The Effect of Paramilitary Protest Policing on Protestors' Trust in the Police: The Case of the “Occupy Israel” Movement

Abstract

The use of paramilitary methods in civil policing tasks has become common in Western police agencies. Despite propositions that such methods should undermine the relationship between the police and the public, the effect of paramilitary policing on public trust in the police has not been empirically tested. In the present study, we examine this question in the context of protest policing, which has become a major concern for Western police agencies. Using a survey of 470 protesters who participated in “Occupy” protest events in Israel in 2012, we find that the perceived use of paramilitary methods has an independent and negative effect on trust, stronger than that of police effectiveness and the “neutrality” component of procedural justice. In-depth interviews suggest that the significance of paramilitarism may be the result of a sense of alienation and criminalization it elicits among protesters who generally perceive themselves as law-abiding citizens.

Paramilitary policing is defined as the use of military methods by civil police agencies for civil policing tasks, and may include using paramilitary policing units (PPUs), or providing officers with military training, equipment, weaponry, and uniform (Kraska 2007). It is frequently argued that this policing style has become common in Western police agencies (Balko 2006; Kraska and Cubellis 1997; Kraska and Kappeler 1997; McCulloch 2001; Rantatalo 2012; Vitale 2005). While the police in Western democracies have emphasized community-oriented policing since at least the early 1990s (Willis 2014), paramilitary methods, which in many ways appear to conflict with community policing, are often perceived as the efficient answer to problems such as organized crime, terrorism, and social protests (Kraska and Kappeler 1997; Murray 2005).

The growing prevalence of paramilitary policing has stimulated a large body of literature in political science, sociology, criminal justice, and criminology, which mostly addresses its definitions, origins, and expressions in the field (e.g., Den Heyer 2014; Hills 2009; Perito 2004; Rantatalo 2012; Salter 2014). As for the effects of this policing style on police-community relationships, scholars have argued that it alienates police officers from the citizens they serve and undermines procedurally-just treatment, including transparency and accountability (Hill and Beger 2009; Kraska and Cubellis 1997; Pino and Wiatrowski 2006). At the same time, the effects of paramilitary policing on public views of the police, and specifically on trust, have not been tested empirically. This question is critical because of the growing use of paramilitary policing practices, and because of the critical implications of trust in the police in terms of public compliance and cooperation (e.g., Gallagher et al. 2001; Murphy 2014; Reisig et al. 2007; Tyler 1990, 2004). What is more, understanding the effect of paramilitary policing on trust can contribute to our broader understanding of the predictors of public trust in the police.

In this study, we address this question in the context of protest policing, using demonstrations organized by the “Occupy Israel” movement in 2012. By “protest policing,” we are referring to the variety of tactics used by police at protest events, including selecting the best unit to handle the specific crowd, evaluating the potential for violence, negotiating with protest leaders, integrating undercover officers in the crowd, and conducting arrests. It is considered a complicated and potentially dangerous task, and thus often involves some level of paramilitary response (e.g., Della Porta and Tarrow 2012; Gillham 2013; Rafail et al. 2012). What is more, in the last decades scholars have documented a significant rise in both the number of protest events and in the heterogeneity of citizens participating in protests (Bédoyan et al. 2004; Norris et al. 2005). Thus, protest policing is an important and useful arena for examining the effects of paramilitary policing on public trust.

We begin by reviewing the concept of paramilitary policing and its specific application in protest policing. We then review the literature on public trust in the police and hypothesize about the potential effects of paramilitary policing on trust. Our study context, sample, survey, and main variables are described next. Using survey data collected in the 20 largest protest events organized by “Occupy Israel” in 2012, we find that the perceived extent to which the police used paramilitary methods at the event has an independent, negative, and relatively strong effect on protesters' trust in the police. Based on in-depth interviews with protestors, we suggest that the employment of paramilitary methods elicits feelings of alienation and criminalization among citizens who are used to viewing themselves as normative and “on the same side” as the police.

Paramilitary Policing—Background

Police agencies are often characterized as quasi-military organizations (e.g., Bittner 1970; Murray 2005), but in the last few decades, the term “paramilitary policing” has been used to describe a more specific phenomenon: the employment of PPUs, military methods and equipment by civil police organizations for civil policing tasks (e.g., DeMichelle and Kraska 2001; Hills 1995; Kopel and Blackman 1997; McCulloch 2001; Weber 1999). Peter Kraska, arguably the most prominent scholar discussing paramilitarism in American policing, refers to paramilitary policing as a continuum, ranging from zero to full paramilitarism. This continuum is expressed primarily by the integration of PPUs in civil police organizations. These units are easily distinguishable from “regular” police due to their martial gear and weaponry, military jargon, and excessive use of force (Kappeler and Kraska 2015; Kraska 1997, 1999).

Data suggest that the employment of paramilitary methods in American policing has expanded: while in 1982 only 59 percent of American police agencies had PPUs, this figure rose to 78 percent in 1990 and to 89 percent in 1997 (Kraska and Cubellis 1997). In 2014, Kraska estimated that PPUs are used about 50,000 times per year in the United States.¹ Moreover, paramilitary methods are employed by “regular” patrol units and small-town police agencies (Crank et al. 2010; McCulloch and Pickering 2009; Pickering and McCulloch 2010). Several reasons for this trend have been identified, including police involvement in counterterrorism in the post-9/11 era (Kennison and Loumansky 2007; McCulloch 2004; Perliger et al. 2009), private market manufacturers' encouragement of “dual-use” of technologies and weaponry by both military and civil agencies, agencies' motivation to off-load military surplus onto local law enforcement² (Katsineris 2016; Maguire 2015), and police attempts to bolster their self-esteem and public image by using “prestigious” military weaponry (Kraska 2007).

