As can be seen in Figure 2, there is a relatively strong difference between those with the lowest and the highest Patriarchal Values in terms of their participation in community militia activities. To the left, among the men with the least patriarchal values, a relatively small share of the men participated in the community militiasnamely, 79 out of 216, corresponding to 37%. To the right, a much larger share of the men with the most Patriarchal Values participated, 181 out of 304, or 60%.
Similarly, in Figure 3, about 24% (52/218) of the men with the least Patriarchal Values participated in the  In summary, Study 1 clearly suggests that men with patriarchal values are more likely to self-select into organized violence through participation in paramilitaries.

Study 2. The Conscription Lottery as a Natural Experiment
Study 2 analyzes the conscription lottery as a natural experiment. Here we investigate the effect of military socialization on patriarchal values among men who hoped to avoid military service by participating in the lottery. In Table 2 we present the balance of several control variables that refer to conditions determined before participation in the conscription lottery, and in the last row we show the treatment effect on Patriarchal Values.
The first row shows that the average age of the treatment and control groups differ when all observations are used. This is natural because the lowest age in our sample is 21. Those between the ages of 21 and 23 who participated in the lottery and drew the red ticket, meaning that they will have to do military service, will most likely be with their units and not available for interview. Therefore, the very youngest respondents are more likely to belong to the category of men who did not have to serve. When the 259 men aged 21-23 are dropped, the treatment and control groups are balanced, as shown in the second row. We control for age in an alternative test in the supplementary materials (Table A22). The next two rows show that Muslims and Buddhists are equally likely to draw the red ticket and that this also holds if only Malay-speaking Muslims are compared with the rest of the sample (i.e., Thai-speaking Muslims and Buddhists together). Similarly, the groups are balanced with respect to the variables Father Served, Talked Before 18, and Mother Beaten. Those who served were significantly less prone to say that they were beaten at home in childhood. The difference is small and may well be a random effect, but we control for Beaten at Home in an alternative test in the supplementary materials (Table A22). The variables chosen for checking balance reported in Table 2 were chosen on the grounds that the variable values were likely determined before the experiment took place.
We now turn to the experiment. The final row shows that there is no significant difference in Patriarchal Values, indicating that being drafted and having to serve has no effect on attitudes toward gender equality. The high number of observations (2,241) means that even a very small effect would have been detected. Given that we can trust that the treatment of drawing the red ticket and serving in the military as a consequence was randomly assigned, we have very strong grounds for concluding that there is no causal effect going from military socialization to patriarchal values. As far as we know, this is the first time that the effect of military service on patriarchal values has been studied using a natural experiment, and the absence of any effect despite the strong design and high number of observations is an important finding.
As a robustness test we also tested for a treatment effect on each of the five components of Patriarchal Values separately but found none (supplementary materials Table A24).
In an alternative test we relaxed the assumption of treatment homogeneity and took into account that the men we surveyed may have had very different experiences of relevance for military socialization. In these alternative tests, we first limited the treatment to only those conscripts who were exposed to conflict-related violence during their service. This means that the assumption that the treatment can be viewed as if randomly assigned becomes possible to question because conscripts who were first randomly enlisted by lottery may then self-select into more exposed roles-for example, by going on patrols or being more alert and active on duty. Also, it is likely that officers will select conscripts into types of roles based on perceived traits such as mental stability. Despite these concerns, we believe that the additional null results reported below further strengthen our conclusions.
We asked the men who served 10 questions about whether they had experienced different forms of violence during their military service-for example, seeing a killed soldier, being close to a bomb, and firing one's weapon at the enemy. Many of the conscripted men had experienced one or more types of violence, ranging from 5% who had been present at their military base when it was attacked to 25% who had seen damage from bombs or other attacks. We created a variable denoting whether the respondent who was enlisted by lottery had experienced any of these forms of violence during their military service and used that indicator as an alternative delimitation of the treatment. Forty-one percent had experienced violence in this sense, but this treatment had no significant effect on Patriarchal Values. We also tested whether the service branch made any difference, first by restricting the enlisted military service treatment to those who served in the army and, second, by delimiting the treatment to those who were enlisted by lottery and served in the navy.
Neither the army only nor the navy only treatment had any significant effect (supplementary materials Table A23). As shown in the supplementary materials ( Figure A8), the number of missing values on the indicator for Patriarchal Values is not significantly different in the treatment group (Enlisted by lottery) and the control group (Not enlisted by lottery). Therefore, we conclude that missingness is unlikely to be a threat to the results of Study 2. In addition, our data contain several other variables of potential interest, such as income and education. These variables were used as controls and in robustness tests in Study 1, but we will not deal with them here because there are sound reasons to expect that serving in the military may affect the variables that are realized after the application of the treatment (e.g., men's future income may be affected by whether or not they served in the military). We will report the effect of the natural experiment on some of these variables in work in progress. Here, we will just mention one finding to illustrate that the experiment does show that there are significant effects of military socialization: men who served are significantly more likely to answer that a man who has served in the military should be more respected than a man who has not served (supplementary materials Table A24). This result and others like it strengthen our confidence that the experiment is valid and does indeed pick up other effects of military socialization. Although the failure to reject the null hypothesis should not automatically lead to the conclusion that there is no effect, this study has a particularly strong design. Consequently, we have great confidence in the result that attitudes to gender equality are largely unaffected by military service in the particular context studied.

