PROTESTING THE POLICE: HOW SITUATIONAL THREATS ELICIT POLICE REPRESSION AT PROTEST EVENTS TARGETING THE POLICE. By Gabrielle Maria Horne May, 2022 Director of Thesis: Dr. Bob Edwards Major Department: Sociology ABSTRACT Protests challenging the police pose a significant reputational threat to law enforcement. The threat hypothesis states that police repression is motivated by the desire to maintain social control, and when threats arise, police will act to incapacitate them. Reputational threat literature has established that police are more likely to be present and intervene at protests with goals challenging them. However, an investigation of police action at only protests that target the police has yet to be conducted. Using the Armed Conflict Location & Event Data Project (ACLED), all protest events from the month of June 2020 with goals targeting the police (n=5,900) – in support of BLM, anti-police brutality, against the death of George Floyd or other minority individuals killed by police, and other goals seeking to remove the social control of police – were analyzed to investigate how police intervene (arrest, use force/violence, or both arrest and use force/violence) at these events when situational threats are present. The results of this study illustrate police are more likely to intervene at protests targeting them when situational threats were present. Furthermore, the forms of situational threats present elicited different forms of policing interventions. PROTESTING THE POLICE: HOW SITUATIONAL THREATS ELICIT POLICE REPRESSION AT PROTEST EVENTS TARGETING THE POLICE. A Thesis Presented To the Faculty of the Department of Sociology East Carolina University In Partial Completion of the Requirements for the Degree Master of Art in Sociology by Gabrielle Maria Horne May, 2022 © Gabrielle Maria Horne, 2022 PROTESTING THE POLICE: HOW SITUATIONAL THREATS ELICIT POLICE REPRESSION AT PROTEST EVENTS TARGETING THE POLICE. By Gabrielle Maria Horne APPROVED BY: Director of Thesis __________________________________________ Bob Edwards, PhD Committee Member __________________________________________ Mamadi Corra, PhD Committee Member __________________________________________ Kirk Miller, PhD Committee Member __________________________________________ Dmitry Tumin, PhD Chair of the Department of Sociology __________________________________________ Mamadi Corra, PhD Dean of the Graduate School __________________________________________ Paul J. Gemperline, PhD TABLE OF CONTENTS LIST OF TABLES ……………………………………………………………………. vi LIST OF FIGURES …………………………………………………………………… vii INTRODUCTION ……………………………………………………………………. 1 THREAT LITERATURE ……………………………………………………………. 5 Reputational Threats …………………………………………………………… 9 Situational Threats …………………………………………………………… 11 METHODOLOGY …………………………………………………………………… 16 Sample …………………………………………………………………… 20 Coding Methodology …………………………………………………………… 21 Dependent Variable …………………………………………………………… 22 Independent Variables …………………………………………………… 25 Control Variables …………………………………………………………… 27 Statistical Analysis …………………………………………………………… 29 RESULTS ……………...…………………………………………………………… 31 DISCUSSION …………………………………………………………………………... 38 CONCLUSION ……...…………………………………………………………… 44 REFERENCES ……...…………………………………………………………… 45 APPENDIX A: “DIRTY HARRY” OR USE OF FORCE/VIOLENCE POLICE INTERVENTION …………………………………………………………………… 49 APPENDIX B: CONFRONTATIONAL TACTICS……………………………………. 50 APPENDIX C: PROTESTER CLASHES WITH THE POLICE………………………. 51 APPENDIX D: COUNTER-PROTESTERS & LONE WOLVES……………………… 52 FIGURE 1: FREQUENCY OF JUNE 2020 PROTEST EVENTS TARGETING THE POLICE BY STATE …………………………………………………………………………… 53 FIGURE 2: FREQUENCY OF PROTEST EVENTS BY DAY OVER JUNE 2020…… 54 FIGURE 3: FLOWCHART OF THREAT LEVELS OF OPERATION………………... 55 FIGURE 4: CYCLE OF DISORDER FROM SITUATIONAL THREATS AT PROTEST EVENTS TARGETING THE POLICE…………………………………………………. 56 TABLE 1: UNIVARIATE DISTRIBUTION OF ALL STUDY VARIABLES………… 57 TABLE 2: DESCRIPTIVE STATISTICS OF STUDY VARIABLES BY POLICE RESPONSE TYPE ……………………...…………………………………………………………… 58 TABLE 3: MULTINOMIAL LOGISTIC REGRESSION OF SITUATIONAL THREATS ON POLICE RESPONSES…………………………………………………………………. 59 LIST OF TABLES 1. Univariate Distribution of All Study Variables…………………………………. 57 2. Descriptive Statistics of Situational threats by Police Response Type………..… 58 3. Multinomial Logistic Regression of Situational Threats on Police Responses…. 59 LIST OF FIGURES 1. Frequency Of June 2020 Protest Events Targeting the Police by State ………… 53 2. Frequency of Protest Events by Day over June 2020…………………………… 54 3. Flowchart of Threat Levels of Operation……………………………………….. 55 4. Cycle of Disorder from Situational Threats at Protest Events Targeting the Police……………………………………………………………………………. 56 PROTESTING THE POLICE: HOW SITUATIONAL THREATS ELICIT POLICE REPRESSION AT PROTEST EVENTS TARGETING THE POLICE. INTRODUCTION On 25 May 2020, George Floyd - an unarmed black man - was killed by a law enforcement officer in Minneapolis, Minnesota. Following his death, America erupted with protest activity at levels not seen since the 1960s and 1970s. Protests in support of the Black Lives Matter (BLM) movement, against police violence, and over the death of George Floyd and other black men and women killed by law enforcement officers occurred in all 50 states and the District of Columbia. Demonstrations occurring in response to police brutality are not a new social phenomenon in the United States. Although the 2020 wave of protest activity marks a new volume of public criticism over police use of force and violence; the social context under which these protest events occurred provided a unique circumstance which aided but also restricted the ability of individuals to collectively engage in protests. SARS-CoV-2, also known as the novel Coronavirus (COVID-19), a disease-causing virus spread throughout the world. Beginning in Wuhan, China on 12 December 2019, the first case of COVID-19 was noted in the United States on 20 January 2020 (CDC 2022). Two months before the death of George Floyd, on 11 March 2020, the World Health Organization (WHO) publicly declared COVID-19 a pandemic, and in the following four days, then President Donald J. Trump used his powers to declare a nationwide emergency and all cruise ships, public school systems, and individual state governments began shutting down and instituting “lock down” procedures (CDC 2022). The unique circumstances stemming COVID-19 created a protesting dilemma. Although the federal government and a majority of state governments closed businesses and schools, expanding individuals political opportunity to protest, those who decided to engage in collective action placed themselves at a greater risk of becoming infected with COVID-19. In a working paper by Dave et al. (2021), researchers examined the relationship between 2020 protests participation the epidemiological risk-factors associated with the spread of COVID-19; their findings reflected an elevation in American’s remaining at home, rather than participating in protests, following the initial onset of Black Lives Matter (BLM) protests. The BLM movement is a multi-generational and multi-ethnic social movement specifically protesting the social issue of police brutality (Harris 2021). Protest events that target the police through their goals, such as anti-police brutality, pro-police reform, or others that challenge the legitimacy of law enforcement create a protest policing paradox. This paradox stems from the responsibility of the police to maintain social order while representing the source of contention, at these events the presence of police can antagonize protesters and potentially lead to confrontational or violent behaviors. Equally, police tasked with maintaining control at protests that target them have demonstrated an increased likelihood of more repressive protest policing interventions (Reynolds-Stenson 2018). The 2020 protest wave provides a robust sample of events to examine how police respond to protests that specifically target them. The current policing strategy of protest policing is called ‘strategic incapacitation’ (Gillham 2011; Gillham and Marx 2018; Gillham, Edwards, and Noakes 2013). Under this approach, police are tasked with identifying, neutralizing, and incapacitating threats, or potential threats, at a protest event. Strategic incapacitation relies upon preemptive law enforcement practices, this aspect of the policing strategy means protest policing extends beyond the bounds of a protest event to the hours and days before the event (Gillham 2011; Gillham, Edwards, and Noakes 2013). The priority of threat identification, neutralization, and incapacitation means policing responses to protests are dependent on the threats associated 2 with, or present at, an event. Protest policing scholars use the threat hypothesis to explain how policing interventions at a protest are determined by how police perceive threats and their desire to maintain public control and order (Earl, Soule, and McCarthy 2003; Davenport, Soule, and Armstrong 2011; Myers-Montgomery 2016; Nassauer 2014; Reynolds-Stenson 2018; Soule and Davenport 2009). Protest policing is guided by the identification, analysis, and neutralization of threats. Although uniquely operationalized across policing discourse, threats are sources, or potential sources, of civil disorder that police utilize mechanisms of social control against to reestablish social control (Blalock 1967). In protest policing literature, threats are broken into two groups (1) reputational threats and (2) situational threats (Earl 2003; Earl, Soule, and McCarthy 2003; Davenport, Soule, and Armstrong 2011; Reynolds-Stenson 2018; Soule and Davenport 2009). Reputational threats are conceptualized at threats to the authority and legitimacy of elite institutions (Earl, Soule, and McCarthy; Reynolds-Stenson 2018). When protesters espouse goals that challenge elite institutions they single a threat to the social order, and police are tasked with the responsibility to neutralizing the threat and regain control (Earl 2003). For this analysis, reputational threats were examined through protest goals that target the police; this is a unique form of threat that directly challenges the legitimacy of police as an institution (Reynolds- Stenson 2018; Trinker, Kerrison, and Goff 2019). Situational threats are behavior that potentially provoke police interventions at protest events because they disrupt the social order through civil disturbance, danger, or violence (Hichman, Piquero, Garner 2008; Kappeler 1992; Earl 2003; Earl, Soule, and McCarthy 2003; Davenport, Soule, and Armstrong 2011; Reynolds-Stenson 2018). This analysis seeks to understand how police respond to situational threats at protest events that already have the reputational threat of goals targeting the police present. 3 To examine protest policing interventions at events that specifically target them and the situational threats that elevate the likelihood of police repression at such events, all protest events that occurred during the month of June 2020 in the U.S. were examined across four types of police intervention (no intervention, arrests, use of force/violence, or both arrests and use of force/violence). Over 22,000 protests happened over the course of 2020, 6,816 protests occurred in June, and of those 5,993 targeted the police through their goals. This explosion of protest activity can be traced to the death of George Floyd on 25 May 2020. Protests targeting the police stem from contempt of law enforcement, such events challenge the inflated self-image of officers, called the ‘thin blue line’ (Lyles and Esmail 2016:171). When the self-image and legitimacy of police are challenged police presence, action, and abuse become more likely (Lyles and Esmail 2016; Reynolds-Stenson 2018). Police were challenged multilaterally by this protest wave with events occurring in all 50 states that targeted the police and the presence of situational threats at particular events presented further challenges to protest policing. However, there was geographic heterogeneity among the number of protests in each state (range: 640 California - 14 North Dakota and South Dakota) (See Figure 1), meaning the level of threat experiences by police varied across the nation. 