While most of this literature discusses American policing, studies suggest similar developments in other Western democracies, particularly the UK (Bunyan 1985; Hills 1995; Jefferson 1990), Australia (McCulloch 2001; McCulloch and Sentas 2006), and South Africa (Tait and Marks 2011). The boundaries between military and civilian policing are fading in other parts of the world as well, including Mexico, Turkey, and India (Davis 2008; Durna and Hancerli 2007; Goldsmith 2008; Mayfield et al. 2013; Verma 2008). However, in these examples, militarization refers to active involvement of army units in civilian policing tasks. Our focus is on PPUs, which, in spite of their military appearance, are an integral part of civil police agencies. As defined by Rantatalo (2012: 51), PPUs are “organisations within law enforcement bodies that in differing degrees are modelled after the military, but with the statutory powers and legitimate status of the police.” In most Western democracies (including Israel, our study site), military forces have no jurisdiction over civilians within the borders of the state. Our focus in this study is on the paramilitarism of civil police units and actions.

At first glance, the rising popularity of paramilitary policing appears to contradict the major development of community policing. However, research findings suggest that this is not necessarily the case, as the two apparently opposite approaches often co-exists in police agencies (Kraska 2001, 2007). Moreover, police officers perceive them as complementary: using interviews with American police officers, DeMichelle and Kraska (2001) find that officers view paramilitary methods as efficient tools in achieving what they perceive to be community policing goals: crime prevention, social order, peaceful neighborhoods, and a sense of personal safety.

Paramilitary Policing and Social Protests

Although paramilitary policing is often associated with counterterrorism, such methods have been used in a variety of policing tasks. One important area in which paramilitarism has become particularly dominant is the policing of social protests (Della Porta and Tarrow 2012; Fernandez 2009; Gillham et al. 2013; Juska and Woolfson 2012; McCulloch 2004; Starr et al. 2011; Williams 2011). In Western democracies, the main goal of the police responding to protests is to allow protesters to execute their right to free speech while maintaining public order. This already complicated task has become even more demanding in the last decade, due to a major rise in the number of protests and protesters, and increased variation in the characteristics of the latter—more citizens from different backgrounds want to exercise their right to demonstrate dissatisfaction with various issues (Earl and Kimport 2009; Soule and Earl 2005). The risk of a calm protest escalating to violence, the large crowds involved, and the perceived threat social movements more generally pose to the government have pressured police to provide an efficient and firm response, which, in turn, result in the frequent employment of paramilitary methods (Ericson and Doyle 1999; Soule and Davenport 2009).

There is no single definition in the literature for “paramilitary protest policing.” Here, we use Vitale's (2005) definition because it derives from the perspective of the protesters (rather than the police), which is the focus of our study. This definition is based on what protesters perceive to be political intent to treat them in a military fashion, as reflected, for example, in the presence of PPUs, the use of less-lethal weapons (water cannons, rubber bullets, tear gas), extensive control of public space (massive presence of police officers, police barricades to block street segments), and excessive use of force (Rantatalo 2012). These measures are not exclusive to paramilitary policing, but they are expected to contribute to protestors' perceptions that they are treated by the state in a military (rather than civil) fashion, particularly when more than one of these methods is being used (Kraska 2007).

Clearly the police may not need to employ paramilitary methods. If the event is peaceful and there is no risk of violence, the police may not be present at all, or may simply make sure the protest takes place as planned (Earl 2011). Earl et al. (2003) found that the police were present only in the minority of protest events in New York State during the period of their study (1968–1973). Moreover, if protestors begin to behave violently, approaches such as “negotiated management” suggest dealing with the situation using direct communication with the protesters, keeping arrests to the minimum, and restricting the use of physical responses to civilian/unarmed police (Della Porta and Reiter 1998; Earl 2011; McPhail et al. 1998; Waddington 1994). This approach was successfully used in at least one “Occupy” protest event in Salt Lake City (Scott 2012). Similarly, British procedures instruct police officers to “facilitate peaceful protest” and avoid employing paramilitary methods when unnecessary (Her Majesty's Chief Inspector of Constabulary 2010: 21). In Israel (our study site), police regulations classify protest events into four categories based on the level of disturbance to public order they pose. The two lowest levels do not require any paramilitary responses, and when they are perceived to be needed, their use must be preceded by a warning, approved by a senior commander, and progress gradually based on need (INP 2003). Thus, “protest policing” and “paramilitary policing” are not interchangeable terms—protest policing does not have to involve paramilitary responses.

Nevertheless, when paramilitary methods are used in protest policing, their aim is to achieve outcomes, including maintaining public order and handling civil unrest (Beede 2008). Indeed, evidence suggests that such methods can be effective. For example, Gillham and Marx (2000) conducted ethnographic observations and interviews with protesters during the World Trade Organization in Seattle. They describe how paramilitary methods, such as using tear gas, pepper spray, and heavy deployment of PPUs, were effective in achieving some short-term goals such as evacuating a specific road to allow access. Mouhanna (2009) describes how the French police, and specifically French PPUs (“Gendarmes mobiles”), effectively stopped the violent wave of protests in 2005. At the same time, this is not to say that paramilitary policing does not come with a price. As described below, there are strong reasons to suspect that paramilitary policing may have significant and negative effects on public trust in the police.

