Does Police Militarization Increase Repression

Does Police Militarization Increase Repression? Forthcoming at Journal of Conflict Resolution Martin Stavro and Ryan Welch The University of Tampa Abstract: Does police militarization influence government use of repression? While comparative work examines police militarization, violence, human rights abuses, and torture, it does not integrate militarized police within existing theories of repression. Connecting police militarization and repression, this article argues that police militarization increases the police’s perception of threat as well as their coercive capacity, thereby increasing their willingness to repress. As such, we hypothesize police militarization increases repression. To evaluate the theory, we conduct an international analysis using existing datasets covering 102 countries from 1994 to 2010. Employing several statistical models, the data supports the conclusion that police militarization increases the likelihood of government repression, specifically through extrajudicial killing and torture. Aside from highlighting a consequence of police militarization policy, the findings point to police militarization as an important omitted variable in crossnational models of repression and human rights. 1 Does police militarization increase repression? i Until recently, policing institutions have been conspicuously absent from most cross-national work explaining variation in repression and human rights abuses (Eck, Conrad, and Crabtree 2021; Soss and Weaver 2017).ii Many in that literature concerned themselves with constraining Leviathan (e.g., Moore 2010): Hobbes’ terrifying description of the state resulting from the social contract where individuals delegate security to an entity approaching all-powerful to live free from the constant worry of predation by their neighbor. But, of course, this power makes the state a potential (if not likely) source of predation. Institutions have proven useful in pacifying, if not shackling, the ever-present threat. Constitutions, elections, and courts can protect rights, though unsurprisingly their design matters (Chilton and Versteeg 2016; Cingranelli and Filippov 2010; Keith, Tate, and Poe 2009; Powell and Staton 2009; Davenport 2007a). While these institutions can decrease repression, might other institutions increase it? If so, which design dimensions might make this more li kely? To answer that question, we train our focus on the police. Weber’s (2018) canonical definition of the state, inspired by the aforementioned social contract, includes an apparatus to protect itself (and ostensibly those within it) from internal threats using the police. Leaders of the state, rationally desiring to remain in power, possess an entire institutional security apparatus to protect their interests, even if these interests come at the public’s expense. Policing institutions, then, have potentially countervailing mandates: protect the public and protect the political establishment. We argue (uncontroversially) that the outcomes associated with policing institutions, much like other institutions, depend on their institutional design. The organization of police will affect which mandate they prioritize in times of conflict: public safety or government control. More specifically, we focus on the extent to which the state militarizes its policing institutions. Police become more militarized when they receive military hardware, organize 2 themselves militarily, and adopt a militaristic culture—all of which increase the importance of violence for problem-solving (Kraska 2007). Militarized police receive a signal from the state that violence is an acceptable means of public control. This increases the institution’s, and its agents’, capacity and willingness to use violence when challenged. Militarizing the police provokes Leviathan and sharpens its fangs. Following this argument, we hypothesize that militarized police will more readily repress people under their purview. A recent example from the U.S. illustrates the relationship between police militarization and repression. On 1 June 2020 in Seattle, Washington, protesters marched peacefully toward the Seattle Police’s East Precinct to demand an end to racialized policing. Those protesters encountered a barricade initially manned by everyday patrol officers on bicycles. At first, police and protesters interacted without serious altercation, but video evidence highlights a critical event preceding the police’s use of repression (Pozo 2020; Wall Street Journal 2020). Riot officers, outfitted in combat helmets, gas masks, batons, pepper spray, tear gas launchers, and heavy body armor, replaced the bicycle officers. Following a riot officer’s attempt to remove an obstructive umbrella, the Seattle Police, without warning, iii deployed pepper spray, tear gas, and blast balls into the crowd resulting in several injuries to members of the public and the police, including 12 requiring medical attention. The increased militarization of the police on duty directly preceded the repression. We seek to understand why and test the extent to which this episode hints at a more general relationship between police militarization and repression. To do so, we develop a theoretical argument using insights from several political science fields, most notably the conflict literature focused on repression, and the politics of policing, as well as other social sciences (e.g., criminology and sociology). Taking as a starting point that the state desires to decrease perceived threats and that repression represents one of the tools to do so 3 (Poe 2004), we argue that police militarization increases the police’s perceived threat by setting them in a warrior mentality (Koslicki 2021; Lawson 2019; Magaloni and Rodriguez 2020), increasing their willingness to use violence in response to threat (Kraska 2007; Soule and Davenport 2009; della Porta 2011), and bolstering their capacity to do so (Delehanty et al. 2017; Earl, Soule, and McCarthy 2003; Lawson 2019). These mechanisms make violence a more viable option for police, thereby increasing the amount of observable repression. To test our hypothesis, we analyze a 102-country cross-national time-series dataset across the years 1994 to 2010. While we rely on an oft-used measure for state repression (Cingranelli, Richards, and Clay 2014), we pull our measure for police militarization from the relatively new State Security Forces dataset (De Bruin 2021b). Estimating several statistical models while controlling for potential confounders, we find evidence that police militarization increases the likelihood of repression. Leveraging the disaggregated components of the CIRI dataset, we find that increases in explicitly violent tactics of extrajudicial killing and torture drive the results. The paper adds evidence to the importance of institutional design for rights protection. Much of the literature has focused on the importance of democratic institutions’ ability to increase rights respect with a fairly clear result: authoritarian governments more readily repress, and democratic institutions can protect the population from government forces (Davenport 1999; Davenport and Armstrong 2004; Mesquita et al. 2005). But what if the security institutions of a country are designed in such a way that their authoritarian nature is not only tolerated but encouraged?iv Instead of looking at how political and legal institutional design protects human rights (Chilton and Versteeg 2016; Powell and Staton 2009; Welch, DeMeritt, and Conrad 2021; Cingranelli and Filippov 2010), we assess the extent to which the militarized design of security institutions decreases protection. By doing so, we join those concerned with police and 4 repression (Curtice 2021; Curtice and Behlendorf 2021; della Porta 2011; Earl, Soule, and McCarthy 2003; Eck, Conrad, and Crabtree 2021; Soss and Weaver 2017; Soule and Davenport 2009) by focusing on the militarization of the police. Much recent work explores the constabularization of military forces, wherein they perform policing duties for the sake of political order or control (Harig 2020; Blair and Weintraub 2023; Durán-Martínez and Soifer 2021; Passos 2021). While related to our work, we differ by turning our lens to the police and what happens when they become more like the military, rather than incorporating members of the military into the policing institution. v Much of the existing work on police militarization focuses on specific countries or regions—the U.S. (Delehanty et al. 2017; Lawson 2019; Lowande 2021), Iraq (Revkin 2022), Latin America (Flores-Macías and Zarkin 2021; Magaloni and Rodriguez 2020)—but we uncover more general patterns of militarization’s effects on repression across more than 100 countries over nearly twenty years.vi Why Governments Repress Over the last few decades, scholars have made important strides in understanding why governments repress.vii Borrowing from much of that work, we begin by conceptualizing the state as a strategic actor who represses to advance its interests while minimizing costs (Davenport 2007b; DeMeritt 2016). Poe (2004) provides a helpful theoretical account wherein the state seeks to maximize the ratio of perceived strength to threat. When met with threats, the government responds with either accommodation or repression (Poe 2004). The decision between these options depends on their relative costs and benefits (Moore 2000; Poe 2004). Since repressing is often cheaper than accommodating, threats to the government are usually met with repression, an expectation borne out so often empirically that Davenport (2007b, 7) termed it the “Law of Coercive Responsiveness.” 