4 THREAT LITERATURE Blalock’s threat hypothesis of police control states that, when a minority population threatens social order, the dominant population will mechanize social control to reassert their power (Blalock 1967). Police are repressive state agents that are instruments of elites, tasked with the responsibility of maintaining social order and the law (Earl 2003; Lyle and Esmail 2016). Therefore, police are motivated by the goals of the elite making them a facet of domination and power. Scholars have utilized the threat hypothesis to explain how police are motivated by racialized repression, where the number of minority residents is positively correlated with the level of threat felt among the majority population. However, once a minority population becomes the majority, the presence of threat associated with racialized conceptions of crime will decrease (Miller 2009). A recent report by the U.S. Commission of Civil Rights (2018) cited an elevation in use of force among law enforcement agencies (LEAs) in America. Additionally, the instances of police use of force were disproportionately concentrated in minority communities (racial and ethnic minorities, LGBT, physical and mentally disabled peoples, low income, and those at the intersection of these populations). Correll, et al. (2002) found all law enforcement officers are affected by racial bias, pointing to experiences with cultural stereotypes and the extent of contact with the individuals they police determining the level of bias, rather than personal racial prejudice. The threat hypothesis has been used to explain environmental influences on a variety of other forms of state violence, such as arrests (Eitle, D’Alessio, and Stolzenberg 2002; Parker, Stults, and Rice 2005), disparities in sentencing (Lowery, Burrow, and Kaminski 2016), and traffic enforcement (Novak and Chamlin 2012). Miller (2009) used the threat hypothesis to explain the likelihood of LEAs adopting anti-profiling policies, noting the positive association between aggressive forms of policing, like the use of deadly force, and the increase of black residents in a predominantly white community. Smith (2003) similarly used the threat hypothesis in his investigation of officer diversity and police-caused homicides. The threat hypothesis has been adopted by protest policing scholars to explain social control at protest events. Primarily, law enforcement repression, or social control, at a protest is examined through police presence and intervention in response to situational threats attendant to the protest. The policing interventions designed to achieve social control at protests include limited action, such as traffic control or collecting information from protesters; arrests; use of force or violence, such as less-lethal weapons or assault; or a combination of two or more interventions (Earl 2003; Earl, Soule, and McCarthy 2003; Reynolds-Stenson 2018; Soule and Davenport 2009). In protest policing literature, the conceptualization of threat is not limited to minority presence. However, the presence of racial minorities at a protest event has been examined both as a form of threat (Davenport, Soule, and Armstrong 2011) and weakness (Earl 2003), and the presence of minorities has been consistently associated with elevated levels of police repression. Although the weakness hypothesis sounds contradictory to the threat hypothesis, both associate elevated numbers of minority individuals at protests events with an increased likelihood of police repression. Rather, the weakness hypothesis argues police exploit the perceived inability of minorities to respond to repressive policing tactics (Earl, Soule, and McCarthy 2003; Davenport, Soule, and Armstrong 2011). Weakness is limited to groups that lack resources, an organized structure, or the ability to react to police repression; therefore, social movement organizations (SMOs) that are led by minority groups would be excluded from the weakness hypothesis. 6 Davenport, Soule, and Armstrong’s (2011) examined policing responses at protests events using the illustrated that, although increasing the likelihood of presence and intervention, African American led protest did not always elicit distinct interventions from those at predominantly white protests. Trends of severe police repression (arrests and use of force/violence) at predominantly minority protests ended in the 1970s; while in the 1980s police were more likely to not intervention at such protests in comparison to majority-white events (Davenport, Soule, Armstrong 2011). The inconsistency of policing interventions at protests based on the race and ethnicity of participants points to other protest elements driving police repression at contemporary demonstrations. For protest policing, the threat hypothesis is applied in a generalized context to examine multiple protest elements as forms of threat. Adding to the traditional threat hypothesis which specifically focuses on racial composition and conflict, the protest policing threat hypothesis states that police actions at a protest occur in response to the perception of threat at an event, and threats can be both reputational and situational (Earl 2003; Earl, Soule, and McCarthy 2003; Davenport, Soule, and Armstrong 2011; Myers-Montgomery 2016; Nassauer 2014; Reynolds- Stenson 2018; Soule and Davenport 2009). Police identify threats by assessing whether the characteristics of a protests can become sources of disorder that jeopardize law enforcement’s legitimacy or ability to maintain control and order. Soule and Davenport (2009) suggested that the policing of protest events occurs in a two-stage process; first police decide whether to attend an event. Then, once present, police identify sources of threat and choose whether to act and what form of intervention will best achieve social control. Police presence at a protest is consistently more likely than police taking action (Reynolds-Stenson 2018; Soule and Davenport 7 2009). Additionally, egregious forms of protest policing, such as police-caused fatalities, are extremely uncommon (Earl, Soule, and McCarthy 2003). Threatening protest characteristics have been conceptualized as reputational threats – revolutionary or racial goals – and situational threats – large protest events, the presence of confrontational tactics (Earl, Soule, and McCarthy 2003), property damage, violence, and the presence of counter-demonstrators (Reynolds-Stenson 2018; Soule and Davenport 2009) and/or third-party agitators (Myers-Montgomery 2016). Earl (2011) noted the goal of the police is to identify and neutralize threats objectively, meaning ideal protest policing interventions will not be subject to the biases of individual officers. However, scholars have noted that reputational threats trigger unneutral protest policing intervention, with protest goals targeting the police increasing the likelihood of police presence and repression at such events (Reynolds-Stenson 2018). Because police function as an institution (Crank and Langworthy 1992), reputational threats affect law enforcement officers who do not directly experience challenges to their authority or legitimacy, which increases the likelihood of police adopting immoral and repressive practices (Trinker, Kerrison, and Goff 2019). Therefore, the subjectivity involved in the identification and neutralization of threats at protests that challenge police’s legitimacy requires an investigation to better understand which particular forms of threat, as such events, drive repressive and force-reliant policing interventions. 8 Reputational Threats In protest policing literature, reputational threats are conceptualized as protest goals that threaten the authority and legitimacy of elite institutions (Earl, Soule, and McCarthy 2003; Reynolds-Stenson 2018). Protests espousing radical goals that target elite institutions have an increased likelihood of police presence and intervention when compared to protests with non- revolutionary goals (Earl 2003; Earl 2011; Earl, Soule, and McCarthy 2003; Reynolds-Stenson 2018; Soule and Davenport 2009). Protests that challenge the police are a direct reputational threat to law enforcement as an institution, this form of threat can also be understood as “threats to legitimacy” (Crank and Langworthy 1992; Terrill and Paoline 2015; Trinker, Kerrison, Goff 2019). Threats to police’s legitimacy are closely tied to stereotype threat, where individual officers are threatened by the association to racism or police brutality (Trinker, Kerrison, Goff 2019; Steele and Aronson 1995). When the legitimacy of police is threatened and individual officers are fearful of confirming negative stereotypes of their profession, the likelihood of repressive and immoral policing increases (Trinker, Kerrison, Goff 2019). At protests targeting the police, the likelihood of police presence and repression is higher than other revolutionary goals (Reynolds-Stenson 2018). Anti-police brutality protests pose a unique dilemma to law enforcement, due to the risk of reputational damage when policing such events. At these events, police face a crisis of legitimacy, where the protest goals challenge law enforcement’s authority, which represent a threat in the eyes of police triggering the desire of police to quell dissent and act on their responsibility to maintain social control and order (Crank and Langworthy 1992; Habermas 1973). The crisis of legitimacy stemming from the 2020 protest wave targeting the police transcends the localized depictions by Crank and Langworthy (1992), with the entire institution 9 of law enforcement having its legitimacy challenged, which is then experienced on an individual level by police officers (Trinker, Kerrison, Goff 2019). However, a majority of American’s support protest that target the police, which is correlated with lowered support of police action at protest event with such goals (Lyle and Esmail 2016; Metcalfe and Pickett 2021). However, when threats and/or violence arise at a protest event, no matter the protest goals, the public will become more supportive of repressive police action (Metcalfe and Pickett 2021). Therefore, when tasked with policing events that target them, law enforcement must maintain order while minimizing the risk of public criticism over their actions (Earl, Soule, and McCarthy 2003). Although past scholars have noted the association between anti-police brutality goals and increased police presence and intervention at such events, situational threats (the protest actors present and the behaviors they exude) are stronger predictors of police action than the goals of a protest event (Reynolds-Stenson 2018). 10 Situational Threat Situational threats are behaviors that threaten social order or individual officers (Hickman, Piquero, Gerner 2008; Kappeler 1992). These threats are commonly conceptualized as civil disobedience, property damage and/or vandalism, contentious behaviors, and violence (Earl, Soule, and McCarthy 2003; Hickman, Piquero, Gerner 2008; Reynolds-Stenson 2018). The diversity of situational threats illustrate that a variety of actions can be perceived as threatening by police at protest events. Additionally, situational threats can emerge from the presence of particular protest actors, with counter-demonstrators and third-pard agitators, like lone wolves, agitating both protesters and the police. Situational threats that disrupt public order have been ubiquitously identified as the strongest predictor of police repression at protests (Earl, Soule, and McCarthy 2003; Reynolds-Stenson 2018; Soule and Davenport 2009). Although both situational and reputational threats are significant predictors of police repression, situational threats have a greater probability of repression. Soule and Davenport (2009) found the intensification of situational threats at a protest event was positively associated with the severity of police intervention. Leading to the developed of the conflict repression nexus which stats, police actions are dependent on the behaviors of protesters and characteristics of the protest event (Soule and Davenport 2009). Waddington’s Flashpoints model can be used to explain the development of disorder at the interactional level, though its suggestion that particularly contentious situational threats, like protester-initiated violence toward the police, trigger further disorder (King and Waddington 2006). It is the compounding of threats, rather than a single breaking point, that increases the likelihood of aggressive police actions. The situational threats examined in this analysis are confrontational tactics, protester clashes with the police, and the presence of counter-protesters and/or lone wolves. 11 Protests with confrontational tactics present have been consistently associated with an increased likelihood of police presence and action (Earl, Soule, and Davenport 2003; Reynolds- Stenson 2018; Nassauer 2014; Soule and Davenport 2009). Confrontational tactics are situational threats that disrupt social order, these include forms of protest, such as riots or sleep-ins, and activities or behaviors of protesters like property damage, verbal and physical altercations, looting, etc. Following the findings of past literature, protests events with confrontational tactics in June 2020 should have the highest likelihood of severe police intervention, when compared to other forms of situational threat. Protester clashes with the police have yet to be conceptualized as a situational threat in protest policing literature. Volatile interactions between protesters and police, that were initiated by protesters, embody confrontational protest behaviors directed specifically at the police. Protester-initiated clashes represent a direct challenge to the authority and power of law enforcement officers. Protest policing scholars have found that situational threats increase the likelihood of police action and reputational threats, such as anti-police brutality protests, have the same effect on police action (Earl, Soule, and McCarthy 2009; Davenport, Earl, and Armstrong 2011; Myers; Montgomery 2016; Nassauer 2014; Reynolds-Stenson 2018; Soule and Davenport 2009). These findings lay the foundation to predict that volatile protester interactions with police represent a niche form of situational threat that may elicit recurring and aggressive forms of police action. Furthermore, following the findings of Reynolds-Stenson (2018) that situational threats are stronger predictors of repression than the goals of a protest, protester clashes with the police should be associated with a higher likelihood of intervention when compared to generalized confrontational tactics. 