Paramilitary Policing and Public Trust in the Police

Trust in the Police and Its Predictors

Trust is an emotional element in a relationship between two parties; in the case of policing - between a citizen and a police officer. The level of trust attributed to the officer reflects the citizen's subjective belief that the officer will behave in desirable ways (Hawdon 2008). As indicated by Tyler and Huo (2002: 58), “trust in a person's motives or character refers to his or her internal, unobservable characteristics that are inferred from his or her observable actions.” Beyond its normative importance (Hough et al. 2010; Jackson et al. 2012a, 2012b; Tyler 1998), trust in public institutions is critical because of its potential effect on citizens' behavior through the expectations it dictates. When citizens trust the police, they expect officers to act in valued ways, such as protecting citizens' rights, investigating and mitigating crime, and obeying the law. These expectations, in turn, may motivate cooperation with the police, even when the citizen does not appear to gain direct instrumental benefits (Tyler 2016).

There is wide empirical support for the proposition that trust promotes public cooperation with the police (Jackson and Bradford 2010; Jackson et al. 2012a, 2012b; Sargeant et al. 2014; Tyler 2016). For example, Tyler (2005) conducted phone interviews with a sample of 1,653 New Yorkers representing both Whites and minorities, and found trust to be a predictor of willingness to cooperate with the police, including reporting crime, providing information about suspects, and reporting dangerous or suspicious activities. Sargeant et al. (2014) conducted a similar survey among 10,000 citizens in two major cities in Australia, and found trust to be the primary predictor of willingness to report crime and suspicious activities, help the police identify suspects, provide information, and assist the police if asked to do so.

Despite the consistent findings, it is important to note some of the complexities of this literature. First, the measured outcomes in these studies are attitudes—declared willingness to cooperate with the police, not actual behavior. Second, public trust in the police has been studied both as an independent attitude (e.g., Jonathan-Zamir and Weisburd 2013; Murphy et al. 2009; Tankebe 2009a, 2009b) and as part of an index measuring police legitimacy (along with obligation to obey and/or normative alignment with the police; e.g., Huq et al. 2011; Papachristos et al. 2012; Sunshine and Tyler 2003; Tyler et al. 2014). This inconsistency appears to derive from lack of consensus regarding the conceptual and operational definition of police legitimacy.

Further elaboration on this debate is beyond the scope of the present review (e.g., Bottoms and Tankebe 2012; Jackson and Gau 2016), but it is important to clarify that there is no disagreement about the critical role of trust in police-community relations. In addition to normative arguments about trust being an essential component of the relationship between citizens and authorities in democratic societies (Goldsmith 2005; Tyler and Huo 2002), trust is critical for practical reasons—the police cannot function effectively without public cooperation, which, in turn, cannot be achieved only through deterrence or other instrumental considerations (e.g., Tyler 2004). Given the critical importance of trust, and given that most studies in Israel (our study site) have focused on trust rather than on the broader concept of legitimacy (Aviv 2014; Hasisi and Weitzer 2007; Jonathan-Zamir and Weisburd 2013), we use trust as the dependent variable in our study.

Two models have been suggested as the foundations of trust in the police. The “performance-based approach” submits that citizens expect the police to “do their job”: prevent crime, mitigate its consequences and maintain public order. Thus, when they are perceived to be doing so, they earn public trust (Sunshine and Tyler 2003; Tyler 2005). On the other hand, the process-based approach (“procedural justice/fairness”; e.g., Tyler 1990; Tyler and Huo 2002) is based on the proposition that people care more about how they are treated than about the outcome of the interaction, because events such as encounters with police provide a setting in which people gain important information about their status and self-worth. Being treated fairly sends the message that the individual is valued (Antrobus et al. 2015; Blader and Tyler 2015; Jonathan-Zamir et al. 2015a, 2015b). In turn, assessments of procedural justice were found to develop from four key characteristics of the process: (1) participation—was the individual involved given the opportunity to contribute to the decision-making process?; (2) neutrality—were decisions made impartially, using objective criteria? (often expressed as transparency); (3) respectful treatment—was the individual treated with politeness and courtesy, and were their rights acknowledged and respected?; and (4) displays of trustworthy motives—did the police demonstrate that they are truly concerned with the well-being and quality of life of the citizens involved or society more generally? (Blader and Tyler 2003; Tyler and Fagan 2006).

Studies have found both police effectiveness and procedural justice to be robust predictors of trust, but procedural justice was often found to be the primary predictor (Gau et al. 2012; Hinds & Murphy 2007; Jonathan-Zamir and Weisburd 2013; Kochel et al. 2013; Mazerolle et al. 2013; Tankebe 2009a, 2009b). It should be noted, however, that some have argued that the strong relationship between procedural justice and trust is an artefact of them being two dimensions of the same theoretical construct. These critics highlight the importance of empirically distinguishing between procedural justice and trust when testing the predictors of the later (Gau 2011; Maguire and Johnson 2010; Reisig et al. 2007; Tankebe 2013; Tankebe et al. 2015), and, as detailed below, we follow this approach.

The Hypothesized Effects of Paramilitary Policing on Trust in the Police

It is frequently argued that paramilitary policing may undermine the relationship between the police and the public because it involves forceful and offensive strategies, which may be perceived as an expression of an authoritarian state (Kraska 2007; Kraska & Cubellis 1997; McCulloch 2001; Pino and Wiatrowski 2006; Wiatrowski and Pino 2008). Emsley (1983) adds that using paramilitary methods to “conquer” an indignant civil crowd is at odds with liberty and democratic principles, and Waddington (1987) argues that forceful policing methods portray officers as agents of an enemy government. Thus, there are strong reasons to suspect that paramilitary policing would undermine public trust in the police, but little empirical evidence.