5 Although governments often view repression as cheaper than accommodation (DeMeritt 2016), costs can vary depending on several factors. For example, the capacity of a state to profitably enact repression, as well as its willingness to do so, affect relative costs. The state’s capacity to repress depends on the human capital of a “bureaucracy of individuals willing to commit human rights violations” (Poe 2004, 26), or the possession of violent specialists (Tilly 2003). But even with the capacity to repress, the government must possess the will to do so. We conceptualize will as the government’s comfort with, and propensity to, violence. For example, Gurr (1986) argues that governments relying on repression become more willing to repress over time as the bureaucracy locks in past strategies for future use. The literature reviewed thus far helps us understand why governmental actors repress. To do so, scholars often make a simplifying assumption that the state is unitary. However, once the state decides to pursue repression, it does so by delegating to agents who can carry out the charge (Mitchell 2004). Thus, policymakers turn to agents with coercive capacity built into their charges. The governmental institution best positioned to meet internal threats in the modern state is the police. This is where we turn our attention. Specifically, we will examine how the design of policing institutions affects the principal-agent relationship between state leadership and the police, in turn affecting repression. viii Agents of Repression: Police and Militarized Police The police represent a natural institutional actor to carry out repression in the face of internal threats. They possess coercive capacity and a legitimate charge to employ violence: preserving public safety. However, the extent to which coercive agents—the police included—carry out the state’s repressive mandates influences how successful state repression will be. Information asymmetries and divergent preferences between principals and agents can undermine the state’s 6 position as it seeks to shift strength-threat ratios in its favor (Poe 2004). For example, agency loss can have drastic consequences for the state: rogue agents may commit unsanctioned violence (e.g., Butler, Gluch, and Mitchell 2007) increasing threats to the state through public backlash; on the other hand, agents may fail to repress, undermining the state’s perceived strength (e.g., DeMeritt 2015). Police, specifically, face two contradicting mandates: protect the public but violate its physical integrity upon state orders (Terpstra 2011; Curtice and Behlendorf 2021). These contradictory mandates can exacerbate delegation issues between state leadership and policing institutions. Especially problematic for a state desiring repression would be an agent that either misperceives the importance of repression (thereby devaluing public safety) or one that chooses public safety over repression. State decision-makers that want to ensure repression is carried out can mitigate agency loss issues through institutional design, specifically by militarizing the police.ix Militarized police should be more willing to repress on the state’s behalf for two reasons. Militarization increases 1) the police’s threat perception and 2) their capacity to use violence. Threat Perception Recall that police officers face two potentially competing demands: protecting the public and protecting the state. When the government perceives threats, it is more likely to prioritize selfprotection, increasing the likelihood it issues repressive orders to the police. By militarizing the police, the government can ensure that a policing institution shares its threat assessments, lowering agency loss between them, thereby resulting in more repression. Police militarization occurs when law enforcement agencies tasked with policing functions adopt military hardware, culture, organization, and operations (Kraska 2007). These 7 four dimensions reinforce one another, increasing the policing institution’s propensity to view itself, and those it interacts with, in militaristic terms. Military equipment used in military-style operations reinforces a mindset where officers view themselves at war with their adversaries (civilians). This phenomenon promotes a “warrior-cop” mindset rather than a public-servant mindset (Balko 2014). These mindsets can be imagined as opposite poles on a one-dimensional line: an officer can either think like a warrior or a public servant, with tension between the two ends. Policing institutions can be conceptualized along this spectrum as well. The militarized mindset generated by increased militarization increases the police’s propensity to view itself as beset on all sides with enemies; as such, it becomes an organization ready to wage war against these enemies, as opposed to a public-serving institution (Stoughton 2015; Lawson 2019). For example, in interviews conducted by Simckes et al. (2019, 525), militarized police characterized themselves as being in “constant attack mode.” Inflated threat perception, introduced by the process of police militarization, makes officers more willing to repress by aligning the goals of guarding public safety, enacting state repression, and self-preservation (Flores-Macías and Zarkin 2021). Actual threats being equal, militarized police will perceive inflated threats in comparison to their less militarized counterparts (Kraska 2007). This gap between actual and perceived threats increases the likelihood that the state will repress as the police turn to violence more often to solve their problems. x Two examples illustrate the threat perception-repression pathway. On the 13th of June 2020, protesters assembled in Paris, France, to demonstrate against police brutality in the wake of George Floyd’s death. For three hours, the protesters were peaceful and confined to the Place de la République. The protest, however, was unauthorized, and riot police were dispatched to break up the protesters. The riot police (a form of militarized police), perceiving the protesters as 8 threats needing to be dealt with (rather than peaceful demonstrators), charged protesters and fired tear gas into the crowds. What was an otherwise peaceful protest quickly devolved into skirmishes with the police (FRANCE 24 English 2020; Reuters 2020), highlighting that militarized police’s inflated perception of threat can act as a type of stereotype bias through the dissent-repression nexus. Militarized police see peaceful protesters as threats and, through escalated policing (della Porta 2011), push protesters into becoming threats. Like the example in France, on the 1 st of June 2020, protesters assembled on a highway in Philadelphia to demonstrate against police brutality. Among the first responders to the scene was a state trooper (a less militarized form of police). Radio transmissions from the trooper indicate that they perceived the protest to be peaceful. SWAT teams quickly responded to the protests by trapping protesters on the highway. The SWAT officers perceived the protesters far differently than the state trooper, quickly responding with force despite the trooper’s comments. In one extreme moment captured on video, a SWAT officer engages a protester sitting down on the highway by pulling down their mask and pepper-spraying them at point-blank range. Like in France, militarized police saw peaceful protesters as enemy combatants, not as civilians under their protection (The New York Times 2020). Capacity Police militarization aligns policing with the material, cultural, operational, and organizational facets of the military (Kraska 2007). The state’s offering of military structure and devices to its policing institutions encourages repression by reducing the cost of capacity-building. Similar to small arms imports which negatively impact physical integrity rights by way of making repression cheaper (Brender and Pfaff 2018), militarized police are more willing to repress as the state supplies the instruments to do so. The process of militarization provides a signal to the 9 police on how to solve problems in line with Maslow’s (1969) law of the instrument. When the principal (the state’s decision-makers, policymakers, leaders, etc.) gives their agent (the police) a hammer, the agent presumes the desired solution involves hammering. To the agent, every problem begins to look like a nail. In the case of police militarization, when the state offers the police mine-resistant ambush-protected vehicles or grenade launchers, the police interpret the use of these instruments as acceptable and preferred by the state. After all, if the state did not intend these instruments, designed for violent coercion, to be used by the police to further its goals in a repressive manner, then why would the state provide these tools to the police? Provided that the police are specialists in violence and bureaucrats of coercive force (Curtice and Behlendorf 2021; Poe 2004; Tilly 2003), they presume the provision of military equipment greenlights repression. Looking to please the state while reducing the costs of decision-making by using capacity-building as a heuristic, militarized police will rely on violence more readily. This frequent deployment of violence will end in repression more often. Repression will become increasingly entrenched in a type of bureaucratic inertia (Gurr 1986) as the police see exploring new avenues of problem-solving as costly: the implements militarized police possess favor repression, and the state’s provision of these implements suggests approval. Providing capacitybuilding to militarized police and leaving them to their own devices, when exacerbated by information asymmetries (Cook and Fortunato 2022; Kirisci 2022), emboldens the willingness of militarized police to repress as the assumption of the state’s tacit approval of repression goes unchecked. The provision of militarized policing equipment to Venezuela by Chinese firms provides an illustrative case example. The New York