12 The presence of counter-protester or lone wolves at a protest catalyze violence and amplify the likelihood of negative outcomes (Earl and Davenport 2009; Meyers-Montgomery 2016). Counter-demonstrators represent ideological challenges to protesters, the contention they ferment at protest events leads to their presence being a source of threat to police. The presence of counter-protesters at a protest increases the likelihood of police presence and action (Meyers- Montgomery 2016; Reynolds-Stenson 2018; Soule and Davenport 2009). Lone wolves are rogue protest actors with no ideological connections to the protest, they represent a multifaceted threat to police because they are commonly armed, deliberately agitate protesters, and intentionally victimize the police (ACLED 2021a; Hamm and Spaaj 2015). Although traditionally described as individual actors, lone wolves during the 2020 protest wave sometimes operated in conglomerates of 2 to 30 or 50 people. In a report by the ACLED (2021a), it was noted that political violence initiated by lone wolves at demonstrations increased over 2020. Considering the evidence provided by protest policing literature on police actions using the threat hypothesis, the recent wave of protests targeting the police while bolstering minority communities may have increased the level of threat perceived by police creating a desire to quell mainstream challenges to LEAs. Recent research examining racial and ethnic group participation in BLM protests has found that white people were more likely to have participated in a demonstration focused on racial equality than minorities in the U.S. (Barroso and Minkin 2020). Furthermore, BLM or other protest events with racialized goals were more likely to occur in localities with an average income above $150,000 and with participants under the age of 35 (Buchanan, Bui, and Patel 2020). These findings demonstrate that contemporary social movement activity challenging the police is guided by a multiracial “national alliance” (Rickford 2016:38). However, protests with situational threats provide legitimate sources of public concern 13 and decrease the potential reputational damage associated with police action at such an event, which may open the door for police to be more repressive (Metcalfe and Pickett 2021). Therefore, I hypothesize that: Hypothesis 1: Protests targeting the police with situational threats will have a higher likelihood of severe police action when compared to protests without situational threats. Protest policing scholars traditionally examine the characteristics of LEAs alongside the characteristics of a protest event. Militarization is a common LEA characteristic that scholars have examined to predict police repression, conceptualized as the process of LEAs adopting military gear and tactics and utilizing them on civilians. The militarization of police agencies has been positively correlated with an increased likelihood of police deploying less-lethal and lethal weapons at protest events (Myers-Montgomery 2016; Soule and Davenport 2009). Soule and Davenport (2009) analyzed militarization through the presence of paramilitary units, while Myers-Montgomery (2016) pointed to LEA participation in the Department of Defense (DoD) 1033 Excess Use Program (1033 Program) as a measurement of militarization. The 1033 Program was congressionally authorized in the early 1990s and allowed police agencies to apply for and acquire retired military hardware (Defense Logistics Agency N.d.). The equipment provided through the DoD 1033 Program include protective equipment, tanks, various forms of less-lethal weaponry, and automatic weapons (Myers-Montgomery 2016). An additional agency-specific characteristic that has been operationalized by protest policing scholars is the number of active officers in an LEA, which is inversely correlated with the likelihood of repression (Nassauer 2014). Understaffed agencies may perceive themselves as incapable of adequately responding to an event leading them to utilize more aggressive responses at protests. 14 Scholars have established that police-caused deaths are strong predictors of social movement activity (Williamson, Trump, and Einstein 2018), however, its utilization to explain police repression at protests is underwhelming in protest policing literature. Traditionally, literature has utilized prior brutality allegations by LEA to explain the likelihood of repression; however, it was not significant in models measuring forms of intervention (Earl 2003; Earl, Soule, and McCarthy 2003). Although police-caused deaths are a significant predictor of protest activity, the association between such police actions has not yielded significant results when examining police repression at protest events. Agency specific policing characteristics are relevant contextual factors to account for when examining protest policing. However, the reliability of such measures predicting the likelihood of police repression at protest events is questionable. Therefore, the presence of reputational threats at all protests targeting the police paired with the presence of situational threats, or lack thereof, will be stronger predictors of police actions at protests compare to LEA characteristics. 15 METHODOLOGY To examine protest policing approaches during the 2020 wave of demonstrations, the Armed Conflict Location & Event Data Project1 (ACLED) dataset was utilized. The ACLED collects information on all instances of political violence, strategic developments, and demonstration events in the United States and abroad (ACLED 2021b). Political violence, strategic developments, and demonstrations events capture various forms of protests across the spectrum of violence. At a peaceful protest, demonstrators remained predominantly non-violent, although, violence may have been directed at them by other protest actors. Violent demonstrations, which may have begun as peaceful, resulted in disruptive behavior such as looting, protester clashes with police, vandalism or property damage, physical altercations, etc. To build the ACLED, researchers reviewed a collection of over 2,800 sources ranging from international media outlets to ‘new media’ – online social media platforms which undergo a verification process to confirm the legitimacy of the source and its information (ACLED N.d.a). ACLED news sources span a variety of geographic regions and contain viewpoints across the political spectrum. By drawing from an assortment of outlets with contrasting viewpoints and characteristics, researchers were able to decrease the risk of political bias (ACLED N.a.b). Information found to be incorrect or that developed over time is amended in the ACLED weekly. The location, date, fatality count, or the actors involved in a protest are the most common forms of data corrected in the ACLED (ACLED N.d.b). Utilizing news media as a source to collect protest data is not a new research method in social movement literature. Earl, Soule, and McCarthy (2003) collected data from the New York Times (NYT), to examine police responses to protests in New York. Reynolds-Stenson (2018) 1 Accessed September 3, 2021 used the Dynamics of Collective Action (DOCA) dataset to analyze policing behavior at anti- police brutality protests, DOCA collects information from daily editions of the NYT to record hard news items about protest events. Soule and Davenport (2009) also utilized DOCA to investigate forms of protest policing between 1960 and 1990. Although a common research method, secondary data analyses utilizing news data are susceptible to selection and description bias (Earl, Soule, McCarthy 2003; Reynolds-Stenson 2018; Soule and Davenport 2009). Selection bias refers to the underreporting of actions and the social actors preset at a protest event, and the events that are noted represent an oversaturated and unreliable sample of excessively violent events or those sponsored by prevalent social movement organizations (SMOs). Description bias concerns the validity of information, it is caused by inaccurate reporting on the actions of protesters and the sequence of events at a demonstration to make it more ‘newsworthy’. Common methods to overcome description biases involve focusing on the “hard facts” or “hard news items” in news reports (Reynolds-Stenson 2018:53; Soule and Davenport 2009:9). Hard facts ensure accuracy and validity through the identification of irrefutable protest information. The hard facts of a protest include the who (organizations, police, and individuals present), what (the kind of protest event and tactics used), when (date), where (geographic location), and why (goals articulated) of an event (Reynolds- Stenson 2018; Soule and Davenport 2009). The ACLED recorded the date, location, involved actors, and the characteristics (goals of the protest, behaviors of the protest actors, and the sequence of events) of the event. Although the dataset documented those present and their behaviors at the protest, the data does not specifically denote police presence, likely due to the description bias of the reporting media source. Police presence was only documented in the ACLED if police intervened in a protest. 17 Therefore, this analysis of protest policing will focus on policing interventions rather than traditional parings of presence and intervention (Earl, Soule, and McCarthy 2003; Reynolds- Stenson 2018; Soule and Davenport 2009). Additionally, the ACLED recorded the size of a protest for a nominal number of events, rendering its inclusion in the analysis inopportune. Although protest size is a traditional form of threat operationalized in protest policing literature, the exclusion of this variable is not detrimental to the analysis as previous scholars have indicated that news records of demonstration size can tamper with the validity of data due to the increased possibility of description bias (Reynolds-Stenson 2018; Soule and Davenport 2009). The ACLED coding methodology for protest events relied on the date, location, and characteristics of a protest. If multiple events with the same behaviors and goals occurred in the same location on the same day – simultaneously or staggered – they were coded as a single event in the data. For example, protesters in Kings, New York organized 11 peaceful protests on the same day across the city, all were in support of BLM, against police brutality, and in solidarity with the death of George Floyd; all protests were grouped into a single case. Contrastingly, protests of similar nature that occurred in the same geographic vicinity and day with differing goals were coded as separate events. On June 12, 2020, two peaceful protests occurred in Riverside, California, one protest supported Blue Lives Matter2 and the other was in support of BLM, each of these protests was coded as a separate event. Additionally, if multiple protests with varying degrees of violent and peaceful protester activity occurred on the same day and geographic location they were coded separately. For instance, a peaceful protest in the morning and a violent demonstration in the evening were coded as distinct events. Therefore, the ACLED 2 Blue Lives Matter is a social movement characterized by its support of law enforcement that began in response to the BLM movement and anti-police rhetoric. 18 coded protest events dependent on the date, geographic location, and categorization of protest behavior – whether it was mostly peaceful or violent – and goals. 