In the context of crowd policing in football games, studies identified that paramilitary policing was perceived by fans as an attack on their democratic rights, which led to violent responses (Hoggett and Stott 2010; Stott and Drury 2000; Stott et al. 2007, 2008). Using observations, Jefferson (1990) found a similar effect in protest policing in the UK. Drury and Reicher (2000) also analyzed the behavior of police and protesters in the UK. Based on interviews with protesters (1999) and observations of protest events (2000), they found that protesters expected the police to defend their democratic rights, but when the police used forceful methods they felt that they were denied these rights, and, in turn, viewed the police and the “system” they represent as illegitimate. This promoted unity among protesters and encouraged violent behavior. More recently, Gillham et al. (2013) observed police responses in “Occupy” protest events in NYC, and found that the police were employing “strategic incapacitation”: controlling space, surveillance, and information-sharing. This approach resulted in violence, including blocking streets and assaulting police officers. Recently, Maguire et al. (2016) conducted a survey among 136 “Occupy DC” protestors in Washington. They measured protesters' perceptions of police behavior, procedural justice, and support for violence. Many testified that the police used TASER, stun-guns, pepper spray, and other weapons against them. In turn, the use of these tactics encouraged support for violence. This relationship, however, was not significant when procedural justice was accounted for.

These studies suggest that police use of paramilitary methods in crowd-control tasks affects protestors' views, particularly attitudes about using violence against the police. While violent attitudes can suggest lack of trust, the effect of paramilitary policing on trust in the police has not been measured directly in previous studies, nor were other relevant factors controlled for. In this study we directly examine whether protestors' assessment of the extent to which paramilitary policing methods were employed in a particular demonstration affects their trust in the police, while accounting for the two main predictors of trust identified in the literature—procedural justice and police effectiveness.

Study Context

The Israel National Police, Protest Policing, and Paramilitary Policing in Israel

The Israeli model of policing was initially inherited from the British mandate in Palestine, which ended in 1948 with the establishment of the State of Israel (Herzog 2001; Weisburd et al. 2009). Similar to police agencies in North America and Europe, the Israel National Police (INP) operates in a Western, democratic society, and is thus restrained by law, court orders and other regulations, and obligated to preserving civil rights. The INP is a national and centralized organization, primarily due to the small size of the country. Over the years, the INP has been influenced to a great extent by trends in American policing, including the adoption of community policing in the 1990s, Compstat, and more recently—hot spots policing and problem-oriented policing (Gimshi 1999; Shalev 1999; Weisburd and Amram 2014; Weisburd et al. 2009). Thus, despite historical and geographical differences, there is general agreement among policing scholars that the Israeli model of policing resembles those of other Western democracies, although some differences, particularly in terms of centralization, were also noted (Hasisi et al. 2009; Jonathan-Zamir et al. 2014, 2015a, 2015b; Perry and Jonathan-Zamir 2014; Weisburd et al. 2002, 2009).

Protests in Israel are policed under the command of the relevant geographical district of the INP. When policing a demonstration, the police commander in charge aims to achieve three main goals: maintain public order; allow the protest to take place; and protect the protesters. To do so, she may use a variety of non-military methods, including negotiating with protest leaders, communicating directly with the protesters using a sound system, and blocking street segments. These measures are similar to those used by other democratic police agencies (e.g., Button et al. 2002; Gorringe et al. 2012; Noakes et al. 2005; Rafail et al. 2012). Depending on the number of protesters, preliminary risk assessments, and the resources available to the district, the event commander may also call for the help of national PPUs (see below).

Due to our focus on paramilitary policing, two specific features of the Israeli context warrant additional elaboration. First, although the INP was established as a highly militarized organization (on the basis of the British Mandate Police), in the second half of the twentieth century the INP has gone through a “demilitarization revolution,” shifting the focus to human rights and citizen-oriented innovations. This was expressed, for example, in the establishing of the “Civil Guard” to engage Israeli citizens in peace keeping within their communities and the adoption of community policing (Friedmann 1992; Gimshi 1999; Herzog 2001; Hod 1996; Shadmi 1998; Weisburd et al. 2002). At the same time, similar to militarization trends in the United States, over time PPUs in Israel grew in numbers and missions (Herzog 2001). We are referring to two national PPUs that are an inherent part of the INP and participate in civilian policing tasks: the Riot-Control Special Unit (“YASAM”) and the Border Guard (“MAGAV”). Both are paramilitary in training and weaponry and wear khaki-green uniform, which make them easily distinguishable from “regular” police officers (termed the “Blue Police” due to their blue uniform; see Weisburd et al. 2009).³

A second important feature is the personal involvement of most Israeli citizens with the Israeli Army. The compulsory military service in Israel means that most adults, particularly those from the majority Jewish community, have served in the Israel Defence Force (IDF) for two to three years. Should we expect this kind of personal experience to affect their views of paramilitary policing? In his article reviewing militarization and demilitarization trends in the INP, Herzog (2001) speculates that the compulsory military service is one source of police militarization, suggesting that military experience encourages acceptance of paramilitary policing. Similar hypotheses were made by Black and Kari in a US study (2010); however, they found no differences between veterans and non-veterans in attitudes toward different policing styles (paramilitary vs. community policing). We return to this point in the Discussion section of the article when considering the generalizability of our findings.

“Occupy Israel” Movement and Protests

In this study, we take advantage of protest events organized by the Israeli branch of “Occupy.” This international protest movement started in Europe and North America in 2010 and later spread throughout the world, playing a major role in the recent wave of global resistance against neoliberal austerity following the 2008 financial crisis (Calhoun 2013; Kilibarda 2012). In line with other branches of “Occupy,” the goal of the Israeli branch is to “create a social and economic, deep and comprehensive, change in Israel.”⁴ The first “Occupy Israel” demonstration took place in Tel-Aviv on July 14, 2011. Since then the movement organized over 20 demonstrations in 2011 and over 30 in 2012, most in four central cities: Tel-Aviv, Jerusalem, Haifa, and Beer-Sheva.