Times (2017) reported that the Venezuelan government acquired high-tech police vehicles sold by Chinese firms to modernize their policing 10 strategy. These vehicles were heavily armored and integrated powerful water cannons mixed with liquid tear gas to quell protests, presenting an effort not only to modernize policing but militarize it. Acquiring these vehicles and providing them to policing institutions enables the Venezuelan government to increase the likelihood of repression. Police will see the provision of these vehicles, designed to dish out violence, as reflecting Venezuela’s stance toward dealing with demonstrators. Moreover, the power of the tools provided to the police, combined with the protection this increase in capacity provides (sitting behind an armored vehicle as opposed to manning a line against protesters), makes violence cheap. Addressing the Professionalism Counterargument Some claim that militarization is a byproduct of a larger professionalization movement in policing. They argue that as policing institutions modernize, they acquire more advanced equipment, including materials and weapons. We should expect those officers to act with higher levels of restraint because the acquisition of advanced weaponry and vehicles requires higher levels of training (see den Heyer 2014 for a representative piece). We see at least two problems with this argument based on assumptions not borne out empirically. First, although receiving more lethal hardware should require more training, it is not clear training increases occur. This is an empirical question that requires more research, but the differences in military and policing institutions lead one to doubt police receive enough training for militarized hardware. When the military is not engaged in active combat, much of their time is spent in training to remain combat-ready. However, police do not possess such “down time” that can be translated into training. They are on active patrol most of the time. The response of the Toronto Police to the 2010 G20 protests provides a case example illustrating the disparity between police militarization and appropriate training. Toronto Police deployed a long-range 11 acoustical device (LRAD) during the G20 protests for crowd control. These acoustic devices can induce nausea, abdominal pain, vomiting, loss of hearing, and even vibrate the bones of target s. LRADs are a military technology originally developed “by the US military for use against Somali pirates and Iraqi insurgents” that found their way into the armories of Canadian police departments (Kitchen and Rygiel 2014, 209). Whereas U.S. military personnel spend most of their workdays training to use devices like LRADs (amounting to hundreds, if not thousands, of workhours), the mandated training received by Toronto Police consisted of one topic out of six in a module taking no longer than 2.5 hours. Although the exact figure of the time spent on the LRAD is not public information, the first topic in the training module where LRAD was bundled took up one-third of the 2.5-hour allotment, leaving the remaining two-thirds for LRAD training alongside four other modules (Morden 2012, 254–56). Secondly, the professionalization argument assumes proper training. In the U.S., police training “tends to be short [see above], classroom-based, and rarely emphasize[s] de-escalation” (Tecott and Plana 2016). Firearm training occurs over seven times more frequently than deescalation (Chang 2016; Tecott and Plana 2016), even though training based on procedural justice reduces violent interactions between officers and the public (Owens et al. 2018). Amnesty International cites a similar absence of procedural training and regulation as a contributing factor to police abuses in Brazil, such as the misuse of military-grade equipment in urban areas (Neder 2017, 5). Moreover, evidence suggests that as departments acquire more militarized hardware, they find ways to use them in the community to train. This strategy leads to heavy-handed policing for non-violent offenses that inevitably results in more violence (Balko 2014). For example, an interviewee in one of Kraska’s studies with police officers voiced how individual officers see militarization creeping into everyday policing: “Why serve an arrest warrant to some 12 crack dealer with a .38? With full armor, the right shit, and training, you can kick ass and have fun” (Kraska 2001, 143). This attitude is documented beyond the U.S. in counter-narcotic operations by Brazilian police where heavy weapons, helicopters, and armored vehicles are disproportionately employed in favelas and marginalized communities (Neder 2017, 5). These ideas of improper training and use of force by militarized police contra their military counterparts are best illustrated by the 2022-2023 anti-government protests in Peru. Beginning in December 2022, protesters took to the streets in Peru after President Pedro Castillo was ousted from office following a foiled coup attempt. National police and soldiers were deployed to quell protests by Castillo’s supporters. In images analyzed by McDonald, Tiefenthäler, and Surdam (2023), police officers are seen outfitted in heavy gear: ballistic armor, bulletproof helmets, riot shields, 12-gauge shotguns, and Kalashnikov assault rifles. One particularly damaging event for the professionalization thesis was the militarized police response at Juliaca on 9 January 2023. Per the New York Times: “Images, documents, and testimonies collected by the Times provide a detailed account of the military and police response to protesters at the local airport [in Juliaca] that day, and suggest police officers on the front lines were responsible for most of the gunshot injuries and deaths. Footage and official documents also point to police officers’ involvement in several more civilian shooting deaths in the city center that evening” (McDonald, Tiefenthäler, and Surdam 2023, para. 31; emphasis added). According to the professionalization hypothesis, we should see militarized police act similarly to the military; instead, this example supports the hypothesis that militarized police are the source of increased repression. Peru’s militarized police possessed a common threat and comparable capacity to the soldiers on-scene, and yet these police officers, according to journalistic 13 accounting, engaged in the bulk of repression (i.e., extrajudicial killing). This case suggests that police militarization provides the ingredients of repression (military force and mindset) without instilling the benefits of professional discipline. While police militarization should require 1) more specialization and 2) proper training methods, the available evidence suggests that does not occur. And although training varies between countries (and within them) on average, we believe the assumptions underlying the professionalization argument are empirically suspect. xi How Police Militarization Increases Repression By conceptualizing the state as a strategic actor and bearing in mind that states undertake police militarization as an institutional design choice to influence the choices of its agents, police militarization influences an overall increase in repression. Recall that police militarization is the material, cultural, organizational, and operational emulation of the military by the police (Kraska 2007), and that repression consists of physical integrity violations to impose costs dissuading actors from taking certain actions, i.e., opposing the state (Davenport 2007b). Police militarization inflates threat perception through the lens of militarism. However, provided that strength-threat imbalances have remedies beyond repression (Poe 2004), why does police militarization lead to repression? It decreases the relative costs of repression. The state’s decision to militarize the police bolsters the capacity of a state’s primary domestic coercive apparatus. When militarized, violence and repression become cheaper for policing institutions. Moreover, before police militarization, the duty of policing is complicated by a balance between serving the public and serving the state. However, capacity-building through police militarization signals that the state’s preferred repertoire of solutions consists of coercive options. Becoming reliant on military tactics as a function of minimizing decision-making costs and bureaucratic inertia, 14 militarized police begin using coercive force more broadly and inappropriately, constituting repression. Thus, the inflated threat perception and capacity-building brought about by police militarization increases the willingness of police to repress, devoid of protections that might be afforded through military-style professionalization. With the motives and tools in place, police militarization increases repression. Figure 1 provides a visual representation of our theory. H1: An increase in police militarization leads to an increase in government use of repression. 