19 Sample To examine policing interventions at protests targeting the police during June 2020, all protests in support of BLM, with anti-police brutality goals, over the death of George Floyd or other black men and women killed by law enforcement, and other goals challenging LEAs as an institution in the ACLED were analyzed. Therefore, all protest events included in the sample had the reputational threat of targeting the legitimacy of police. June marked the height of demonstration activity during 2020, a total of 6,816 protests occurring over the course of thirty days and 5,993 targeted the police with their goas. (See Figure 2 for a distribution of events by day in June). The high volume of protest activity can be traced to the murder of George Floyd by police on 25 May 2020, with the frequency of protest events increasing by over 300 percent in June compared to the previous month. The drastic increase of protests paired with demonstrations targeting law enforcement posed numerous challenges for protest policing during this period. Protest events that did not have goals that targeted police were dropped from the sample (n=796). Additionally, events that lacked adequate details regarding the actors involved, description of behaviors, or protest goals were excluded (n=24). Alaska, Hawaii, and the District of Columbia did not have any 1033 Program data, leading to their exclusion from the analysis, a total of 93 events occurred in these geographic locations over the month of June 2020. The remaining protests created a total sample of 5,900 events targeting law enforcement. The presence of situational threats varied across events; with a majority not having any present. 20 Coding Methodology Building the ACLED data so that it was applicable to the foci of this study required new variables to be constructed from the original data. Using the Notes3 variables in the ACLED, the actions exhibited by police during a protest were used to construct the outcome variable focused on police actions at protests. Likewise, the same text description of the protest event in the original Notes variable were examined to identify the characteristics of the protest event to code protester behaviors that represent situational threats, including the presence of counter- demonstrators and/or lone wolves. The violent protest events were first examined due to the diversity of police actions and protester behaviors and goals. After the themes of policing interventions and situational threats emerged, the entire sample was coded. Following the initial coding phase, four more reviews of the dataset were conducted to ensure consistency in the coding of variables. Other datasets were merged using a many-to-one merge on the geographic location variables (county and state or FIPS codes). Many-to-one merges were used because the LEA characteristics and control variables could correspond to many observations in the master ACLED datafile, but uniquely identified observations in the original dataset. Once the variables were merged, some variables were rescaled to produce meaningful relative risk ratios for the multinomial logistic regression analysis, if the variables were not rescaled the risk ratios were nominal (=1.00). 3 The Notes variable in the ACLED dataset contains all the hard facts of a demonstration (the goals, behaviors of protest actors, and the sequence of events over the course of a protest. 21 Dependent Variable To create the dependent variable, Earl, Soule, and McCarthy’s (2003) operationalization of police responses was used as a framework to analyze forms of police intervention. Three out of the five categories of police responses were present in the ACLED: “Ounce of Prevention/Legal Eagles,” “Dirty Harry,” and “Calling All Cars” (Earl, Soule, and McCarthy 2003:590). However, adjustments were made to the police intervention categories to fit the logical groupings of the recorded police responses. The forms of police action at June 2020 protest events operationalized for this study are as follows. First, was no intervention or the “Nothing Newsworthy” approach (n=5,556). The ACLED did not record police presence unless law enforcement took action, therefore, this variable reflects protests with no mention of police action. However, a nominal amount of crowd management cases (barricades, traffic direction, or conducting surveillance) are also represented in this policing approach. Unfortunately, crowd management policing actions were too scarce to be independently categorized (n=30). The oversaturation of events with no reported police presence and underrepresentation of crowd management is likely due to the description bias of the reporting media outlets. Any protests events with no police action or limited police action, like separating protesters and counterprotesters or erecting barricades, were coded as a 0. An example of “Nothing Newsworthy” is, “on 6 June 2020, people marched in Dallas (Texas) in support of the Black Lives Matter movement and against police brutality” or “on 4 June 2020, around 125 people and students supporting Black Lives Matter marched around Elk Grove Village Hall in Elk Grove Village (Cook, Illinois) against police brutality. 9 counter-protesters were present at the scene. Police arrived and separated the two groups.” 22 The second policing approach was “Legal Eagles,” which captures instances of police making at least one arrest at a protest event (excluding the use of force and/or violence). Any protest where police arrested a protest actor (activists, counterdemonstrators, lone wolves, journalists, and/or passersby), with no use of force present, was coded as a 1 (n=179), they were the most common form of police action other than “Nothing Newsworthy.” For example, “on 4 June 2020, hundreds of people marched in Sacramento (Sacramento, California) in support of the Black Lives Matter movement and against police brutality and the killing of George Floyd. One person was detained by California Highway Patrol after he was found with at least one gun bag and was expressing opposition to the demonstration and support for the Boogaloo Bois movement.” “Dirty Harry” was the third protest policing approach, any events where officers used force and/or violence against protest actors, without any arrests, to contain an event or regain control fell under this approach. The Dirty Harry category is modeled from Earl, Soule, and McCarthy (2003), but the conceptualization of this variable also includes the frameworks of Earl’s (2003) description of coercive repression along with Soule and Davenport’s (2009) use of force and/or violence variable. Any event where police deployed less-lethal or lethal weapons (See Appendix A) and/or coercive repression (intimidation or physical assault) with individuals at the protest was coded as Dirty Harry (n=73). For example, “on 15 June 2020, a crowd of 200- 300 people staged a march outside police headquarters in Richmond (Virginia) in support of the Black Lives Matter movement. At least two city councilors also attended. Several demonstrators were armed with knives and one had a sword. Police forces targeted demonstrators with pepper spray, flash-bang grenades, a chemical irritant, and rubber bullets to disperse the demonstration.” 23 “Calling All Cars” was the final police response operationalized in this study, this approach involved a combination of arrests and use of force and/or violence. This was the third most common form of police response (n=92). An example of the “Calling All Cars” policing approach is, “on 22 June 2020, in an overnight demonstration, more than 150 people staged a march in Richmond (Virginia) in support of the Black Lives Matter movement and against police brutality and the death of George Floyd. Around midnight, police officers arrived at the demonstration. People threw objects and rocks at the police, blocked off the street, and set up tents, although the time of this event is unknown. Police dispersed the crowd with tear gas, rubber bullets, and pepper balls. A journalist was sprayed with tear gas by police. 12 people were arrested.” 24 Independent Variables The situational threats measured in this study are event-level characteristics of the protest events, which included the presence of confrontational tactics, protester clashes with the police, and independent measures for counter-protester or lone-wolf presence. The variation in the occurrence of situational threats at protest events targeting the police are the main independent measures of the analysis. Although all events in the sample challenged the legitimacy of police, a majority did not have situational threats present. Confrontational tactics were conceptualized as protester tactics and behaviors that present situational threats to law enforcement’s ability to maintain social control. In this analysis, confrontational tactics were operationalized as a dummy variable (n=574). Any events with instances of disruptive or disorderly conduct directed at fellow protesters, counter-protesters, lone-wolves, passersby, journalists, or public and private property were coded as a 1 (See Appendix B). Volatile protester behaviors directed at law enforcement officers were separately coded. I operationalized a measure of protester-police clashes to examine whether police interventions are more likely when protesters directly engage with officers violently or aggressively. A protester-police clash was conceptualized as an aggressive or confrontational engagement between protesters and police, that was initiated by protesters (See Appendix C). Volatile protester interactions with the police occurred at 141 events. Although substantially less frequent than confrontational tactics, these interactions showcase a particular form of situational threat directed specifically at the police. The behaviors exerted by counter-protesters and lone-wolfs at protests were similar, with both groups either being present or engaging protesters through verbal and physical altercation (See Appendix D). The counter-protester and lone-wolf variables were independently coded as 25 (0) not present or (1) present and/or engaging volatilely with demonstrators, police, or other protest actors. In June 2020, counter-protesters were present at 210 protest events and lone wolves were present at substantially fewer demonstrations (n=128). Lone wolves are a unique measure for this analysis, although conceptualized in previous literature, the operationalization and statistical analysis of these protest actors has yet to be conducted. The ACLED does not collect information on LEA characteristics, therefore the characteristics of police agencies were merged with the ACLED data from other sources4. Combining Earl, Soule, and Davenport’s (2003) variable which measured police resources through per-capita county expenditure on LEAs and Myers-Montgomery’s (2016) emphasis of 1033 Program participation leading to militarization, the variable measuring the total value of 1033 equipment aggregated by county5 was operationalized with all values rescaled by $10,000 (mean: 2,525,000.00). The total number of law enforcement officers by county was also included in the analysis and was rescaled by 1,000 officers (mean: 598). The final police measure was the total number of police killings from 2013 to 24 May 2020 (mean: 20.31). 4 Police characteristics were merged from Patrick Gillham’s “occupy_blm_data”. Received April 1, 2022. 5 The total dollar amounts are from May 1990 to September 2020. 26 Control Variables All control variables, excluding federal force involvement and executive action, were merged from 2020 U.S. Census data at the county level. The federal force involvement and executive action variables were operationalized using ACLED data. Protests where the presence of federal forces were explicitly mentioned or any events that occurred within the localities and dates when federal forces were deployed were coded as a 1 for the dummy variable. Any event that took place within the geographic bounds and dates of a curfew – at the local and/or state level – were coded as a 1 for a curfew being present. Protests did not have to occur during the “active hours” of a curfew to be coded as having a curfew present, for instance, any protest that occurred in Raleigh, High Point, or Greensboro North Carolina during the dates of 1 June through 8 June while curfews were instated would be coded as a 1. The control variables drawn from 2020 U.S. Census data include the county population – which was rescaled by 100,000 –, the percentage of black or African American residents, the percentage of Hispanic residents, the percentage of the population over 18 with a college degree, the median family income – rescaled by $1,000 –, and the median age. Figure 3 provides a flowchart explaining the relation to each variable in this study to the operationalization of different levels and forms of threat. Protests targeting the police occurred on a national level, emphasizing the necessity to understand what protest elements police perceive as threatening and the associated likelihood of repressive actions when threats arise. Blalock’s (1967) threat hypothesis examines policing practices in relation to community level characteristics, attributing a positive correlation between the likelihood of police repression and the percentage of minority residents in a community. For this analysis, the county-level demographic characteristics will control for elements of Blalock’s threat hypothesis. 