“Occupy” represents the only example in the last few decades of a protest movement that developed following a global crisis, with common features across Western democracies (Calhoun 2013). Thus, it is reasonable to expect much similarity in the characteristics of “Occupy Israel” protesters and their American or European counterparts. At the same time, contrary to the American branch, “Occupy Israel” was not associated with either the political right or left, because in Israel the distinction between the two is based primarily on security-related debates, not economic ones (Rosenhek and Shalev 2013). “Occupy Israel” expressed goals that were deliberately designed to create an “all-Israeli” identity, and surveys suggest that this aim was achieved: 88 percent of all Israelis supported the actions of “Occupy” (Yediot Aaharonot 2011). Thus, by focusing on “Occupy” protests, we were able to study the relationship between paramilitary policing and trust in the police in a context that represented a broad range of political ideologies.

Methodology

The Sample

Protesters were sampled in all “Occupy” protest events that took place in Tel-Aviv, Jerusalem, Haifa, or Beer-Sheva in 2012. However, in order to allow for adjustment of the standard errors due to the nested nature of the data (protestors are clustered within protest events), we decided on a minimum threshold of 20 protesters from each event. Thus, 12 protest events in which fewer than 150 protestors took part (as estimated by the police) were excluded from the final analysis. This left us with the 20 largest events (see Table 1).

Table 1. Demonstrations and Protesters

^a Based on police estimations as published in the Israeli press.

The first author and one/two research assistants attended all protest events. They arrived at the site 30 minutes before the official starting time, and stood separately in central locations until the conclusion of the event. They approached every protester who passed them and asked him/her to provide their email address in order to participate in a survey. A week later a survey was emailed to those who agreed to participate. Between 87 percent and 95 percent of the protesters approached provided their email addresses (M = 91 percent). The age and gender of those who refused were documented and compared to the final sample, and no significant differences were found. Of 653 emailed surveys, a total of 470 were returned, resulting in an overall response rate of 72 percent, which is considered very good for online surveys (Cobanoglu et al. 2001; Nulty 2008).

Table 2 presents the characteristics of the protesters in our final sample, in comparison to the general population in Israel. It reveals that protesters in our sample are similar in their sociodemographic characteristics to those described in the literature (Schussman and Soule 2005): relatively educated, similar in their income to the general population, and slightly overrepresented by men and by the majority group (in our case Jews). Our data also indicate that most protesters in our sample define themselves as secular (85.4 percent), reside in central Israel (“Gush-Dan”; 71.8 percent), are members of a social movement engaged in political protest (71.8 percent), and have participated in more than five demonstrations in the year preceding the survey (67 percent). In our model, we control for these and other factors that may affect trust in the police.

Table 2. Sample Characteristics

^a Data obtained from the Israeli Central Bureau of Statistics (CBS), the Statistical Abstract for 2012; see www.cbs.gov.il.

^b Due to the way data are reported by the CBS, the frequencies for the Israeli population apply to citizens who are 20 years old and older.

^c Due to the way data are reported by the CBS, the frequencies for the Israeli population apply to citizens who are 20 years and older. The median was calculated from data reported categorically.

^d Due to the way data were reported by the CBS, the median for the Israeli population applies to citizens who are 15 years old and older.

The Questionnaire and Main Variables

Our questionnaire was made up of 84 items. In the present analysis, we use items from three themes detailed below: the extent to which paramilitary policing methods were used during the demonstration (as perceived by the protester); trust in the police and its main predictors as identified in the literature (procedural justice and police effectiveness); and control variables, including sociodemographic characteristics and previous experiences with the police and protests. The questionnaire was pre-tested prior to data collection using a sample of 23 protesters from the first “Occupy” protest event in Israel in 2012; the questionnaires were filled out in the presence of the first author immediately following the conclusion of the event, and items that were unclear to the protestors or did not appear to tap the theme they were designed to reflect were subsequently modified.

Paramilitary Protest Policing

As reviewed earlier, in the context of protest policing, paramilitary methods include the presence of PPUs, excessive use of force, and heavy deployment of riot-control measures: rubber bullets, water cannons, cavalry, and tear gas (Vitale 2005). While these measures are not exclusive to “paramilitary policing,” each is expected to contribute to one's sense that she was treated by the state in a military fashion, particularly when multiple tactics are used (Kraska 2007). Following this conceptualization, we created a formative index⁵ by adding five yes (1)/no (0) items: Were you verbally assaulted by the police?; Were you physically assaulted by the police?; Did the police use riot control measures—rubber bullets, water cannons, cavalry and tear gas?; Were Border Guard PPUs (“MAGAV”) present at the event?; Were riot-control PPUs (“YASAM”) present at the event? (N = 456; range: 0–5; M = 1.84; SD = 1.50).

Notably, the items and overall index reflect subjective views regarding the use of paramilitary methods. We recognize that protesters may perceive or remember the same situation differently, or be exposed to different levels of paramilitarism at the same event. Although understanding what the police did objectively is clearly important, protesters' subjective interpretations are precisely what should affect their trust in the police, and are thus the focus of the present analysis.