15 Figure 1. Visual Representation of Theory 16 Assessing Police Militarization’s Influence on Repression To test the hypothesis, we focus on the country-year as the unit of analysis, estimating ordered logit, partial proportional odds, and linear regression models. We draw data from a variety of sources whose spatial-temporal domain consists of 102 countries from 1994 to 2010 (n = 1650).xii Dependent Variable: Government Use of Repression To operationalize the dependent variable, government use of repression, we use a recoded version of the CIRI Physical Integrity Rights Index (Cingranelli, Richards, and Clay 2021). This index aggregates four components into a comprehensive measure of a government’s respect for physical integrity: torture, extrajudicial killing, political imprisonment, and disappearance. Originally, each component is scored from 0 to 2, then aggregated into a single index ranging from 0 to 8, with 0 representing complete disregard for physical rights, or maximum state-based repression, and 8 representing complete respect for physical rights, or no state-based repression (Cingranelli and Richards 2010, 413). For easier interpretation, we recode the CIRI Physical Integrity Rights Index by inverting the scale, having higher values represent more repression. On this recoded variable, 8 represents maximum repression, and 0 represents minimum repression. xiii Interested readers can find descriptive statistics for this and all variables in the online appendix.xiv Explanatory Variable: Police Militarization We operationalize police militarization using De Bruin's (2021b) State Security Forces dataset, which records police militarization on a dichotomous scale: 0 for the absence of national militarized police and 1 if present. A country’s police force is militarized if it possesses at least 17 two of the following: command and control centers; elite squadrons modeled after the military; special operations; barracked housing; or long-range deployment (De Bruin 2021a, 2–3). See De Bruin (2021a) for detailed coding and methodological documentation. Most of the country-years in our sample (64%) contain militarized police (figure A1; table A2). As figure 2 shows, militarization levels vary from year to year, with a peak in the early 1980s that began decreasing until the mid-2000s, when it began increasing again. xv Figure 2. Global Police Militarization over Time, 1965-2010 Control Variables To guard against spurious findings, we include control variables related to the dependent and independent variables. First, we include a count of the number of protests in a country-year. Doing so accounts for the law of coercive responsiveness (Davenport 2007b), wherein states respond to dissent with 18 repression. Furthermore, the number of protests may affect the extent to which a country militarizes its policing institutions in an attempt to prepare itself for confrontation (della Porta 2011). We measure the number of protests using ICEWS event data (Boschee et al. 2015). Next, we include a country’s level of democracy in a given year. A robust literature finds democracies repress less often than their less-democratic counterparts (e.g., Poe and Tate 1994; Davenport 2007a; Hill and Jones 2014). Regime-type may also affect if and how a state militarizes its policing institutions, with governments accountable to the public facing more pushback in doing so. We include the Polity IV measure to control for democracy (Marshall, Gurr, and Jaggers 2015). We account for another institutional constraint: judicial independence. Several scholars have shown that judicial independence decreases repression (Keith 2002; Powell and Staton 2009; Simmons 2009). An independent judiciary also presents an additional veto player against police militarization. To control for judicial independence, we include Linzer and Staton’s (2015) latent variable measure that estimates judicial independence using informat ion from several sources. National human rights institutions (NHRIs), domestic institutions tasked with promoting and protecting human rights, also provide institutional protection from repression (Cole and Ramirez 2013; Welch 2017; Welch, DeMeritt, and Conrad 2021). Further, while enacting their rights protection mandate, many NHRIs monitor and comment upon police activity within the state, which may influence the extent to which police (do not) militarize. We include a dichotomous indicator for NHRI presence in a given country-year taken from the National Human Rights Institutions Data Collection Project (Welch, DeMeritt, and Conrad 2021). 19 We include standard controls used in other repression models, logged GDP and population (World Bank 2015), as well as a lagged dependent variable. Statistically Modeling Police Militarization and Repression To test the hypothesis and hedge against concerns over modeling decisions, we estimate and present results from ordered logit, partial proportional odds, and linear regression models. xvi We present first differences using the ordered logit model to demonstrate the influence of police militarization within each level of repression. Across all models, the findings remain unchanged: police militarization increases repression below the alpha level of 0.05. Results Before presenting results from the statistical models, we observe the bivariate relationship between repression and police militarization. The descriptive visualization provided in figure 3, a violin plot of country-year distributions along the repression scorexvii subset by police militarization, tells a compelling story: police militarization shifts observations toward more repressive rankings. The median repression score for country-years without police militarization is 2, while the median for country-years with police militarization is 4. 20 Figure 3. Violin Plot of Repression Index by Police Militarization Note: Dots represent the median for categories. 21 Do these results hold in controlled statistical models? Table 1 displays the results of three models: ordered logit (model 1), partial proportional odds logit (model 2), and linear regression (model 3). We drop the coefficient for judicial independence and the constant from the partial proportional odds model since both fail the parallel regression assumption.xviii All models include controls. Interested readers can find uncontrolled models for OLS and ordered logit in table A4 of the online appendix. Recall that the dependent variable is the inverted CIRI index scored from 0 to 8 with higher values indicating greater repression. The results from table 1 are in line with our hypothesis and the visualization in figure 3. The coefficients of models 1 and 2 represent the logodds change from no police militarization to police militarization. Model 3’s coefficient represents the average repression index difference between a country-year with and without police militarization. When police militarization occurs, the likelihood of a country-year scoring higher on repression increases. The linear regression in Model 3 confirms the fit of the model, explaining 78% of the variation when accounting for controls. Without controls, the model has a poor fit, explaining only 7.8% of the variation in OLS regression and outside the 0.2-0.4 range in the ordered logit Pseudo-R2 (see table A4 in the online appendix). 22 Table 1. Analysis of Repression Outcomes Dependent variable: Repression Index (Inverted CIRI Physical Integrity Index) ordered partial proportional OLS logistic odds (1) (2) (3) Police militarization 0.349*** (0.101) 0.308*** (0.103) 0.161*** (0.058) Number of protests 0.006*** (0.001) 0.005*** (0.001) 0.002*** (0.0004) Polity IV score -0.023* (0.013) -0.032** (0.014) -0.019** (0.008) Latent judicial independence -1.593*** (0.373) -0.685*** (0.207) NHRI presence -0.210** (0.096) -0.206** (0.097) -0.145*** (0.056) Log GDP -0.498*** (0.104) -0.478*** (0.105) -0.239*** (0.059) Log population 1.139*** (0.146) 1.145*** (0.147) 0.656*** (0.082) Repression t-1 1.153*** (0.041) 1.138*** (0.041) 0.686*** (0.017) -0.673* (0.375) Constant Observations R2 Adjusted R2 McFadden’s Pseudo-R2 Log Likelihood 1,650 1,650 0.351 0.646 -2,280.061 1.065 (df = 1641) 754.042*** (df = 8; 1641) Residual Std. Error F Statistic Note: 1,650 0.786 0.785 * p<0.1; **p<0.05; ***p<0.01 23 Yet, if police militarization alone can explain 7.8% of physical integrity abuses, this may represent a key step in improving human rights and an important omitted variable in crossnational models of repression. Converting the log-odds coefficients for police militarization into odds ratios illustrates the likelihood that a country-year with militarized police falls into higher categories of repression. Table 2 provides the odds ratios corresponding to models 1 and 2 from table 1. 24 Table 2. Odds Ratios of Ordered Logit Analysis Dependent variable: Repression Index (Inverted CIRI) ordered partial proportional logistic odds (1) (2) Police militarization 1.417 1.361 Number of protests 1.006 1.005 Polity IV score 0.977 0.968 Latent judicial independence 0.203 NHRI presence 0.811 0.814 Log GDP 0.608 0.620 Log population 3.125 3.143 Repression t-1 3.168 3.119 Observations Log Likelihood 1,650 1,650 -2,280.061 Note: standard errors, p-values, and OLS model omitted for odds ratios (see table 1 instead). 25 The odds ratios suggest country-years with police militarization are roughly 36 to 42% more likely to fall into a higher category of repression than those without. However, odds ratios do not provide specific information on the predicted probabilities of falling into a particular repression level, only the probability of moving upward in levels. Table 3 reports the first differences for police militarization based on model 1, holding all other variables constant at their medians. Figure A5 in the online appendix visualizes the first differences and confidence intervals for all variables. 26 Table 3. First Differences in Predicted Probabilities for Repression Level Due to Police Militarization Variable Police militarization Repression Level First Difference Lower CI Upper CI 0 -0.004* -0.022 -0.000 1 -0.024* -0.077 -0.002 2 -0.027 -0.064 0.031 3 -0.007 -0.067 0.062 4 0.026 -0.047 0.069 5 0.021* 0.001 0.054 6 0.011* 0.000 0.048 7 0.004* 0.000 0.020 8 0.001* 0.000 0.006 Note: * p < 0.05; lower and upper values computed at 95% confidence interval. 