27 Reputational threats, operationalized as protest goals targeting the police, function on the institutional level by such goals challenging the legitimacy of police as an institution. All protest events examined in this analysis have the reputational threat of goals targeting the police present. Situational threats are event-specific elements of protests, based on the social actors present and the behaviors they exert, like confrontational tactics, protester clashes with the police, and the presence of counter-demonstrators and lone-wolves. Within the protest events being examined, the presence of situational threats was heterogenous, allowing for an investigation of police interventions at protest events targeting the police based on the presence of situational threats. 28 Statistical Analysis This study relied on publicly available data that did not contain identifiable information, therefore, did not require Institutional Review Board review. The data was obtained from the ACLED, U.S. Census, and Patrick Gillham’s “Occupy BLM Data” dataset6. Summary statistics of all study variables were gathered using univariate analyses. Bivariate comparisons were conducted across each form of police intervention on all independent and control variables. Pearson chi-square tests were performed on dichotomous variables with counts and percentages. Kruskal-Wallis tests were used for all continuous variables7, displayed as medians and interquartile ranges (IQR), the IQR reflect the 25th and 75th percentile medians. The statistical procedure for each variable in the bivariate analysis is noted with a unique symbol in Table 2. A multinomial logistic regression model, with Nothing Newsworthy set as the base outcome, was performed with relative risk ratios (RRRs) expressing the association between each protest and police characteristic and the likelihood of being in the one category of the outcome variable (Legal Eagles, Dirty Harry, or Calling All Cars) rather than the base. Statistical significance was defined as any P value less than 0.05. RRRs express the likelihood of being in the one category of a variable rather than the base, an RRR >1.00 reflects an increased likelihood of being in the one category, while an RRR <1.00 indicates a decreased likelihood of being in the one category. The equation to translate RRRs to a percentage of increased or decreased likelihood is: RRR – 1.00 * 100 = RRR% 6 Proper citation will be updated pending citation guidelines from Patrick Gillham. 7 The Kruskal-Wallis tests were selected for continuous variables due to the large standard deviations produced when Pearson chi-square tests were used. Using Kruskal-Wallis tests, the differences in medians and IQRs provided a more reliable measure of distribution. 29 Therefore, an RRR =1.00 means there is no difference in likelihood of being in the one category compared to the base (i.e. RRR=1.00 translates to 0%). Example interpretations of RRRs in the multinomial logistic regression are provided in the results. Analyses were performed using Stata17.0/BE (StataCorp, LP, College Station, TX). 30 RESULTS Table 1 provides the univariate descriptive statistics for all study variables; dichotomous variables have the frequency and frequency percentage recorded while continuous variables have the median and IQR. Nothing Newsworthy was the most common policing response to protest events (n=5,556). Legal Eagles was the second most common form of protest policing intervention (n=179), force-reliant interventions were the last common (Calling All Cars n=92; Dirty Harry n=73). Among the forms of situational threats at protest events, confrontational tactics had the highest frequency (n=574), followed by counter-protesters (n=210), protester clashes with the police (n=141), and lone wolves had the lowest frequency (n=128). The rarity of situational threats at protests challenging the police may impact the RRRs produced in the multinomial logistic regression. However, this should not lead readers to discount the results that have been provided. Rather, they emphasize the unique association between situational threats and police repression and discount egregious depictions of violent behaviors being a norm at protest events targeting the police. Additional univariate statistics include LEA characteristics at the county-level. The median total value of 1033 Program equipment was $1,230,000 (IQR: $281,300, $34,211,000) and counties had a median of 1,600 law enforcement officers (IQR: 500, 4700). The total number of citizens killed by police from 2013 – 24 May 2020 had a median of 5 (IQR: 1, 15). Federal agencies were involved at 22 protest events (0.37%) and curfews were present for 96 protests (1.63%). Sociodemographic control variables included the county population, which had a median of 3,920,000 (IQR: 1,190,000, 9,570,000), the percentage of black or African American residents (median: 8.02; IQR: 3.16, 17.73), and the percentage of Hispanic residents (median 10.68; IQR: 5.83; 21.82). The percentage of the county population over the age of 18 with a college degree had a median of 45.86 (IQR: 34.77, 56.69) and the median age of the counties was 38.30 (IQR: 36.1, 41). The median family income was $803,200 (IQR: $695,500, $956,200). The median family income of counties in the sample was higher than the national average (Shrider et al. 2021). Situational threats and LEA characteristics are summarized by type of police response in Table 2. Police were most likely to use Calling All Cars (50%) interventions at protests when protesters engaged in confrontational tactics, followed by Legal Eagles (42%), Dirty Harry (36%), and Nothing Newsworthy (8%). Demonstrations where protesters clashed with the police were similarly met with an increased likelihood of repressive police action (only 1% of events fell under with Nothing Newsworthy, while 46% resulted in Calling All Cars). Counter- protesters and lone-wolf presence at a protest were more likely to be met with Legal Eagles interventions (counter-protesters 10%; lone-wolf 21%). Police were least likely to do nothing at protests with a lone-wolf presence (1%). Interestingly, police did not use Dirty Harry interventions at any event where counter-protesters were present. Confrontational tactics and protester clashes with the police at demonstrations targeting the police elicited the highest likelihood of severe police action (Calling All Cars), while counter-protesters and lone wolves were most likely met with Legal Eagles police actions. TABLE 2 ABOUT HERE The LEA characteristics were measured at the county-level. The total value of 1033 Program equipment was significantly associated with the type of police response at a protest, with force-reliant interventions, like Dirty Harry (median $1,960,000; IQR: 550,000, 4,950,000) and Calling All Cars (median $1,790,000; IQR: $490,000, $3,660,000), occurring in counties 32 with higher total values of 1033 Program equipment. In counties where the value of 1033 Program equipment was lower, Legal Eagles (median 1,400,000; IQR: $480,000, $3,680,000) and Nothing Newsworthy (median $1,200,000; IQR: $280,000, $3,420,000) were the primary forms of intervention. The total number of law enforcement officers followed the same distribution among policing interventions as 1033 Program equipment value; counties with larger amounts of law enforcement officers were dramatically more likely to intervene than do nothing. Protests in counties with the highest amount of law enforcement officers were the most likely to be met with the Dirty Harry policing intervention (median 311; IQR: 133, 647). The median number of law enforcement officers only had a differential of 10 officers between Calling All Cars (median 232; IQR: 105, 527) and Legal Eagles (median 243; IQR: 85, 788) policing approaches. Nothing Newsworthy was the most common protest policing action in counties with the lowest amount of law enforcement officers (median 156; IQR: 53, 445). In counties with the highest total number of citizens killed by law enforcement officers, protests were most likely to be met with Dirty Harry police interventions (median 8; IQR: 2, 23). The second highest median of police caused deaths was associated with Calling All Cars policing actions at protests (median 6.5; IQR: 2, 29). Police were more likely to do Nothing Newsworthy (IQR: 1, 14) or utilize Legal Eagles approaches (IQR: 2, 20) in counties with a median of 5 police-caused deaths. All demographic characteristics, excluding federal force intervention and executive action, were measured at the county-level. Police were more likely to use Calling All Cars interventions when federal agencies were present (7% when compared to 1% among Nothing Newsworthy). Police responses followed the same trend when curfews were present (Calling All Cars 16%, compared to 1% with Nothing Newsworthy), however, the second most common response was Legal Eagles (12%) followed by Dirty Harry (11%). 33 Protesters were more likely to be met with Dirty Harry (IQR: 390,000, 1,300,000) and Calling All Cars (IQR: 410,000,1,400,000) policing interventions in counties with larger populations, each had a median of 810,000 for the county population. Nothing Newsworthy was the most common intervention in counties with the lowest population (median 380,000; IQR: 110,000, 950,000). Counties with elevated percentages of black or African American residents were not significantly associated with protest policing responses (P=0.378). Although, the population percentages of black of African American residents was higher in counties where protests were met with Legal Eagles (median 10; IQR: 5, 22), Dirty Harry (median 13; IQR: 7, 26), and Calling All Cars (median 11; IQR: 7, 23) policing interventions. The percentage of Hispanic residents in a county was not significantly associated with type of police interventions at a protest event; however, the percentage of Hispanic residents was higher in counties where police utilized Legal Eagles (IQR: 7, 24) and Calling All Cars (8, 27) protest policing interventions, with each having a median of 13. Protests were met with Dirty Harry policing interventions in counties with the highest The percentage of college graduates was higher in counties were police utilized repressive interventions. Protests met with Dirty Harry interventions occurred in counties with the highest percentage of college graduate (median 52; IQR: 45, 64), followed by Calling All Cars (IAR: 42, 60) and Legal Eagles (39, 63); both had a median of 48. Protests met with Nothing Newsworthy (IQR: 69, 96), Legal Eagles (IQR: 71, 92), and Dirty Harry (IQR: 71, 90) policing interventions occurred in counties with higher median family incomes; all had a median of 80. The “Calling All Cars” intervention, had the lowest family income with a median of 77 (IQR: 72, 88). The final control variable, median age, showed protests were met with force-reliant interventions, Dirty Harry (IQR: 35, 38) and Calling All Cars (IQR: 35, 38 in counties with younger populations, both had a median of 37. 34 Table 3 displays the results of the multinomial logistic regression8 of forms of police intervention by situational threat and LEA characteristics, along with the control variables of this analysis. The significance of situational threats eliciting repressive police responses is consistent with the hypothesis that protests targeting the police with such threats will have a higher likelihood of severe police action. Furthermore, law enforcement officers are continuously more likely to exude more repressive actions when protesters clashed with the police. Although generally, police were more likely to do nothing at protests events (n=4,866), when situational threats were present police were more likely to repress and use force against protsters. The evidence of confrontational tactics, protester clashes with the police, and lone-wolf presence each being associated with an increased likelihood of Legal Eagles, Dirty Harry, and Calling All Cars interventions (P<0.001) confirms the hypothesis that situational threats elicit more repressive police action at protest events targeting the police. Although both protester clashes with the police and confrontational tactics increased the likelihood of repressive police actions, clashes with the police were consistently greater predictors of intervention. Interestingly, lone-wolf presence consistently elicited more repressive police action compared to counter- protester presence. The situational threat of use of confrontational protest tactics at a protest events targeting the police was associated with the police being 5.03 times more likely to use arrests (Legal Eagles), 3.76 time more likely to use force (Dirty Harry), and 5.55 times more likely to use both arrests and force (Calling All Cars). Protester clashes with the police produced exponentially higher RRRs across each form of police intervention compared to other situational threats. 8 A multinomial logistic regression with county random-effects was attempted on the data. However, collinearity between the protester clashes with police and confrontational tactics variables prevented the models from converging. When either protester-initiated clashes or confrontational tactics were removed from the model, the results produced the same level of significance and direction of association. 