Trust in the Police and Its Expected Predictors

Items in this theme were designed as statements based on past research conducted both in Israel (Jonathan-Zamir and Weisburd 2013) and elsewhere (e.g., Gau 2011; Reisig and Lloyd 2009; Tyler 2002). Respondents were asked to rank their agreement with statements on a scale ranging from 1 (“do not agree”) to 5 (“completely agree”). Due to frequent criticisms of the measurement and operational definition of key terms in this literature (see above), the construction of the scales in this theme was guided by factor analysis (an examination of the Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy suggested that the sample was appropriate for factor analysis; KMO = 0.90), which confirmed that trust in the police, procedural justice, and police effectiveness are indeed distinct constructs (see Table 3).

Table 3. Factor Analysis Differentiating Trust in the Police, Procedural Justice, and Police Effectiveness

Note: N = 470; Extraction method: Alpha factoring. Rotation method: Oblimin with Kaiser Normalization; The loaded items are marked in bold.

As can be seen in Table 3, “trust” (factor 1) was constructed by averaging six statements, such as I have trust in the Israel National Police (Cronbach's α = 0.80; N = 444; range: 1–4.83; M = 2.41; SD = 0.75). As suggested by the factor analysis, “procedural justice” in our data should be treated as two separate scales: “participation” + “respectful treatment” (factor 4) and “neutrality” (factor 3).⁶ The scale for participation/respect was made up of five statements such as: Police officers in the INP give people the opportunity to express their views before making decisions; and The Israeli police treat people with respect (Cronbach's α = 0.83; N = 446; range: 1–4.60; M = 2.03; SD = 0.72). The scale for neutrality was constructed using four statements, including: The INP usually discriminates against citizens based on their ideology (reversed) (Cronbach's α = 0.67; N = 443; range: 1–4.50; M = 1.78; s.d. = 0.74). The final scale in this theme taps protesters' evaluations of police effectiveness in the context of protest policing (factor 2),⁷ and was composed by averaging five statements such as: The INP is efficient in reacting to violence during demonstrations (Cronbach's α = 0.82; N = 411; range: 1–5; M = 2.55; SD = 1.13).⁸

Control Variables

Previous studies found sociodemographic characteristics to affect trust in the police (e.g., Jonathan-Zamir and Weisburd 2013; Reisig et al. 2007), and we thus account for these in our analysis. We also control for previous contact with the police, membership in the “Occupy” social movement, and participation in previous demonstrations (see Appendix A for descriptive statistics of control variables).

In-Depth Interviews

The final item of the questionnaire asked the participants if they would be willing to be interviewed about their personal experiences during the protest event. These interviews are not the main focus of the present article, but are used to help interpret the quantitative findings. They are particularly useful in illuminating the mechanisms linking perceived use of paramilitary policing and protesters' trust in the police.

The interviews were conducted with all protestors who agreed—31 men and 21 women. Their age ranged from 23 to 45 (M = 33). All were Israeli Jews, and defined themselves as secular. T-tests revealed no significant differences between the interviewees and the rest of the sample in terms of sociodemographic characteristics and perceptions of the police. The interviews were carried out by the first author within the month following the protest event, at the interviewee's preferred location (e.g., home, café). They lasted between two to six hours, and were recorded and transcribed. We used a semi-structured protocol: all questions were prepared in advance and usually asked in a pre-determined order, but the interviewer could change their sequence or add follow-up questions based on the flow of the conversation. The basic list included 24 questions tapping three main themes: personal background, experiences during the protest event, and views of the police.

All notes and transcripts were analyzed using thematic analysis (Arksey and Knight 1999; Patton 1990). This form of analysis included six stages: (1) preliminary read; (2) preliminary coding; (3) preliminary themes; (4) theme validation; (5) final themes; and (6) coding report. These themes were used to interpret the quantitative findings, as detailed in the next section.

Findings

We begin by examining the prevalence of paramilitary policing in “Occupy Israel” protest events (as perceived by the protestors). We find that over 80 percent of protesters witnessed what they perceived to be some level of paramilitary responses (only 18.85 percent had the score of “0” on the “use of paramilitary policing methods” index), and 16.63 percent witnessed very high levels (4 or 5) (34.64 percent had the score of “1”; 13.59 percent—“2”; 16.22 percent—“3”; 9.86 percent—“4”; 6.77 percent—“5”). At the same time, our data also indicates that the police did not employ all paramilitary methods to the same extent: an overwhelming majority of respondents (76.20 percent) stated that they saw Border Guard PPUs (“MAGAV”) at the event, and 57.10 percent claim the police used riot-control measures. Less than one third (30.10 percent) stated that they were physically assaulted by the police, and only 13.90 percent were verbally assaulted. Twenty percent testified that riot-control PPUs (“YASAM”) were present.

Our main interest is in identifying the role perceived paramilitary policing plays in forming protestors' trust in the police. In Table 4, we present a series of Ordinary Least Squares (OLS) regression models (see Appendix B for correlation matrix). In model I, we test whether the perceived use of paramilitary methods indeed undermines trust. In model II, we add the sociodemographic characteristics and previous experience with the police/protests. In model III, we add the “classic” predictors of trust in the police—procedural justice and police effectiveness, and examine if paramilitary policing retains an independent, statistically significant effect.

Table 4. Predictors of Protesters' Trust in the Police

^a *p < 0.05; **p < 0.0l; ***p < 0.001.

Note: Standard errors are adjusted for the clustering of the data (by protests).

Tolerance levels were larger than 0.510 for model II and 0.490 for model III, suggesting no multicollinearity problems (Menard 1995).