CONCLUSION
This article addresses the broader fundamental question of whether patriarchal values drive decisions to partake in violence or if violent experiences, instead, influence patriarchal value systems. By picking out two central components of militarized masculinities-the valorization of violence and its connection to the male gender role-we investigate whether individuals who are already socialized into holding patriarchal values are more likely to choose potentially violent roles for themselves or whether it is instead the case that the misogynistic aspect of holding patriarchal values is strengthened by involuntarily serving in the military in a conflict zone. While recognizing the mutual reinforcement of these processes as well as the importance of different military contexts, we think it is useful to analytically distinguish between self-selection and military socialization as hypotheses for how patriarchal values become part of militarized masculinities.
We make use of a survey of men residing in the conflict-affected southern provinces of Thailand. In the first study, we found a strong and robust relationship between holding patriarchal values and actively volunteering into paramilitary services in active armed conflict. Although this first part of our article is not an experimental study, it lends support to the interpretation that patriarchal values are causes of military masculinities rather than consequences of military socialization, suggesting that self-selection is an important factor for understanding militarized masculinities and participation in violence.
In the second study, we use a natural experiment made possible by the conscription lottery in Thailand to compare survey responses of men who were involuntarily enlisted through the lottery to do Military Conscription Service with the responses of those men who participated in the lottery but were exempted from military service by lottery. We find no significant difference in patriarchal values between the treatment and control groups. In other words, we find no support for military socialization into patriarchal values.
There are important limitations to the study. We cannot rule out military socialization as a potential source of military masculinities, but we do conclude that involuntary service in the armed forces of Thailand had no effect on patriarchal values, despite conscripts serving in a conflict zone. In this study, we thus report that military service does not affect misogyny, but other aspects of military masculinities, such as valorization of hardiness and self-sacrifice, which we did not examine, may be affected. Earlier studies have also pointed to the importance of the severity of experiences of violence, and it is important to remember that military service, even in a conflict zone, is likely to be a less transformative experience than more traumatic and disorganized fighting in the very worst warscapes. We did take into account whether conscripts had themselves experienced violence during their service in our study, but other conflicts may be different in this respect. These issues are important topics for future research.
The workings of the military in Thailand remain important subjects for further study. Our finding that military service has no effect on patriarchal values does not imply an absence of military socialization overall. Rather, ongoing work suggests that military service in Thailand seems to be associated with changes in other values. The null result regarding the effect of military service on patriarchal values also raises questions about the socialization of the young men who never serve in the military. An important topic for future research is the extent to which young men are socialized into patriarchal values in civilian life in Thailand. Furthermore, our study highlights that in the study of gender and war it is fruitful to go beyond gender binaries and essentialism. By studying variations in the degree to which men embrace patriarchal values, we get a much better understanding of what explains the association between men and violence than simplistic comparisons between monolithic blocs of men and women. Moreover, using fine-grained individuallevel data allows us to close in on the causal mechanisms that drive participation in violence. We have focused on men and militarized masculinity because the overwhelming majority of the combatants in the conflict in the Deep South of Thailand, as in the rest of the world, are men. However, the study of women combatants and their values seems like a particularly important topic for further research in light of our results. Do women with more patriarchal values also tend to self-select into military service? Are their values likely to change as a consequence of serving in masculine institutions like the military? Based on their qualitative interviews with both men and women soldiers in the Democratic Republic of the Congo, Baaz and Stern (2008, 69) hold that women recruits are likely to be masculinized given the strong masculine character of the military. Further research on this, using various kinds of data and methods, is called for.
All in all, the finding that patriarchal values seem to be important for volunteering to paramilitary forces such as the Community Militias and the Rangers does give support to the interpretation that patriarchal values are ultimately caused by factors outside military organizations and war in the first hand. However, members of military organizations will often be characterized by higher levels of patriarchal values because individuals who are socialized into holding such values earlier in life are consequently more prone to joining military and paramilitary organizations. In other words, military socialization does not seem to add to baseline levels of patriarchal values, but early life socialization affects patriarchal values and later self-selection (as opposed to forced recruitment) into armed violence. Militaries that rely on volunteers are particularly likely to be staffed by people who embrace traditional notions of militarized masculinities, including its component of patriarchal values. Modern militaries that wish to disassociate themselves from traditional military misogyny and broaden the base of recruitment should pay careful attention to these dynamics of selfselection into the military role. Also, our findings render support to the notion that trying to reduce patriarchal values, especially among young men, is a promising strategy for preventing recruitment to violent groups. Patriarchal values seem to be a primary driver that distinguishes those men who willingly take up arms in a situation of violent conflict from those who do not.

SUPPLEMENTARY MATERIALS
To view supplementary material for this article, please visit http://doi.org/10.1017/S0003055422000594.

DATA AVAILABILITY STATEMENT
Research documentation and/or data that support the findings of this study are openly available at the American Political Science Review Dataverse: https://doi. org/10.7910/DVN/OF1JOF .