35 Compared to events where no protester clashes with the police occurred, those that did have clashes with the police were about 27 times more likely to elicit a Legal Eagles response from police, 52 times more likely for Dirty Harry, and 91 times more likely to have a Calling All Cars intervention. Counter-protesters were significant predictors (P=0.002) of Legal Eagles policing intervention, increasing the likelihood of such policing actions by almost 3 times. The situational threat of lone-wolf presence increased the likelihood of Legal Eagles interventions by 22 times, Dirty Harry by 6.76 times, and Calling All Cars by 11.33 times compared to events where these protest actors were not present. The only statistically significant policing characteristics in this analysis was the total number of law enforcement officers by county (P<0.001) and increased the likelihood of police using Legal Eagles interventions at protests by 7%. Interestingly, neither the total DoD 1033 Program equipment values nor total citizens killed by police were associated with an increased likelihood of force-reliant policing approaches. The involvement of federal agencies at a protest event significantly increased the likelihood of police using Legal Eagles by almost 5 times and Calling All Cars by 10 times (P<0.001). The control variable executive action, which measured the presence of curfews, was statistically significant (P<0.001) across all policing interventions. The presence of a curfew increased the likelihood of repressive police interventions (Legal Eagles, Dirty Harry, and Calling All Cars) compared to events outside the bounds of a curfew. Curfews were associated with an increased likelihood of utilizing Legal Eagles interventions by 16 times, Dirty harry by 10 times, and Calling All Cars by 15 times. Additionally, the sociodemographic situational threat variables which measured the percent of the county population with a college degree, median family income, and median age 36 were statistically significant predictors of police intervention. As the percentage of college graduates increased in a county, police were 4% more likely to use force-reliant policing interventions, such as Dirty Harry (P=0.014) and Calling All Cars (P=0.030). Higher median family incomes decreased the likelihood of Calling All Cars police interventions by 4% (P=0.010). Similarly, the likelihood of Legal Eagles as a form of policing intervention decreases by 2% (P<0.001) at the median county age increases. The county population, percentage of black or African American, and percentage of Hispanic residents were not statistically significant indicators of any form of police intervention. The results of this analysis illustrate the significant role that situational threats play in catalyzing police repression at protest events. Traditional situational threats, like confrontational tactics, counter-protester presence, and lone-wolf presence were significantly associated with an increased likelihood of Legal Eagles policing interventions at protests. Additionally, confrontational tactics and lone-wolf presence elicited force-reliant interventions, such as Dirty Harry and Calling All Cars. The new situational threat presented in this research, protester clashes with the police, was significantly associated with an increased likelihood of police repression and use of force at protests. Although all but one of the LEA characteristics were not statistically significant, the total number of law enforcement officers was associated with an increased likelihood of Legal Eagles policing approaches at protest events. The presence of a curfew in the geographic location and during the period of a protest event elevated the likelihood of police repression and use of force at a protest. The involvement of federal agencies at a protest was a significant predictor of Calling All Cars policing interventions at a protest event. 37 DISCUSSION The 2020 wave of protests targeting the police marks another chapter of Americans using their right to protest to demand accountability and reform in law enforcement institutions. Thousands of protests occurred across the United States despite a the COVID-19 global pandemic which posed a significant health risk. Three days after George Floyd’s death, 28 May 2020, the United States reported a total of 100,000 COVID-19-related deaths (CDC 2022). Although stay-at-home practices were increasing among Americans, the nation experienced the largest upheaval in protest activity since the mid 20th century (Dave et al. 2021). Therefore, should this protest wave of demonstrations targeting the police have occurred outside the context of COVID-19, we could expect an increase in frequency and participation of protests targeting the police. While the effects of COVID-19 on social movement activity have yet to be completely understood, the widespread adoption of protest goals targeting the police presented a national threat to police as an institution. Although similar to the wave of protests that followed the death of Michael Brown in 20149, the geographic expanse and frequency of protest events stemming from the death of George Floyd make the June 2020 sample of protest unique. More importantly, traditionally localized protest goals that targeted the police transcend to a national level, with at least 15 protests occurring in each state (See Figure 1) and a heigh of 697 protests occurring on the same day (See Figure 2). The national adoption of such protest goals signaled decreased public support for police as an institution, leading these protests to represent a widespread source of threat to the legitimacy of police (Lasley 1992; Trinker, Kerrison, Goff 2019). Additionally, 9 Michael Brown was an unarmed black or African American teenager who was fatally shot by police in Ferguson, Missouri in 2014. Following his death and the failure of the grand jury to indite the officer alleged to be responsible, protests against police brutality and led by or in support of BLM occurred in Ferguson and spread across the nation (Reynolds-Stenson 2018). with law enforcement officers representing and/or being the source of grievance at protests targeting the police, their presence alone could be a form of threat to protesters and possibly triggering the usage of confrontational tactics or protester clashes with the police (Waddington 2006). The protest policing practices at protest targeting the police, where situational threats were present, triggers a cycle of disorder. The source of the cycle are events of police brutality, after police engage in violence that was deemed unnecessary or excessive by the public, protests targeting the police occur. Protest goals against police brutality increase the likelihood of police presence (Reynolds-Stenson 2018), and once police are present both demonstrators and officers are threatened by the opposing protest actors, once protesters engage in civil disobedience police deploy repressive and force-reliant interventions to regain social control, once again initiating the cycle for the initial cause of the protest event. Figure 4 illustrates the cycle of disorder that stems from the presence of situational threats at protest events targeting the police. The multiracial national alliance that has developed around protest goals targeting the police has possibly signaled a mainstream challenge to the police, which compounded the protest policing paradox (Reynolds-Stenson 2018). At protests targeting the police, law enforcement officers must tread a careful line of social control, where not enough control leads to the delegitimization of policing as a profession and excessive control increases public scrutiny. However, situational threats increase public support of police action at such events, which may alleviate law enforcement’s perceived risk for potential criticism if they take repressive action. This analysis explored the dynamic between police action and the presence of situational threats at protests targeting the police and found that police are more likely to use repressive interventions (Legal Eagles, Dirty Harry, and Calling All Cars) when situational threats were present at protests targeting them. 39 This research builds on the threat hypothesis of protest policing, illustrating that, although protests targeting the police, and particularly those that may traditionally trigger the perception of racial threat - like BLM - represent reputational threats to the police and increase the likelihood of presence and intervention, situational threats are stronger predictors of law enforcement repression at such events (Reynolds-Stenson 2018). I demonstrated that situational threats, such as confrontational tactics and protester clashes with the police, and lone-wolf presence are continuously associated with an increased likelihood of arrests, use of force, and a combination of both interventions at protests. Additionally, this research offers a continuation of the ‘blue approach’ by examining protester-initiated clashes with the police, this form of situational threat compounds the existing reputational threat at protests targeting the police by protesters directly challenging the authority and legitimacy of police while also representing a source of violence that threatens social order. This finding draws attention to a rare, yet significant, form of situational threat that elicits repressive police action. Furthermore, this research begins where Reynolds-Stenson (2018) left off by specifically analyzing protests that target the police. Where Reynolds-Stenson (2018) investigated the difference in policing interventions at anti-police brutality protests compared to other revolutionary goals, the scope of this research aims to take a deeper look into police responses to situational threats at protests events that target them. In Reynolds-Stenson (2018) analysis, reputational threats were the primary independent variable and situational threats were secondary measures. Reynolds-Stenson (2018) operationalized confrontational tactics as a distinct variable from the presence of property damage and violence by protesters. The operationalization of confrontational tactics in this analysis condensed those present in Reynolds-Stenson’s (2018) work, following the models of Earl, Soule, and McCarthy (2003) and Davenport, Soule, and 40 Armstrong (2011); with a single dummy variable for confrontational tactics. Additionally, the situational threats counter-demonstrators, lone wolves, and protester clashes with the police were not accounted for in Reynolds-Stenson’s (2018) analysis. The types of police interventions at protest events were broadened from a dichotomous measure of arrests and/or use of force, by adopting the framework of Earl, Soule, and McCarthy (2003), which permits a more nuanced understanding of how police respond to protest events targeting them. Most importantly, this examination of protest policing is situated within contemporary social contexts, allowing for an analysis of protest policing interventions under the current policing strategy of strategic incapacitation, whereas Reynolds-Stenson (2018) investigated policing interventions under negotiated management; the predominant policing strategy of the 1970s through late 1990s. The most significant finding of this analysis is the exponential increase of more repressive law enforcement interventions at protests targeting the police where protesters clashed with the police. The relative risk of repressive police action with protester clashes is exponentially larger than that associated with confrontational tactics and lone wolves. Although the collinearity of confrontational threats and protester clashes with the police create a limitation when analyzing them as distinct situational threats, the findings of this study suggest these forms of protester behavior should be conceptualized as independent forms of situational threat that elicit police intervention. Additional limitations of this analysis stem from the exclusion of protest size and the racial composition of protest participants as variables, due to a lack of data. The crux of Blalock’s (1967) threat hypothesis is minority presence driving the desire for social control; however, protest policing literature has noted an array of other threats that influence the likelihood of police repression at protest events (Earl 2003; Earl 2011; Earl, Soule, and 41 McCarthy 2003; Davenport, Soule, and Armstrong 2011; Gillham 2011; Gillham and Marx 2018; Gillham, Edwards, and Noakes 2013; Myers-Montgomery 2016; Nassauer 2014; Reynolds-Stenson 2018; Soule and Davenport 2009). Additionally, contemporary protests targeting the police are closely aligned with goals of racial equity and equality; SMOs like BLM are at the intersection of both. Protests occurring in response to police violence are mainly concentrated around instances of police brutality, the inconsistency of such social movement activity and the self-organized nature of these grassroots movements make analyses of the