As can be seen from Table 4, all three models are statistically significant, and explain 14 percent (model I), 19 percent (model II), and 55 percent (model III) of the variance in protestors' trust in the police. All R ² changes are statistically significant at the 0.001 level. Model I reveals that, as expected, the perceived employment of paramilitary methods has a statistically significant and negative effect on protesters' trust in the police. The impact of paramilitary policing remains robust after controlling for personal characteristics and previous experience with the police/protests (model II), and after controlling for the “classic” antecedents of trust (model III). Moreover, perceived paramilitarism was among the strongest predictors in the final model: it had a somewhat stronger effect than the “neutrality” component of procedural justice (with the beta value of 0.15 versus 0.13, albeit in opposite directions), and was more closely associated with trust than police effectiveness (β = 0.11). The only variable in the model with a stronger effect is the participation/respect component of procedural justice.

How can we understand the significant, independent effect of paramilitary policing on trust in the police? The in-depth interviews with protestors provide some insight. First, they suggest that perceived paramilitarism affects trust because it elicits a sense of alienation—the feeling that the protesters and the police are not “on the same side.” They are two separate, or even rival, entities. For example, one interviewee explained: “When it's the “Blue Police” present, everything is calmer… but when PPUs get involved, it's like they are declaring that we are the enemy… and we see them the same way.” Another added: “When PPUs are involved in policing the protest event, they give you this feeling—we do not work for you. You are the enemy [and] we are here to keep you quiet.” After being arrested for the first time, one of the protest leaders explained: “You learn that you are the enemy to them, so you shouldn't expect them to serve you or consider your rights.”

Second, some protesters experienced the use of paramilitary policing as an act of criminalization—as if police perceive and treat them as criminals rather than law-abiding citizens. “Criminalization” takes “alienation” a step forward. Not only are the police and the protestors on opposing sides, the protestors' position is no longer legitimate as it is not within the realms of the law. This approach, in turn, influenced the way protestors felt about the police: “This kind of treatment from PPUs - especially the use of water cannons, horses, batons, and mass arrests - makes you feel as if you are breaking the law, makes you feel like a criminal… although you know that you are just standing at the same place with thousands of people, doing nothing wrong, you feel like a criminal, and then you start to see the police as a criminal would see them.” Another added: “I don't know how criminal arrestees feel, but I imagine it to be very similar… you just stop believing everything you believed before: that the police are there for you, that they can be trusted, that they will protect you.”

Discussion

Despite the growing prevalence of paramilitary policing methods in the post 9/11 era, and important discussions concerning their potential negative effects on the relationship between the police and the public, we are unaware of studies that have empirically examined the impact of this policing style on trust in the police. Protest events in Israel, in which the police employed paramilitary methods to varying degrees, allowed us to examine this question. We find that paramilitary protest policing significantly undermined protestors' trust in the police. Given the nature of paramilitary policing and previous studies on crowd control, this finding is not surprising, although to our knowledge it is the first empirical evidence of the negative relationship between paramilitary policing and trust.

Given the nature of paramilitary methods, it is reasonable to suspect that they would undermine trust because they are perceived as procedurally unjust. Paramilitarism may also weaken perceptions of police effectiveness, because it may elicit violent reactions on part of the protestors. The mediating effect of procedural justice and effectiveness was indeed demonstrated in our models: the regression coefficient of paramilitary policing dropped from −0.24 in model II to −0.09 in model III, where these two factors were accounted for. At the same time, even in this model paramilitarism showed a statistically significant effect on public trust. Moreover, it was the second-strongest predictor in the model, overtaking police effectiveness and neutrality, and surpassed only by the participation/respect component of procedural justice (although we acknowledge that the difference between the effect of paramilitarism and that of neutrality and effectiveness is small).

Thus, our study makes an important contribution to the study of the antecedents of public trust in the police by identifying a new and important predictor—paramilitary policing. Moreover, while the literature suggests that we should rely primarily on just procedures to promote trust in the police, our findings suggest that when using some policing tactics (such as paramilitary ones), procedural justice may not be enough. In this sense our findings add to a recent, important debate—is public trust primarily about what the police do or about how they do it? Our findings echo those of Jonathan-Zamir et al. (2016): procedurally just treatment is critical, but what police do (in our case—use PPUs or riot control measures) also affects citizen attitudes, irrespective of how they do it.

As suggested by the in-depth interviews, two main feelings mediate the relationship between perceived paramilitary protest policing and protesters' trust in the police. The first is a sense of alienation. This proposition is in line with the argument that paramilitary policing leads officers to view the city as a “war zone” and its inhabitants as the “enemy” (Hill and Beger 2009), positioning police and protesters as rival groups on opposite sides of an imaginary border. This perception is in line with a key component of the classic police culture—the “us versus them” mentality (Kappeler et al. 1998; Paoline 2003; Willis and Mastrofski 2016). Moreover, paramilitarism contributes to a sentiment of criminalization, echoing Muzzatti's argument about paramilitarism in the context of counterterrorism: “[paramilitarism] is used by the police to criminalize a wide range of nonviolent political and social activists committed to progressive social change” (2005: 120). Thus, it appears that the strong, independent effect of paramilitary policing on trust stems from the new position protestors find themselves in: as noted earlier, protesters in our sample are mostly normative, middle-class citizens. Given their profile, it is not surprising that many have had little experience with the police: 58.70 percent had no personal contact with the police in the year preceding the demonstration, and, of those who had, only 5 percent were involved in a police investigation. Being subjected to paramilitary policing placed them, perhaps for the first time, on opposite sides from the police. The police were no longer perceived as “protectors” or “service providers,” but as distant government agents, treating protesters as criminals. Such feelings, in turn, undermined trust, independent of their views regarding police fairness or effectiveness.