27 In six out of nine levels, police militarization has a statistically significant effect. Police militarization decreases the predicted probability that a country-year will fall into the lowest repression levels, 0 and 1, and increases the predicted probability of a country falling into the four highest levels of repression, 5 through 8. Though the effect sizes are small, ranging from a predicted 0.1% increase in probability for level 8 to a 2.4% decrease in predicted probability for level 1, our data allow us to state confidently that the changes are significant from 0. The control variables behave as expected. Table 1 highlights that increased dissent (proxied by protest count) increases repression (Davenport 2007b; DeMeritt 2016), increased population increases repression (Poe and Tate 1994), and that past repression has an inertial influence on future repression (Gurr 1986; Poe and Tate 1994). As anticipated, increased democracy, latent judicial independence, GDP (Hill and Jones 2014; Powell and Staton 2009; Simmons 2009; Poe and Tate 1994), and the presence of NHRIs (Cole and Ramirez 2013; Welch, DeMeritt, and Conrad 2021) decrease repression. To ensure robustness, we substitute the dependent variable, CIRI, with the Political Terror Scale (PTS; Wood and Gibney 2010, 369). As a further robustness check, we specify a two-way fixed effects model. The results remain similar (Table A6, Table A15). Police Militarization Increases the Probability of Torture and Extrajudicial Killing A benefit of the CIRI Physical Integrity Index is its disaggregation of repression into four metrics: torture, extrajudicial killing, disappearances, and political imprisonment. To determine police militarization’s influence on each tactic, we estimate ordered logits by component. xix Table 4 displays the results. Table A7 in the online appendix displays the partial proportional 28 odds models for components violating the proportional odds assumption. Modeling follows the methodology outlined above with the lagged prevalence of a tactic replacing lagged repression.xx The results of table 4 demonstrate police militarization has a positive and significant effect on the use of extrajudicial killing and torture by states. Converting the significant effects into odds ratios yields 1.962 for extrajudicial killing and 1.548 for torture. In terms of percentages, a country-year with police militarization is 96.2% more likely to experience a higher incidence of extrajudicial killing and 54.8% more likely to experience a higher incidence of torture compared to a country-year without. The torture ordered logit passes the proportional odds assumption.xxi Results from the torture model support Magaloni and Rodriguez’s (2020) findings that police militarization increases torture while extending their claims beyond Mexico. Provided the magnitude and direction of the latent judicial independence coefficient compared to police militarization, Magaloni and Rodriguez’s (2020) emphasis on judicial checks as a curb to torture holds as well. 29 Table 4. Analysis of CIRI Tactic Outcomes Dependent variable: (1) Political Imprisonment (2) Extrajudicial Killing (3) Police militarization -0.069 (0.160) 0.091 (0.130) 0.674*** (0.126) 0.437*** (0.130) Number of protests 0.001 (0.001) 0.007*** (0.002) 0.007*** (0.002) 0.010*** (0.002) Polity IV score -0.004 (0.020) -0.128*** (0.019) 0.041** (0.016) -0.004 (0.018) -1.641*** (0.591) -0.498 (0.484) -1.997*** (0.453) -1.870*** (0.478) NHRI presence 0.071 (0.150) -0.259** (0.128) -0.345*** (0.121) -0.241* (0.128) Log GDP -0.386** (0.156) -0.355*** (0.134) -0.545*** (0.128) -0.439*** (0.139) Log population 1.335*** (0.216) 1.216*** (0.187) 1.231*** (0.178) 1.070*** (0.189) Disappearance t-1 2.459*** (0.114) Disappearance Latent judicial independence 2.215*** (0.101) Extrajudicial killing t-1 2.041*** (0.114) Torture t-1 Note: (4) 2.106*** (0.103) Political imprisonment t-1 Observations McFadden’s Pseudo-R2 Torture 1,654 0.424 1,654 0.462 1,658 0.411 * 1,658 0.365 p<0.1; **p<0.05; ***p<0.01 30 Moreover, police militarization’s relationship with extrajudicial killings suggests a general increase rather than one unique to the United States (Delehanty et al. 2017; Lawson 2019; Masera 2021). In terms of the repression literature, militarized police serve strategic functions for states. For one, it may institutionalize certain forms of repression (torture, extrajudicial killing) behind the veil of professionalism (Kappeler and Kraska 2015; Ron 1997). Further, police militarization could be one method by which governments substitute repressive tactics (DeMeritt and Conrad 2019; Moore 2000). For example, governments dissatisfied with the efficacy of political imprisonments conducted by conventional police may opt for police militarization to shift repression towards torture and extrajudicial killings. Aside from implications to the existing literature, the results in table 4 offer a finer view of the theory described above. The influence of police militarization on extrajudicial killing is in line with the state’s capacity-building efforts; by militarizing the police, the state offers more instruments of violence, thereby increasing the likelihood of extrajudicial killing occurring. As far as police perceive these instruments as reflecting what the state wants, police will use the material component of militarization for violent ends such as extrajudicial killing. However, police militarization’s impact on torture is a critical reflection of perceived threat and will. Previous work suggests that when the public perceives a heightened threat, they are more approving of torture (Anderson and Richards 2018; Conrad et al. 2018). The same may apply to militarized police officers: convinced by politicians and a militaristic culture that they are facing a heightened threat (Lawson 2019; Magaloni and Rodriguez 2020), militarized police, like the public, become more willing to torture. Additionally, the inflated threat perceived by militarized police may lead them to believe the public will be more accepting of torture, thereby posing less risk of backlash. Combined with assumed state approval, police militarization results in 31 increased torture. Though these are mostly post-hoc interpretations, the results corroborate the sentiment of our proposed theory: militarized police’s increased capacity and will towards violence lead them to perpetrate the most explicitly violent types of state repression. Disentangling Repression and Police Militarization One concern might be that repression and police militarization are endogenous; repression (or anticipation thereof) prompts states to militarize their police, police militarization encourage s more repression, and the relationship between the two is a vicious circle. In this case, police militarization would not increase repression. To account for this concern, we estimate several logistic regression models to assess whether repression (in the current country-year as well as lagged a year to account for anticipated repression) influences police militarization. The State Security Forces dataset includes a dichotomous variable recording newly militarized police forces in a country-year (De Bruin 2021a; 2021b). We would expect that if states anticipating repression are prone to militarize their police forces, either current or past repression would positively correlate with transitioning the current police to a militarized state. xxii We present the results in table 5. We include various models with and without controls for robustness. Repression does not exhibit a statistically significant impact on police militarization, regardless of the model specification, giving us more confidence that our findings are uncovering a relationship in which police militarization leads to repression, and not the other way around. 32 Table 5. Repression’s Influence on New Police Militarization Dependent variable: (1) (2) Newly Militarized Police (3) (4) (5) (6) Repression 0.329 0.194 0.251 (0.392) (0.267) (0.238) Repression t-1 -0.177 (0.378) Number of protests 0.001 0.0003 (0.006) (0.006) 0.001 (0.006) Polity IV score -0.078 -0.072 (0.115) (0.115) -0.088 (0.114) Latent judicial independence 3.916 3.819 (3.273) (3.205) 3.761 (3.298) NHRI presence 0.780 0.789 (0.918) (0.919) 0.719 (0.912) Log GDP -1.663* -1.573 (0.993) (0.974) -1.737* (0.987) Log population 0.629 0.526 (1.345) (1.338) 0.868 (1.316) Constant 4.685 4.358 -6.261*** -6.321*** -6.082*** (6.183) (6.037) (0.654) (0.645) (0.611) 4.315 (6.124) Observations Log Likelihood Akaike Inf. Crit. 1,650 1,659 2,839 2,965 2,859 -36.649 -36.886 -76.187 -76.769 -76.847 91.297 89.772 158.373 157.537 157.693 1,656 -37.054 90.108 Note: 0.195 (0.121) -0.062 (0.238) 0.151 (0.120) * 0.052 (0.264) p<0.1; **p<0.05; ***p<0.01 33 Conclusion Much of the scholarship on institutions and repression concerns constraining Leviathan (Moore and Welch 2015). However, by thinking through the importance of institutional design, we have highlighted a way in which Leviathan’s chains are loosened. Specifically, we turned our attention to the institution most directly involved with repression—the police (della Porta 2011). While the police’s raison d’être relies on coercion, the extent to which they are militarized affects state repression. Building on the existing repression literature and drawing lessons from more recent work on police militarization, we presented a theory for how police militarization increases repression. The state’s decision to militarize the police, a strategic choice meant to empower the domestic security apparatus against (perceived/potential) threats, ironically increases the amount of threat perceived by the police, while also increasing their capacity to respond to those threats violently. Our statistical analyses support our theoretical expectations: increased police militarization increases repression. Our findings present opportunities to constrain Leviathan. One way to decrease repression is to rein in police militarization. That insight carries valuable information for the public and policymakers. Domestically, (democratic) publics can demand their lawmakers address militarization through oversight and legislation. In the U.S., where much of the police