racial composition and size of SMOs like BLM difficult (Rickford 2016). Moreover, the adoption of such goals within white communities in America has created a notable complexity for police when approaching protests that target the police (Barroso and Minkin 2020). The most notable limitation of this study is the exclusion of police presence as an outcome. This variable is traditionally examined alongside policing interventions, with scholars noting the decision to be present at a protest is distinct from those determining police action (Earl, Soule, and McCarthy 2003; Reynolds-Stenson 2018; Soule and Davenport 2009). Reynolds-Stenson (2018) found police presence at anti-police brutality protests is significantly more likely than that at protests with other revolutionary or non-revolutionary goals. Furthermore, during the period of Reynolds-Stenson’s (2018) study, protest policing was under the directive of negotiated management, a strategy that was characterized by impartiality and communication with protesters (Gillham 2011). The contemporary protest policing strategy of strategic incapacitation is characterized by preemptive policing tactics that specifically focus on the neutralization and removal of legitimate or potential threats (Gillham 2011; Gillham, Edwards, and Noakes 2013). Therefore, the reputational threat of protests targeting the police 42 may increase the likelihood of police presence at such events, before situational threats are accounted for. Although protests that challenge the police increase attention to police conduct, the actions of police at such protest events display the desire of police to control and incapacitate situational threats is stronger than to do nothing and mitigate the associated risk of criticism. The findings of this analysis provide evidence that police will consistently deploy repressive and force-reliant interventions to situational threats at protests targeting the police. Moreover, the significant association between protester clashes with the police eliciting more repressive police interventions illustrates that particular forms of protester behaviors dynamically impact the actions taken by police. Future research needs to be conducted on the actions of police at protests targeting them when situational threats are not present. The significant impact that situational threats have on police action may signal that law enforcement officers behave differently at protests targeting them when threats are not present. Additionally, social movement scholars should investigate how protesters respond to other demonstrators clashing with the police, and the impacts of such protester behaviors on public support for protests or SMOs. Furthermore, the increased presence of lone wolves at protest events requires immediate attention in social movement and protest policing literature. Lone wolves agitate both police and protesters, making them social actors that significantly influence police actions at protests. However, further attention is required to explain how protesters react to lone wolves, as well as creating a deeper understanding of where these protest actors are coming from and how they organize as individuals or in groups. 43 CONCLUSION This analysis of protest policing is unique because it solely examines protests that challenge the legitimacy of police by having goals in support of BLM, against police brutality, over the death of George Floyd or other black or African American men and women killed by law enforcement, or other ideologies calling for the removal, reallocation, or reformation of police. Additionally, the selected sample of protests occurred during a global pandemic, which potentially created an opportunity for some Americans to engage in protest activity, but COVID- 19 also presented a significant health risk that likely decreased participation levels in collective activities that required public settings (Dave et al. 2021). The extraordinary aspects of the protests examined in this analysis are the frequency and national distribution of protests targeting the police. The frequency of protest events during the month of June 2020 is almost equivalent to, if not larger than, past protest policing studies which aggregated 10 and 35 years of protest data (Earl, Soule, and McCarthy 2003; Reynolds-Stenson 2018). The findings of this research offer insight on how police respond to situational threats at protest events that target the police. The most significant result of this analysis is the continuous increased likelihood of police repression and use-of-force at protests when confrontational tactics, protester-initiated clashes with the police, and lone wolves were present. Moreover, protester clashes with the police were the strongest predictors of police intervention compared to all other forms of situational threat. Situational threats are significant predictors of police intervention at protest events. The form of intervention police chose to utilize was dependent on the type(s) of situational threat present. Therefore, not only do situational threats significantly predict an increased likelihood of police repression at protest events target the police, but different types of situational threat drive differing forms of protest policing interventions. REFERENCES ACLED. 2019. “Armed Conflict & Event Data Project (ACLED) Codebook.” Retrieved September 1, 2021 (https://acleddata.com/acleddatanew/wp- content/uploads/dlm_uploads/2019/04/ACLED_Codebook_2019FINAL_pbl.pdf). ACLED. 2021a. “ACLED Methodology and Coding Decisions around Political Violence and Demonstrations in the United States of America.” Retrieved November 20, 2021 (https://acleddata.com/acleddatanew/wp-content/uploads/dlm_uploads/2020/10/USA- Methodology-v6-September-2021.pdf). ACLED. 2021b. “FAQs: ACLED US Coverage.” Retrieved November 20, 2021 (https://acleddata.com/acleddatanew/wp-content/uploads/2021/11/ACLED_FAQs-US- Coverage_March-2021.pdf). ALCED. N.d.a “ACLED Methodology.” Retrieved November 20, 2021 (https://acleddata.com/acleddatanew/wp- content/uploads/dlm_uploads/2019/04/Methodology-Overview_FINAL.pdf). ACLED. N.d.b “FAQs: ACLED Sourcing Methodology.” Retrieved November 23, 2021 (https://acleddata.com/acleddatanew/wp- content/uploads/dlm_uploads/2020/02/FAQs_ACLED-Sourcing-Methodology.pdf). Barroso, Amanda and Rachel Minkin. 2020. Recent protest attendees are more racially and ethnically diverse, younger than Americans overall. Pew Research Center. Washington D.C.: Pew Research Center. Retrieved March 29, 2022 (https://www.pewresearch.org/fact-tank/2020/06/24/recent-protest-attendees-are-more- racially-and-ethnically-diverse-younger-than-americans-overall/). Blalock, H. M. (1967). Toward a theory of minority group relations. New York: John Wiley. Buchanan, Larry, Quoctrung Bui, and Jugal Patel K. 2020. “Black Lives Matter May Be the Largest Movement in the U.S. History.” New York Times, July 3. Retrieved March 29, 2022 (https://www.nytimes.com/interactive/2020/07/03/us/george-floyd-protests-crowd- size.html). Center for Disease Control (CDC). 2022. “COVID-19 Timeline.” Atlanta, GA: David J. Sencer CDC Museum. Retrieved April 20, 2022 https://www.cdc.gov/museum/timeline/covid19.html#:~:text=January%2020%2C%2020 20%20CDC,18%20in%20Washington%20state Correll, Joshua, Bernadette Park, Charles Judd M., and Bernd Wittenbrink. 2002. “The Police Officer’s Dilemma: Using Ethnicity to Disambiguate Potentially Threatening Individuals.” Journal of Personality and Social Psychology 83(^):1314-1329. Crank, John P. and Robert Langworthy. “An Institutional Perspective of Policing.” The Journal of Criminal Law and Criminology 83(2):338-363. Earl, Jennifer. 2003. “Tanks, Tear Gas, and Taxes: Toward a Theory of Movement Repression.” Sociological Theory 21(1):44-68. Earl, Jennifer. 2011. “Political Repression: Iron Fists, Velvet Gloves, and Diffuse Control.” Annual Review of Sociology 37:261-284. 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. Eitle, David, Stewart J. D'Alessio, and Lisa Stolzenberg. 2002. "Racial Threat and Social Control: A Test of the Political, Economic, and Threat of Black Crime Hypotheses." Social Forces 81(2):557-576. Dave, Dhaval M., Andrew I. Friedson, Kyutaro Matsuzawa, Joseph J. Sabia, and Samuel Safford. 2021. Black Lives Matter Protests and Risk Avoidance: The Case of Civil Unrest During a Pandemic. National Bureau of Economic Research. Cambridge, MA: National Bureau of Economic Research. Retrieved April 20, 2022 https://www.nber.org/system/files/working_papers/w27408/w27408.pdf Davenport, Christian, Sarah Soule A., and David Armstrong A. 2011. “Protesting While Black? The Differential Policing of American Activism, 1960 and 1990.” American Sociological Review 76(1):152-178. Defense Logistics Agency (DLA). N.d. “DLA Disposition Services.” Retrieved January 10, 2022 (https://www.dla.mil/DispositionServices/Offers/Reutilization/LawEnforcement/Program FAQs.aspx Gillham, Patrick F. 2011. “Securitizing America: Strategic incapacitation and the Policing of Protest Since the 11 September 2001 Terrorist Attacks.” Sociology Compass 5(7):636- 652. Gillham, Patrick F., Bob Edwards, and John A. Noakes. 2013. “Strategic incapacitation and the policing of Occupy Wall Street protests in New York City, 2011.” Policing and Society 23(1):81-102. Gillham, Patrick F. and Gary T. Marx. 2018. “Changes in the Policing of Civil Disorders Since the Kerner Report: The Police Response to Ferguson, August 2014, and Some Implications for the Twenty-First Century.” RSF: The Russell Sage Foundation Journal of the Social Sciences 4(6):122-143. Habermas, Jürgen. 1975. Legitimation crisis. London, Uk: Heinemann Education. Hamm, Mark and Ramon Spaaj. 2015. Lone Wolf Terrorism in America: Using Knowledge of Radicalization Pathways to Forge Prevention Strategies. Office of Justice Programs. Washington, DC: U.S. Department of Justice. Retrieved March 9, 2022 46 (https://www.ojp.gov/ncjrs/virtual-library/abstracts/lone-wolf-terrorism-america-using- knowledge-radicalization-pathways Harris, Anthony J. 2021. “The Civil Rights Movement. Million Man March. Black Lives Matter.” International Journal of Multiple Research Approaches 13(1):33-40. Hickman, Matthew J., Alex R. Piquero, and Joel H. Garner. 2008. “Toward a National Estimate of Police Use of Nonlethal Force.” CRIMINOLOGY & Public Policy 7(4):563-604. King, Mike and David Waddington. 2006. “Flashpoints Revisited: A Critical Application to the Policing of Anti-globalization Protests.” Policing and Society 15(3):255-282. Lasley, J. R. 1992. “The Impact of the Rodney King Incident on Citizens attitudes toward police.” Policing and Society 3(4):245-255. Lowery, Patrick G., John Burrow D., and Robert J. Kaminski. 2016. “A Multilevel Tests of the Racial Threat Hypothesis in One State’s Juvenile Court.” Crime & Delinquency 61(1):52- 87. Lyle, Perry and Ashraf Esmail M. 2016. “Sworn to Protect.” Race, Gender & Class 23(3-4):155- 185. Metcalfe, Christi and Justin Pickett T. 2021. “Public fear of protesters and support for protest policing: An experimental test of two theoretical models*.” Criminology 60:60-89. Myers-Montgomery, Jamar. 2016. “Militarized Police and Unpermitted Protest: Implementing Policy that Civilizes the Police.” Cultural Studies ? Critical Methodologies 16(3):278- 286. Nassauer, Anne. 2014. “Effective crowd policing: empirical insights on avoiding protest violence.” Policing: An International Journal of Police Strategies & Management 38(1):2-23. Novak, Kenneth J. and Mitchell Chamlin B. 2012. “Racial Threat, Suspicion, and Police Behavior: The Impact of Race and Place in Traffic Enforcement.” Crime & Delinquency 58(2):275-300. Miller, Kirk. 2009. “The Institutionalization of Racial Profiling Policy: An Examination of Antiprofiling Policy Adoption Among Large Law Enforcement Agencies.” Crime & Delinquency 59(1):32-58. Parker, K. F., Stults, B. J., Rice, S. K. (2005). Racial threat, concentrated disadvantage and social control: Considering the macro-level sources of variation in arrests. Criminology, 43, 1111-1131. 47 Reynolds-Stenson, Heidi. 2018. “Protesting the police: anti-police brutality claims as a predictor of police repression of protest.” Social Movement Studies 17(1):48-63. Rickford, Russell. 2016. “Black Lives Matter.” New Labor Forum 25(1):34-42. Shrider, Emily A., Melissa Kollar, Frances Chen, and Jessica Semega. 2021. Income and Poverty in the United States: 2020. United States Census Bureau. Prince George’s County, Maryland: U.S. Department of Commerce. Retrieved May 31, 2022 (https://www.census.gov/library/publications/2021/demo/p60-273.html). Smith, Brad W. 2003. “The Impact of Police Officer Diversity on Police-Caused Homicides.” The Policy Studies Journal 31(2):147-162. 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. Terrill, William and Eugene A. Paoline, III. 2015. “Citizen Complaints as Threats to Police Legitimacy: The Role of Officers’ Occupational Attitudes.” Journal of Contemporary Criminal Justice 31(2):192-211. Tinker, Rick, Erin M. Kerrison, and Phillip Atiba Goff. 2019. “The Force of Fear: Police Stereotype Threat, Self-Legitimacy, and Support for Excessive Force. Law and Human Behavior 43(5):421-435. U.S. Commission on Civil Rights. 