This gap between the way protesters perceive themselves and their understanding of the way police officers view them adds to the important discussion on “policing as a debate of identities” (Bradford 2014; Bradford et al. 2014; Stott et al. 2012). In the context of protest policing, we find this debate to go beyond personal identities, and include the group identity of the protester as part of a social movement. In this context, paramilitary policing is perceived as condemnation not only of individuals but of the social movement, and in some cases of the democratic idea at the basis of its actions. Thus, our findings draw attention to the importance of considering not only one's personal identity but also her group identity when striving to preserve public trust in various policing contexts.

Returning to the specific context of protest policing, our findings highlight how policing strategies perceived to be professional, effective, and most suitable for some missions (Gillham and Marx 2000; Mouhanna 2009) can backfire. Less trust in the police would likely mean less cooperation and compliance, in both the short and long terms (e.g., Reisig et al. 2007; Tyler and Fagan 2006, 2008). Research on crowd control describes the reaction to paramilitary policing as a “cycle of violence”: citizens react to what they perceive as forceful policing; the police, in turn, respond with more force to what they perceive as disorder, which inflames more violence on part of the crowd. Moreover, clashes in one event may affect the inclination of protesters to respond violently in future events, and the profile of future protesters: if a social movement has a record of violence, “normative” citizens may hesitate to join, while violent activists may be motivated to do so (Drury and Reicher 2000; Earl 2004, 2005; Reicher 1996; Stott and Reicher 1998; Stott et al. 2008). It is important to note that despite the disadvantages of paramilitary approaches, we are not suggesting that they should necessarily be abandoned altogether. Rather, if police do decide to use paramilitarism, the implications on trust should be taken into account. More research is needed to assess how these methods compare to other approaches in terms of costs and benefits.

We began with the question of the effect of paramilitary policing on protesters' trust in the police, but other interesting findings emerge from our data. First, our model was similar overall to those reported in previous studies on public trust in the police: procedural justice was found to be the primary predictor of trust, followed by evaluations of police effectiveness. Thus, social protesters appear to be no different from the general population in the value they ascribe to the quality of treatment they receive from law enforcement and to their effectiveness. At the same time, although often treated in the literature as a single construct (e.g., Gau 2014; Lind and Tyler 1988; Tyler and Blader 2013), “procedural justice” did not behave that way in our data: our factor analysis showed “neutrality” to load on to a separate factor from “participation” and “respect,” and the latter was found to be more influential than the former. We attribute this finding to the unique characteristics of protesters, which may have led to an “expectation-experience gap.” As noted above, similarly to the profile of social activists in the literature (e.g., Schussman and Soule 2005), the protesters in our sample can be characterized overall as normative, middle-class citizens. Thus, we expect that they are used to getting respectful treatment from authorities, and have consequentially experienced a “procedural justice shock” in protest events when, perhaps for the first time, they are treated in a disrespectful manner or are not offered the chance to be heard, as they are used to and believe they are entitled to being heard. Thus, “participation” and “respect” become particularly dominant. Because this population is not a minority group, we suspect that it is less sensitive to issues of neutrality (Tyler and Wakslak 2004; also see the “expressive harm” hypothesis [Risse and Zeckhauser 2004]). In this regard, it is noteworthy that the tense relationship between the police and ethnic or racial minorities has received much attention (Murphy and Cherney 2011; Tyler 2005; Tyler and Wakslak 2004; Waddington et al. 2004). Our findings show that in some circumstances, the relationship between the police and majority communities can be no less sensitive: when well integrated, “privileged” citizens feel the police are denying them what they believe they are entitled to, the gap between them and the police widens and their trust in the police is compromised.

Before concluding, we want to identify some specific limitations of our work in drawing inferences about paramilitary policing. Every study context has unique features, and it is thus important to discuss our specific context in relation to our findings. As noted earlier, similar to many police agencies in the democratic world, the INP has gone through trends of demilitarization and militarization. Moreover, the day-to-day activities and regulations that guide the work of Israeli police officers are highly similar to those of their counterparts in many Western democracies (Weisburd et al. 2002, 2014). At the same time, the compulsory military service in Israel may have affected the attitudes of protestors in our sample. Given the arguments and findings reviewed earlier, we can expect that our protestors' military background either had no effect on their trust in the police (see Black and Kari 2010), or had a positive effect on the way they perceive paramilitary policing, leading to improved trust (Herzog 2001). Thus, in contexts where protestors have little or no military background, we can expect the negative effect of paramilitarism on trust to be similar or stronger. The fact that the INP is a national, centralized police agency may mean, for example, that the use of national PPUs in civil demonstrations has different meanings in Israel than for example in the United States, where border police and other national policing units are not generally integrated with regular police agents. Finally, we studied paramilitary policing as applied to a social protest movement concerned primarily with economic questions, which represented a broad range of social backgrounds and political orientations. Paramilitary policing may have different impacts when it is applied to movements based on ethnic, racial, or political grievances. Accordingly, we encourage future researchers to broaden the study of paramilitarism to other policing contexts.

Conclusions

Although paramilitary policing has become common in Western police agencies, and despite the critical importance of public trust in the police, the effects of this policing style on trust have not been tested. In this study, we find that in the context of protest policing, the perceived employment of paramilitary methods has a negative and relatively strong effect on protesters' trust, independent of the effects of its two “classic” predictors—procedural justice and effectiveness. Based on in-depth interviews, we attribute this effect to a sense of alienation and criminalization felt by citizens who are used to seeing themselves as part of the normative, law-abiding community. Thus, in addition to extending our understanding of the predictors of public trust in the police, our findings provide strong warning to police agencies that are rushing to implement paramilitary methods in challenging arenas such as terrorism, protests, and organized crime.