militarization debate has been highlighted in the wake of racial justice protests, xxiii Congress has held oversight committees that exposed flaws in the current system. And while national legislation has faltered, states (e.g., California) have passed laws meant to constrain the bureaucratic apparatus making militarization decisions, which research suggests should affect policy outcomes (Huber and Shipan 2002). These types of actions can protect the public from the state. Further, to the extent that members of the public wish to enhance human rights globally, 34 they should insist their lawmakers scrutinize foreign aid practices. For example, the U.S. Department of Defense provides police militarization programs for developing countries to maintain order in that country (Buchanan 2011; Kenzer 2020). However, our analysis highlights that order comes at the cost of reduced respect for human rights (see also Aaronson 2022). Officials in foreign policy positions should take seriously the competing objectives of bolstering human rights and increasing stability in service of national security and fiscal interests (e.g., Revkin 2022). For scholars, our findings suggest an important omitted variable for existing models of repression in cross-national research designs. Until now, many of us have failed to account for the institution most directly involved with repression: the police (Eck, Conrad, and Crabtree 2021). But recent scholarship in comparative politics and other fields (Delehanty et al. 2017; Magaloni and Rodriguez 2020; Revkin 2022) suggests we pay more attention to policing in cross-national models of human rights. How policing institutions are designed matters (Greitens 2016; Arriola et al. 2021) and we find militarization increases repression. Other design decisions probably matter too. For example, the amount and type of training varies greatly between policing institutions. Future research should explore this and other institutional design variations , though recent work warns us to be wary of the difficulties in reforming the police that stem from politics and the police themselves (González 2019; 2020; 2023; Flom 2022). Further, future work could better capture the institutional variation of police militarization. For example, Kraska (2007) and Simckes et al. (2019) offer several subdimensions of police militarization. Future work could consider whether material militarization influences repression more than organizational, operational, cultural aspects, or similar questions. Do those dimensions act independently, or reinforce each other? Using the current dichotomous measure from the State 35 Security Forces database (De Bruin 2021a; 2021b), while valuable in determining the general relationship between police militarization and repression, does not allow us to answer those questions.xxiv Another limitation of using the current police militarization measure is its focus on national police forces. Some countries use a decentralized model of policing where local jurisdictions possess powerful policy-making authority (the U.S. being an extreme example). In those cases, some jurisdictions will be more militarized than others, while the national policing institutions also differ. The State Security Forces dataset codes the U.S. police as non-militarized using its coding rules. We account for this using robustness tests that drop the U.S. and recode it as militarized.xxv Furthermore, we know from past work (e.g., Delehanty et al. 2017; Lawson 2019) that militarization varies at the subnational level. We estimated a robustness check that did not include federal systems in the sample, and the general results remain unchanged—increased militarization increases repression. xxvi Finally, we developed a theory of how police militarization increases repression through two theoretical mechanisms: increased threat perception and increased capacity for violence. We contribute to the literature by showing that the implications of our theoretical expectations occur in the data: police militarization increases repression. Future research, such as qualitative case studies, experiments, or mediation analyses, should further probe the proposed mechanisms. 36 Acknowledgements This work was supported by the University of Tampa’s Office of Undergraduate Research and Inquiry’s Summer Undergraduate Research Fellowship. We also thank three anonymous reviewers and the editorial team at JCR for their comments 37 References Aaronson, Ely. 2022. “From Global Interventionism to Domestic Police Militarization: The Transnational Routes of American Policing.” Law & Social Inquiry 47 (1): 369–74. https://doi.org/10.1017/lsi.2021.84. Anderson, Mary R., and David L. Richards. 2018. “Beyond the Media’s Explanation: Examining the Determinants of Attitudes toward Torture.” Journal of Human Rights 17 (3): 289– 302. https://doi.org/10.1080/14754835.2018.1437541. Arriola, Leonardo R., David A. Dow, Aila M. Matanock, and Michaela Mattes. 2021. “Policing Institutions and Post-Conflict Peace.” Journal of Conflict Resolution 65 (10): 1738–63. https://doi.org/10.1177/00220027211013088. Balko, Radley. 2014. Rise of the Warrior Cop. Illustrated edition. New York: PublicAffairs. Blair, Robert, and Michael Weintraub. 2023. “Little Evidence That Military Policing Reduces Crime or Improves Human Security.” SSRN Scholarly Paper. Rochester, NY. https://doi.org/10.2139/ssrn.3925245. Boschee, Elizabeth, Jennifer Lautenschlager, Sean O’Brien, Steve Shellman, James Starz, and Michael Ward. 2015. “ICEWS Coded Event Data.” Edited by Lockheed Martin et. al. Harvard Dataverse. https://doi.org/10.7910/DVN/28075. Brender, Agnes, and Katharina Pfaff. 2018. “Still Tools of Repression? Re-Assessing the Effect of Arms Imports on Physical Integrity Rights.” 22. ILE Working Paper Series. ILE Working Paper Series. University of Hamburg, Institute of Law and Economics. https://ideas.repec.org/p/zbw/ilewps/22.html. 38 Buchanan, John. 2011. “A Field Guide for USAID Democracy and Governance Officers: Assistance To Civilian Law Enforcement in Developing Countries.” Washington, D.C.: United States Agency for International Development. Butler, Christopher K., Tali Gluch, and Neil J. Mitchell. 2007. “Security Forces and Sexual Violence: A Cross-National Analysis of a Principal—Agent Argument.” Journal of Peace Research 44 (6): 669–87. https://doi.org/10.1177/0022343307082058. Chang, Alvin. 2016. “Police Academies Spend 60 Hours on Firearms — and Just 8 Hours on Conflict Management.” Vox. July 7, 2016. https://www.vox.com/2016/7/7/12118906/police-training-mediation. Chilton, Adam S., and Mila Versteeg. 2016. “Do Constitutional Rights Make a Difference.” American Journal of Political Science 60 (3): 575–89. Cingranelli, David L., and Mikhail Filippov. 2010. “Electoral Rules and Incentives to Protect Human Rights.” Journal of Politics 72 (1): 243–57. Cingranelli, David L., and David L. Richards. 2010. “The Cingranelli and Richards (CIRI) Human Rights Data Project.” Human Rights Quarterly 32 (2): 401–24. Cingranelli, David L., David L. Richards, and K. Chad Clay. 2014. “CIRI Human Rights Data Project: Short Variable Descriptions for Indicators in The Cingranelli-Richards (CIRI) Human Rights Dataset.” ———. 2021. “The CIRI Human Rights Dataset, Version 2014.04.14.” Harvard Dataverse. https://doi.org/10.7910/DVN/UKCPXT. Cole, Wade M., and Francisco O. Ramirez. 2013. “Conditional Decoupling: Assessing the Impact of National Human Rights Institutions, 1981 to 2004.” American Sociological Review 78 (4): 702–25. https://doi.org/10.1177/0003122413492762. 39 Conrad, Courtenay R., Sarah E. Croco, Brad T. Gomez, and Will H. Moore. 2018. “Threat Perception and American Support for Torture.” Political Behavior 40 (4): 989–1009. https://doi.org/10.1007/s11109-017-9433-5. Conrad, Courtenay R., Jillienne Haglund, and Will H. Moore. 2013. “Disaggregating Torture Allegations: Introducing the Ill-Treatment and Torture (ITT) Country-Year Data.” International Studies Perspectives 14 (2): 199–220. Conrad, Courtenay R, Jillienne Haglund, and Will H Moore. 2014. “Torture Allegations as Events Data: Introducing the Ill-Treatment and Torture (ITT) Specific Allegation Data.” Journal of Peace Research 51 (3): 429–38. Cook, Scott J., and David Fortunato. 2022. “The Politics of Police Data: State Legislative Capacity and the Transparency of State and Substate Agencies.” American Political Science Review, July, 1–16. https://doi.org/10.1017/S0003055422000624. Curtice, Travis B. 2021. “How Repression Affects Public Perceptions of Police: Evidence from Uganda.” Journal of Conflict Resolution 65 (10): 1680–1708. Curtice, Travis B., and Brandon Behlendorf. 2021. “Street-Level Repression: Protest, Policing, and Dissent in Uganda.” Journal of Conflict Resolution 65 (1): 166–94. Davenport, Christian. 1999. “Human Rights and the Democratic Proposition.” Journal of Conflict Resolution 43 (1): 92–116. ———. 2005. “Understanding Covert Repressive Action: The Case of the U.S. Government against the Republic of New Africa.” The Journal of Conflict Resolution 49 (1): 120–40. ———. 2007a. State Repression and the Domestic Democratic Peace. New York: Cambridge University Press. 40 ———. 2007b. “State Repression and Political Order.” Annual Review of Political Science 10 (1): 1–23. https://doi.org/10.1146/annurev.polisci.10.101405.143216. Davenport, Christian, and David Armstrong. 2004. “Democracy and the Violation of Human Rights: A Statistical Analysis from 1976-1996.” American Journal of Political Science 48 (3): 538–54. De Bruin, Erica. 2021a. “Online Appendix and Codebook: The State Security Forces Dataset, 1960-2010.” ———. 2021b. “Mapping Coercive Institutions: The State Security Forces Dataset, 1960–2010.” Journal of Peace Research 58 (2): 315–25. https://doi.org/10.1177/0022343320913089. ———. 