2018. Police Use of Force: An examination of modern policing practices. Washington: DC: U.S. Commission on Civil Rights. Retrieved February 20, 2022 (https://www.usccr.gov/files/pubs/2018/11-15-Police-Force.pdf). Williamson, Vanessa, Kris-Stella Trump, and Katherine Levine Einstein. 2018. “Black Lives Matter: Evidence that Police-Caused Deaths Predict Protest Activity.” American Political Science Association 16(2):400-415. 48 APPENDIX A: “DIRTY HARRY” OF USE OF FORCE/VIOLENCE POLICE INTERVENTION Less-Lethal Weapons The deployment of the following was coded as less-lethal weapons usage: incendiary canisters; smoke, smoke canisters, smoke bombs, and smoke grenades; taser; pepper spray, mase, and tear gas; non-lethal munition, police projectiles, non-lethal projectiles: pepper balls, rubber bullets, stinger balls, foam-coated police projectile, mace pellets, bean bags, smoke rounds, and foam bullets; flash-bang devices, flashbangs, and flash-bang grenades; stun grenades; unknown chemical agents, a chemical irritant; sound device; baton Use of Force/Violence Police interventions described as the following were considered to be use of force or violence: shooting but not fatality injuring a protester; a protester being treated for a gunshot wound or being injured; beating, shoving, assault without less-lethal weapons possibly leading to injury; using a low flying helicopter to scare off demonstrators; using police dogs; driving a police car through a demonstration. APPENDIX B: CONFRONTATIONAL TACTICS Protester actions considered to be confrontational were: disruptive behavior: disorderly conduct – picketing, trespassing; threatening behavior: making verbal threats, being aggressive, physical altercation/assault; being armed – firearms, knives, swords, or hammers; throwing objects; setting off explosives; reports of shots fired; protesting outside of and throwing/setting off objects at private residences; rioting; looting; property destruction: arson, broken windows/doors, spray painting, vandalism, and damaging/toppling statues; criminal mischief; reports of violent behavior, outbreaks of violence; obstructing a roadway; occupying an area of space or declaring an autonomous zone; multiple protests in one day; “violence throughout the day;” attempting to get a hold of projectiles; protesting after curfew; protesting across state-lines. The types of protests considered to be confrontational were reports of a violent demonstration, sit-ins, silent-ins, and camp-ins; and multi-day protest APPENDIX C: PROTESTER CLASHES WITH THE POLICE Demonstrators exhibiting the following behaviors were categorized as volatile protester clashes with the police: “demonstrators clashed with police,” confronted the police, and expressed grievances with police; throwing objects at police: rocks, bottles, water bottles, glass and glass bottles, frozen bottles, milk jugs, bricks, cinder blocks, shoes, urine and feces, Molotov cocktails, a large steel/metal object, cans of food and full beverages, a gas canister, projectiles, a scooter, paint, debris; setting off fireworks/firecrackers/flares; firing paint balls; flashing lasers at/in the direction of police; protesting to or outside of police buildings; damaging police property (building or vehicles): spray painting, breaking the windows, setting on fire, or throwing objects at/striking; blocking the police station with barricades; assaulting, fighting/physical altercation, shoving, violently pushing back, pepper spraying, shooting, or threatening police officers with weapon: machetes, bats, explosives, and firearms; refusing or encouraging other to refuse police orders, crowd refusing to disperse; surrounding police officers; protester and police standoff; killing a police officer. APPENDIX D: COUNTER-PROTESTERS & LONE WOLVES Counter-Protesters Counter-protesters united by a cause or membership to politically motivated group (militias, extremist groups, etc.) being present or observing a protest; physical altercations: fighting, assaulting, or clashing with protesters; heckling; being armed; standing with police at protest event; shooting a demonstrator; and protesting the protest. Lone-Wolf Individuals or groups of people who are not affiliated with a politically motivated group nor protest cause being present, patrolling, observing, engaging with, or “protecting the community” from demonstrators at a protest. Accelerating or driving a vehicle into a protest; verbal and physical altercations; severely injuring an officer; throwing rocks at protesters; property damage; possessing, brandishing, threatening discharging, or shooting a firearm at protesters or a protest event. FIGURE 1: FREQUENCY OF JUNE 2020 PROTEST EVENTS TARGETING THE POLICE BY STATEa ALABAMA 82 ARIZONA 62 ARKANSAS 46 CALIFORNIA 640 COLORADO 97 CONNECTICUT 157 DELAWARE 25 FLORIDA 313 GEORGIA 152 IDAHO 29 ILLINOIS 241 INDIANA 112 IOWA 90 KANSAS 62 KENTUCKY 103 LOUISIANA 82 MAINE 51 MARYLAND 102 MASSACHUSETTS 195 MICHIGAN 227 MINNESOTA 68 MISSISSIPPI 34 MISSOURI 150 MONTANA 29 NEBRASKA 41 NEVADA 31 NEW HAMPSHIRE 33 NEW JERSEY 132 NEW MEXICO 61 NEW YORK 330 NORTH CAROLINA 209 NORTH DAKOTA 14 OHIO 245 OKLAHOMA 64 OREGON 161 PENNSYLVANIA 240 RHODE ISLAND 22 SOUTH CAROLINA 74 SOUTH DAKOTA 14 TENNESSEE 124 TEXAS 252 UTAH 47 VERMONT 39 VIRGINIA 204 WASHINGTON 200 WEST VIRGINIA 28 WISCONSIN 151 WYOMING 35 0 100 200 300 400 500 600 700 Frequency of Protests aAlaska, Hawaii, and the District of Columbia were excluded from the analysis due to lack of DoD 1033 Excess Use Program data. Geographic Location - State FIGURE 2: FREQUENCY OF PROTESTS EVENTS BY DAY OVER JUNE 2020 Frequency of Protest Events by Day, June 2020 0 100 200 300 400 500 600 700 800 Monday, June 1 462 Tuesday, June 2 503 Wednesday, June 3 469 Thursday, June 4 419 Friday, June 5 411 Saturday, June 6 697 Sunday, June 7 489 Monday, June 8 220 Tuesday, June 9 140 Wednesday, June 10 122 Thursday, June 11 166 Friday, June 12 183 Saturday, June 13 306 Sunday, June 14 201 Monday, June 15 56 Tuesday, June 16 64 Wednesday, June 17 58 Thursday, June 18 56 Friday, June 19 190 Saturday, June 20 141 Sunday, June 21 51 Monday, June 22 39 Tuesday, June 23 41 Wednesday, June 24 46 Thursday, June 25 53 Friday, June 26 64 Saturday, June 27 120 Sunday, June 28 73 Monday, June 29 31 Tuesday, June 30 29 Weekday (Monday – Friday) Weekends (Saturday and Sunday) Date of Protest events, 2020 FIGURE 3: FLOWCHART OF THREAT LEVELS OF OPERATION. National Scale Protest policing at protest events that target the police, like Black Lives Matter (BLM) protests. Community County-level characteristics perceived to be potentially threatening by the police: county population size, average income, percentage of Hispanic and black or African American residents. Institutional Reputational threats, or threats to police's legitimacy, via the goals of protests: BLM or protests in support of BLM, anti-police brutality, against or over the death of George Floyd or other black or African American men and women killed by law enforcement, defund and/or abolish the police, removing the police from schools, or calling for the removal of individual officers due to alleged brutality or racism. Event-Level/Interactional Situational threats that arise at the protest event: confrontational tactics, protester clashes with the police, and the presence of counter protesters and lone wolves. FIGURE 4: CYCLE OF DISORDER FROM SITUATIONAL THREATS AT PROTEST EVENTS TARGETING THE POLICE TABLE 1: UNIVARIATE DISTRIBUTION OF ALL STUDY VARIABLES Variable N (%) or Median (IQR) Mean Std. dev Forms of Police Intervention Nothing Newsworthya 5,556 (94.71%) 0.10 0.46 Legal Eaglesb 179 (3.03%) 0.10 0.46 Dirty Harryc 73 (1.24%) 0.10 0.46 Calling All Carsd 92 (1.56%) 0.10 0.46 Situational Threats Confrontational Tactics 574 (9.73%) 0.10 0.30 Protester Clash with Police 141 (2.39%) 0.02 0.15 Counter-protestersc 210 (3.56%) 0.04 0.19 Lone-wolf 128 (2.17%) 0.02 0.15 Police Characteristics Total 1033 Equipment Value (10,000) 123 (28.13, 342.11) 252.50 366.3 Total Law Enforcement Officers (1,000) 0.16 (0.05, 0.47) 0.60 1.41 Total citizens killed by police 2013 - 24 May 2020 5 (1, 15) 20.31 50.9 Control Variables Federal Agency Involvement 22 (0.37%) 0.004 0.06 Executive Action – Curfew 96 (1.63%) 0.02 0.13 County Population (100,000) 3.92 (1.19, 9.57) 9.00 15.73 Percent Black 8.02 (3.16, 17.73) 12.24 12.18 Percent Hispanic 10.68 (5.83, 21.82) 15.60 14.18 Percent of Population ?18 College Graduate 45.86 (34.77, 56.69) 47.13 16.47 Median Family Income (1,000) 80.32 (69.55, 95.62) 84.29 21.42 Median Age 38.30 (36.1, 41) 38.80 21.42 Abbreviation: IQR, interquartile range; Std. dev., standard deviation. aNothing Newsworthy: protests events with police no intervention. bLegal Eagles: police intervene using arrests, with no use of force. cDirty Harry: police use force and/or violence without arrests. dCalling All Cars: police intervened using both arrests and use of force/violence. TABLE 2: DESCRIPTIVE STATISTICS OF SITUATIONAL THREATS BY POLICE RESPONSE TYPE. Nothing Newsworthya Legal Eaglesb Dirty Harryc Calling All Carsd (N=5,556) (N=179) (N=73) (N=92) N (%) or median N (%) or median N (%) or median N (%) or median Variable (IQR) (IQR) (IQR) (IQR) Situational Treats Confrontational Tactics 426*** (8%) 76*** (42%) 26*** (36%) 46*** (50%) Protester Clash with Police 36*** (1%) 41*** (23%) 22*** (30%) 42*** (46%) Counter-protesters 188*** (3%) 18*** (10%) 0 (0%) 4*** (4%) Lone-wolf 77*** (1%) 38*** (21%) 4*** (5%) 9*** (10%) LEA Characteristics Total 1033 Equipment Value (10,000) 120+++ (28, 342) 140+++ (48, 368) 196+++ (55, 495) 179+++ (49, 366) Total Law Enforcement Officers (1,000) 156+++ (53, 445) 243+++ (85, 788) 311+++ (133, 647) 232+++ (105, 527) Total citizens killed by police 5 (1, 14) 5 (2, 20) 8 (2, 23) 6.5 (2, 29) 2013 - 24 May 2020 Control Variables Federal Agency Involvement 11*** (1%) 3*** (2%) 2*** (3%) 6+++ (7%) Executive Action – Curfew 51*** (1%) 22*** (12%) 8*** (11%) 15+++ (16%) County Population (100,000) 38++ (11, 95) 65++ (21, 109) 81++ (39, 130) 81++ (41, 140) Percent Black 8 (3, 18) 10 (5, 22) 13 (7, 26) 11 (7, 23) Percent Hispanic 11 (6, 22) 13 (7, 24) 12 (7, 19) 13 (8, 27) Percent of Population ?18 College Graduates 46++ (34, 57) 48++ (39, 63) 52++ (45, 64) 48++ (42, 60) Median Family Income (1,000) 80+++ (69, 96) 80+++ (71, 92) 80+++ (71, 99) 77+++ (72, 88) Median Age 38+++ (36, 41) 38+++ (36, 40) 37+++ (35, 38) 37+++ (35, 38) + or *p < 0.05; ++ or **p < 0.01; +++ or ***p < 0.001. Note: Pearson chi-square tests were performed on dichotomous variables, marked with * Note: Kruskal-Wallis tests were used for continuous variables, marked with + Abbreviation: IQR, interquartile range aNothing Newsworthy: protests events with police no intervention. bLegal Eagles: police intervene using arrests, with no use of force. cDirty Harry: police use force and/or violence without arrests. dCalling All Cars: police intervened using both arrests and use of force/violence. TABLE 3: MULTINOMIAL LOGISTIC REGRESSION OF SITUATIONAL THREATS ON POLICE RESPONSES. (n=5,900) Legal Eaglesa Dirty Harryb Calling All Carsc Variable RRR (95% CI) RRR (95% CI) RRR (95% CI) Situational Threats Confrontational Tactics 5.03*** (3.25, 7.80) 3.76*** (1.99, 7.11) 5.55*** (3.08, 10.03) Protester Clash with Police 29.68*** (15.17, 58.05) 52.05*** (24.52, 110.50) 91.02*** (44.50, 186.21) Counter-protestersc 2.99** (1.52, 5.87) 2.00 (0.62, 6.42) Lone-wolf 22.06*** (12.84, 37.91) 6.76*** (2.22, 20.65) 11.33*** (4.68, 27.45) Police Characteristics Total 1033 Equipment Value (10,000) 1.00 (0.99, 1.00) 1.00 (1.00, 1.00) 1.00 (1.00, 1.00) Total Law Enforcement Officers (1,000) 1.71** (1.17, 2.49) 1.32 (0.81, 2.13) 1.26 (0.80, 1.98) Total citizens killed by police 2013 - 24 May 2020 0.99 (0.98, 1.01) 1.00 (0.98, 1.01) 1.00 (0.99, 1.01) Control Variables Federal Agency Involvement 4.92 (0.88, 27.55) 3.94 (0.41, 38.27) 10.45** (1.98, 55.25) Executive Action – Curfew 16.01*** (7.93, 32.32) 10.15*** (3.89, 26.50) 15.00*** (6.10, 36.89) County Population (100,000) 0.97 (0.93, 1.02) 0.98 (0.93, 1.04) 1.00 (0.94, 1.06) Percent Black 1.01 (0.99, 1.03) 1.00 (0.97, 1.03) 1.00 (0.98, 1.03) Percent Hispanic 1.00 (0.99, 1.02) 1.01 (0.98, 1.03) 0.99 (0.97, 1.01) Percent of Population ?18 College Graduate 1.02 (1.00, 1.04) 1.04* (1.01, 1.08) 1.04* (1.00, 1.08) Median Family Income (1,000) 0.99 (0.97, 1.01) 0.98 (0.96, 1.01) 0.96** (0.93, 0.99) Median Age 0.98*** (0.93, 1.03) 0.93 (0.00, 1.19) 0.93 (0.01, 10.23) *p < 0.05; **p < 0.01; ***p < 0.001. Abbreviation: RRR, relative risk ratio; CI, confidence interval. a Legal Eagles: police intervene using arrests, with no use of force. b Dirty Harry: police use force and/or violence without arrests. cCalling All Cars: police intervened using both arrests and use of force/violence. d There were an insufficient number of counter-protester cases to calculate a RRR