2021c. “Policing Insurgency: Are More Militarized Police More Effective?” Small Wars & Insurgencies 0 (0): 1–25. https://doi.org/10.1080/09592318.2021.1980186. Delehanty, Casey, Jack Mewhirter, Ryan M. Welch, and Jason Wilks. 2017. “Militarization and Police Violence: The Case of the 1033 Program.” Research and Politics. https://doi.org/10.1177/2053168017712885. della Porta. 2011. Clandestine Political Violence. New York: Cambridge University Press. DeMeritt, Jacqueline H. R. 2015. “Delegating Death: Military Intervention and Government Killing.” Journal of Conflict Resolution 59 (3): 428–54. https://doi.org/10.1177/0022002713515406. ———. 2016. “The Strategic Use of State Repression and Political Violence.” Oxford Research Encyclopedia of Politics, October. https://doi.org/10.1093/acrefore/9780190228637.013.32. 41 DeMeritt, Jacqueline H.R., and Courtenay R. Conrad. 2019. “Repression Substitution: Shifting Human Rights Violations in Response to UN Naming and Shaming.” Civil Wars 21 (1): 128–52. https://doi.org/10.1080/13698249.2019.1602805. Durán-Martínez, Angélica, and Hillel David Soifer. 2021. “The Drug Trade and State Violence in Internal Conflicts: Evidence from Peru.” Latin American Politics and Society 63 (4): 96–123. https://doi.org/10.1017/lap.2021.39. Earl, Jennifer, Sarah A. Soule, and John D. McCarthy. 2003. “Protest under Fire? Explaining the Policing of Protest.” American Sociological Review 68 (4): 581–606. Eck, Kristine, Courtenay R. Conrad, and Charles Crabtree. 2021. “Policing and Political Violence.” Journal of Conflict Resolution 65 (10): 1641–56. https://doi.org/10.1177/00220027211013083. Engelhardt, Andrew M., Stanley Feldman, and Marc J. Hetherington. 2021. “Advancing the Measurement of Authoritarianism.” Political Behavior, May. https://doi.org/10.1007/s11109-021-09718-6. Fariss, Christopher J. 2014. “Respect for Human Rights Has Improved Over Time: Modeling the Changing Standard of Accountability.” The American Political Science Review 108 (2): 297–318. Flom, Hernán. 2022. “The Politics of Police Violence: Political Competition and Police Killings in Brazil.” Journal of Urban Affairs 0 (0): 1–23. https://doi.org/10.1080/07352166.2021.2018935. Flores-Macías, Gustavo A., and Jessica Zarkin. 2021. “The Militarization of Law Enforcement: Evidence from Latin America.” Perspectives on Politics 19 (2): 519–38. https://doi.org/10.1017/S1537592719003906. 42 FRANCE 24 English, dir. 2020. Thousands Rally in Fresh Paris Protest against Racism and Police Brutality. https://www.youtube.com/watch?v=tGXTMQ7r2QU. González, Yanilda María. 2019. “The Social Origins of Institutional Weakness and Change: Preferences, Power, and Police Reform in Latin America.” World Politics 71 (1): 44–87. https://doi.org/10.1017/S004388711800014X. ———. 2020. Authoritarian Police in Democracy: Contested Security in Latin America. Cambridge Studies in Comparative Politics. Cambridge: Cambridge University Press. https://doi.org/10.1017/9781108907330. ———. 2023. “Reforming to Avoid Reform: Strategic Policy Substitution and the Reform Gap in Policing.” Perspectives on Politics 21 (1): 59–77. https://doi.org/10.1017/S1537592722000135. Greitens, Sheena Chestnut. 2016. Dictators and Their Secret Police: Coercive Institutions and State Violence. Cambridge University Press. Gurr, Ted Robert. 1986. “The Political Origins of State Violence and Terror: A Theoretical Analysis.” Government Violence and Repression: An Agenda for Research 45: 28. Harig, Christoph. 2020. “Soldiers in Police Roles.” Policing and Society 30 (9): 1097–1114. https://doi.org/10.1080/10439463.2019.1650745. Heyer, Garth den. 2014. “Mayberry Revisited: A Review of the Influence of Police Paramilitary Units on Policing.” Policing and Society 24 (3): 346–61. https://doi.org/10.1080/10439463.2013.784304. Hill, Daniel W., and Zachary M. Jones. 2014. “An Empirical Evaluation of Explanations for State Repression.” The American Political Science Review 108 (3): 661–87. 43 Huber, John D., and Charles R. Shipan. 2002. Deliberate Discretion?: The Institutional Foundations of Bureaucratic Autonomy. Cambridge University Press. Kappeler, Victor E., and Peter B. Kraska. 2015. “Normalising Police Militarisation, Living in Denial.” Policing and Society 25 (3): 268–75. https://doi.org/10.1080/10439463.2013.864655. Keith, Linda Camp. 2002. “Judicial Independence and Human Rights Protection around the World.” Judicature 85 (4): 194–200. Keith, Linda Camp, C. Neal Tate, and Steven C. Poe. 2009. “Is the Law a Mere Parchment Barrier to Human Rights Abuse?” Journal of Politics 71 (2): 644–60. Kenzer, Benjamin. 2020. “Do U.S. Policing Programs Help Boost the Militarization of Foreign Police Forces?” Washington Post, July 10, 2020, sec. Monkey Cage. https://www.washingtonpost.com/politics/2020/07/10/do-us-policing-programs-helpboost-militarization-foreign-police-forces/. Kirisci, Mustafa. 2022. “Militarized Law Enforcement Forces, State Capacity and Terrorism.” Terrorism and Political Violence 34 (1): 93–112. https://doi.org/10.1080/09546553.2019.1668376. Kitchen, Veronica, and Kim Rygiel. 2014. “Privatizing Security, Securitizing Policing: The Case of the G20 in Toronto, Canada.” International Political Sociology 8 (2): 201–17. https://doi.org/10.1111/ips.12052. Koslicki, Wendy M. 2021. “Recruiting Warriors or Guardians? A Content Analysis of Police Recruitment Videos.” Policing and Society 31 (6): 702–20. https://doi.org/10.1080/10439463.2020.1765778. 44 Kraska, Peter B. 2001. “Playing War: Masculinity, Militarism, and Their Real-World Consequences.” In Militarizing the American Criminal Justice System: The Changing Roles of the Armed Forces and the Police. Boston: Northeastern University Press. ———. 2007. “Militarization and Policing — Its Relevance to 21st Century Police.” Policing 1 (4): 501–13. Lawson, Edward. 2019. “TRENDS: Police Militarization and the Use of Lethal Force.” Political Research Quarterly 72 (1): 177–89. Linzer, Drew A., and Jeffrey K. Staton. 2015. “A global measure of judicial independence, 1948–2012.” Journal of Law and Courts 3 (2): 223–56. Lowande, Kenneth. 2021. “Police Demilitarization and Violent Crime.” Nature: Human Behaviour 5 (February): 205–11. Magaloni, Beatriz, and Luis Rodriguez. 2020. “Institutionalized Police Brutality: Torture, the Militarization of Security, and the Reform of Inquisitorial Criminal Justice in Mexico.” American Political Science Review 114 (4): 1013–34. https://doi.org/10.1017/S0003055420000520. Marshall, Monty G, Ted Robert Gurr, and Keith Jaggers. 2015. “Polity IV Project: Political Regime Characteristics and Transitions, 1800-2015.” http://www.systemicpeace.org/inscrdata.html. Masera, Federico. 2021. “Police Safety, Killings by the Police, and the Militarization of US Law Enforcement.” Journal of Urban Economics 124: 1–11. Maslow, Abraham H. 1969. The Psychology of Science: A Reconnaissance. Henry Regnery Company. 45 McDonald, Brent, Ainara Tiefenthäler, and James Surdam. 2023. “How Peru Used Lethal Force to Crack Down on Anti-Government Protests.” The New York Times, March 16, 2023, sec. World. https://www.nytimes.com/2023/03/16/world/americas/peru-protestspolice.html. Mesquita, Bruce Bueno de, George W. Downs, Alastair Smith, and Feryal Marie Cherif. 2005. “Thinking Inside the Box: A Closer Look at Democracy and Human Rights.” International Studies Quarterly 49 (3): 439–57. Mitchell, Neil J. 2004. Agents of Atrocity. New York: Palgrave Macmillan US. https://doi.org/10.1057/9781403973696. Moore, Will H. 2000. “The Repression of Dissent: A Substitution Model of Government Coercion.” The Journal of Conflict Resolution 44 (1): 107–27. ———. 2010. “Incarceration, Interrogation, and Counterterror: Do (Liberal) Democratic Institutions Constrain Leviathan?” PS: Political Science & Politics 43 (3): 421–24. https://doi.org/10.1017/S1049096510000661. Moore, Will H., and Ryan M. Welch. 2015. “Why Do Governments Abuse Human Rights?” In Emerging Trends in the Social and Behavioral Sciences: Interdisciplinary Directions, edited by Robert A. Scott and Stephen M. Kosslyn. Hoboken, NJ: Wiley. Morden, John W. 2012. “Independent Civilian Review into Matters Relating to the G20 Summit.” Toronto: Toronto Police Services Board. https://tpsb.ca/g20/ICRG20Mordenreport.pdf. Neder, Renata. 2017. “Brazil: Police Killings, Impunity and Attacks on Defenders: Amnesty International Submission for the UN Universal Periodic Review – 27th Session of the UPR Working Group, May 2017.” Amnesty International. 46 Owens, Emily, David Weisburd, Karen L. Amendola, and Geoffrey P. Alpert. 2018. “Can You Build a Better Cop?” American Society of Criminology 17 (1): 41–87. Passos, Anaís M. 2021. “Breaking the Law to Ensure Order: The Case of Tijuana (2007–2012).” Bulletin of Latin American Research 40 (2): 251–66. https://doi.org/10.1111/blar.13135. Poe, Steven C. 2004. “The Decision to Repress: An Integrative Theoretical Approach to the Research on Human Rights and Repression.” In Understanding Human Rights, edited by Sabine C. Carey and Steven C. Poe. Burlington, VT: Ashgate. Poe, Steven C., and C. Neal Tate. 1994. “Repression of Human Rights to Personal Integrity in the 1980s: A Global Analysis.” American Political Science Review 88 (4): 853–72. https://doi.org/10.2307/2082712. Powell, Emilia Justyna, and Jeffrey K. Staton. 2009. “Domestic Judicial Institutions and Human Rights Treaty Violation.” International Studies Quarterly 53 (1): 149–74. Pozo, Brandon del. 2020. “Opinion | Watch This Protest Turn From Peaceful to Violent in 60 Seconds.” The New York Times, June 26, 2020, sec. Opinion. https://www.nytimes.com/interactive/2020/06/26/opinion/blm-protests-policeviolence.html. Reuters. 2020. “Riot Police Fire Teargas on Anti-Racism Protesters in Paris.” The Guardian, June 13, 2020, sec. World news. https://www.theguardian.com/world/2020/jun/13/riotpolice-fire-teargas-on-anti-racism-protesters-in-paris. Revkin, Mara R. 2022. “How Does Subnational Variation in Repression Affect Attitudes toward Police? Evidence from Iraq’s 2019 Protests.” Violence: An International Journal 3 (1): 85–99. https://doi.org/10.1177/26330024211070318. 47 Ron, James. 1997. “Varying Methods of State Violence.” International Organization 51 (2): 275–300. Simckes, Maayan, Anjum Hajat, Debra Revere, Ali Rowhani-Rahbar, and Dale Willits. 2019. “A Conceptualization of Militarization in Domestic Policing.” Police Quarterly 22 (4). Simmons, Beth A. 2009. Mobilizing for Human Rights: International Law in Domestic Politics. New York: Cambridge University Press. Soss, Joe, and Vesla Weaver. 2017. “Police Are Our Government: Politics, Political Science, and the Policing of Race–Class Subjugated Communities.” Annual Review of Political Science 20 (1): 565–91. https://doi.org/10.1146/annurev-polisci-060415-093825. Soule, Sarah A., and Christian Davenport. 2009. “Velvet Glove, Iron Fist, or Even Hand? Protest Policing in the United States, 1960-1990.” Mobilization: An International Journal 14 (1): 1–22. Stoughton, Seth. 2015. “Law Enforcement’s ‘Warrior’ Problem.” Harvard Law Review 128 (6). https://harvardlawreview.org/2015/04/law-enforcements-warrior-problem/. Tecott, Rachel, and Sara Plana. 2016. “Maybe U.S. Police Aren’t Militarized Enough. Here’s What Police Can Learn from Soldiers.” Washington Post, August 16, 2016, sec. Monkey Cage. https://www.washingtonpost.com/news/monkey-cage/wp/2016/08/16/maybe-u-spolice-arent-militarized-enough-soldiers-are-better-trained-to-deescalate/. Terpstra, Jan. 2011. “Two Theories on the Police – The Relevance of Max Weber and Emile Durkheim to the Study of the Police.” International Journal of Law, Crime and Justice 39 (1): 1–11. https://doi.org/10.1016/j.ijlcj.2011.01.009. The New York Times, dir. 2017. The Anti-Protest Gear Used in Venezuela | NYT Investigates. https://www.youtube.com/watch?v=350q14b_9lA. 48 ———, dir. 2020. Philadelphia Police Violated Their Own Guidelines, Here’s How | Visual Investigations. https://www.youtube.com/watch?v=btyOeHpETf8. Tilly, Charles. 2003. The Politics of Collective Violence. New York: Cambridge University Press. Wall Street Journal, dir. 2020. Video Investigation: How a Seattle Protest Ended in Chaos | WSJ. https://www.youtube.com/watch?v=MMjExztMbyA. Weber, Max. 2018. “Politics as a Vocation.” In Classical Sociological Theory and Foundations of American Sociology, by Allison L. Hurst. Oregon State University. https://open.oregonstate.education/sociologicaltheory/chapter/politics-as-a-vocation/. Welch, Ryan M. 2017. “National Human Rights Institutions: Domestic Implementation of International Human Rights Law.” Journal of Human Rights 16 (1): 96–116. https://doi.org/10.1080/14754835.2015.1103166. Welch, Ryan M., Jacqueline H. R. DeMeritt, and Courtenay R. Conrad. 2021. “Conceptualizing and Measuring Institutional Variation in National Human Rights Instituions (NHRIs).” Journal of Conflict Resolution 65 (5): 1010–33. Welch, Ryan M, Jack Mewhirter, James E Wright II, and Jeongmin Oh. Forthcoming. “Evaluating the Determinants of Support for Police Militarization among Officers.” Politics and Policy. https://12a65534-1710-9c11-a8fcf9f4b30690b9.filesusr.com/ugd/f3f1e9_b8eebc6425304c4f8a2a939f2b0fb661.pdf. Wood, Reed M., and Mark Gibney. 2010. “The Political Terror Scale (PTS): A Re-Introduction and a Comparison to CIRI.” Human Rights Quarterly 32 (2): 367–400. 49 World Bank. 2015. “The World Bank. World Development Indicators 2015.” The World Bank. https://datacatalog.worldbank.org/search/dataset/0037712/World-DevelopmentIndicators. Zarkin, Jessica. 2023. “The Silent Militarization: Explaining the Logic of Military Members’ Appointment as Police Chiefs.” Comparative Politics. 50 Footnotes i For full replication materials, please visit the following GitHub repository: https://github.com/mstavro/DPMIR/ ii For example, consider Hill and Jones (2014)’s model using machine learning models to adjudicate between the most frequently used explanans of human rights. Though, see Conrad, Haglund, and Moore (2013, 2014) for important exceptions in the torture literature. iii The Seattle Police Department Manual (https://www.seattle.gov/police-manual) requires issuing an order to disperse before using these tactics (14.090-TSK-3). iv We define authoritarianism as a strict adherence to tradition and hierarchy often enforced through violence (see Engelhardt, Feldman, and Hetherington 2021 for a recent and detailed discussion about the concept). v See Zarkin (2023) for an example of work that combines elements of both police militarization and constabularization by exploring why governments appoint military personnel in policing leadership positions. vi De Bruin (2021c) represents a notable exception that tests police militarization’s effects generally across a global sample. vii For reviews of the literature see Davenport (2007a) and Moore and Welch (2015). For a comprehensive empirical evaluation, see Hill and Jones (2014). viii We refer to the principal here as state decision-makers, leaders, or policymakers provided our aim is to present a generalizable theory. The specific actor delegating repression to the police through militarization is context-dependent (whether it be the executive, legislators, etc.). 51 ix Policing institutions possess relatively large amounts of autonomy compared to other bureaucracies (González 2019), and as such, they may choose to militarize themselves rather than receive militarization directly from the government. However, at the very least, the government must allow it to take place, or even facilitate it with legislation. Doing so signals to the police that militarization is on the table. Here we remind the reader that our theoretical model is a simplification of the world, but one that can incorporate more complicated processes like this without changing the overarching conclusions x Surely, moments occur when police need to legitimately react with violence. Our argument is that the chances of illegitimate violence for the sake of threat-reduction—repression—occurs more often when officers perceive threats as ubiquitous. xi This strikes us a fruitful area for future research. xii For readers interested in geographic scope and temporal bias on the dependent variable, see the online appendix, specifically the section entitled “The Sample” as well as figures A3 and A4. xiii Much of the present work on repression uses the latent variable from Fariss (2014). However, that measure operationalizes repression in a way that includes politicide, democide, and genocide (303), tactics that militarized police are less likely to employ compared to other state security elements. Specifying torture, extrajudicial killing, imprisonment, and disappearance is a more valid operationalization of the capabilities and mandate of militarized police as opposed to mass killing incidents such as genocide, politicide, or democide. xiv See tables A1 through A3 as well as figures A1 and A2 for descriptive statistics. xv Many of the countries remained militarized throughout the sample (58%); some never militarized (31%), and still others changed militarization status (11%). Of those who changed 52 their militarization status, 62% militarized, while 38% demilitarized. See table A14 for information on each country’s militarization status in the sample. xvi Due to the ordinal nature of the response variable, ordered logit is most appropriate, although scholars do employ linear regression treating categorical variables as continuous when enough categories exist. We use partial proportional odds to address violations of the proportional odds (parallel lines) assumption. xvii Recall we created the repression score by inverting the CIRI index, thus higher values indicate higher levels of repression. xviii In the partial proportional odds model, we treat the effect of those variables that failed the parallel regression assumption (latent judicial independence and the intercept) as nominal. Interested readers can find those coefficient estimates in table A9 of the online appendix. xix Given that each component only has 3 levels (0, 1, and 2), we do not run OLS regression. xx Readers can find descriptive statistics for each repression tactic in tables A10 through A13 of the online appendix. xxi Thus, there is no partial proportional odds model for torture in table A7. xxii Limiting the dependent variable in these analyses to only newly militarized police avoids estimating a model that predicts a string of 1s on the police militarization variable. We find this more suitable because it better captures the decision of whether the government would decide to militarize when currently not militarized. xxiii This context is especially important since racial attitudes affect police officers’ support for militarization (Welch et al. Forthcoming) 53 xxiv Other recent work has taken a similar approach. For example, Flores-Macías and Zarkin (2021, 521) develop what they describe as a continuum, from least to most militarized that steps up along four conceptual categories: non-militarized, militarized, para-militarized, constabularized. To determine which step the police fall upon, they ask four questions about their accountability, weaponry, training, and organization. The answers to these questions then allow them to place a country in a category, one of which encompasses the militarization we are concerned with in this paper. As our supplementary analysis in table A16 suggests, constabularization is a separate concept rather than a higher value along a continuum. Future research should parse this out more. Future research should also dig deeper with militarized police (and other categories) to account for variation among the dimensions that make an LEO militarized. xxv Results in table A8 of the online appendix. xxvi Results in table A5 of the online appendix.