The social psychological processes of ‘procedural justice’: Concepts,critiques and opportunities

Power and ‘procedure’

It is Thibaut and Walker (1975, 1978) who are largely credited with the creation of the now highly influential concept of ‘procedural justice’. ¹ Their seminal work demonstrated that criminal justice procedures which afforded disputants the opportunity to have ‘voice’ within the legal system were perceived as more ‘procedurally fair’ than were those that denied disputants this opportunity (Lind et al., 1997). This preference was maintained even in instances where ‘decision control’ was in the hands of a third party. Thibaut and Walker (1978) argue this motivation for ‘voice’ is a form of process control, based in a desire the disputants possess to shape the consequences of those criminal justice ‘procedures’. Thus, they ultimately imply that individuals value process control because they are embedded in a power relationship and ‘voice’ allows some opportunity to shape the outcomes that they receive.

Leventhal’s (1980) justice judgement model (JJM) similarly emphasizes that individuals value process control. Leventhal (1980) advanced six rules that he argued individuals use to judge procedural fairness. These are the extent to which a procedure: (1) is applied consistently across people and time (the consistency rule); (2) it is not unduly influenced by personal self-interest (bias suppression rule); (3) it is based on as much accurate and informed information as possible (accuracy rule); (4) there is the opportunity to change or overturn incorrect decisions (the ‘correctability’ rule); (5) it reflects the views and concerns of all subgroups and individuals affected (representativeness); and (6) the procedure is concordant with an individual’s moral and ethical values (ethicality rule). An important point to take from Leventhal’s (1980: 32) conceptual development is that far from being fixed and universal, individuals apply ‘procedural rules selectively and follow different rules at different times. The relative weight of procedural rules may differ from one situation to the next, and one procedural component to the next.’ In other words, from the early stages of theoretical development within the field it has been acknowledged that the ‘rules’ governing judgements of ‘fairness’ are not fixed but dynamic and contextually bounded.

Lind and Tyler (1988) critiqued these early theoretical models because they relied too heavily upon the idea of individual self-interested actors instrumentally controlling procedural decision making for personal gain, be it equitable or favourable outcomes. Their group value model (GVM) proposed that in contrast, criminal justice procedures carry important identity-relevant information. Put simply, the way authority is exercised ultimately communicates to people information about their status and standing within important social categories. Lind and Tyler (1988) subsequently argue that ‘socialization’ within nation states or communities engenders relatively stable and universally shared beliefs about what constitutes ‘procedural fairness’. However, while there was a tendency for people to agree with the notion that ‘voice’ or representation in the decision making of the police or courts was an important constituent of ‘fair process’, the GVM suggests that such perceptions are ultimately open to change as and where different patterns of ‘socialization’ are in place. The GVM therefore also emphasizes that the rules governing perceptions of ‘procedural fairness’ are not fixed and universal but are to some extent socially determined and contextually bounded.

‘Procedure’, identity and cooperation

The group engagement model (GEM; Blader and Tyler, 2009; Tyler and Blader, 2003) built upon the GVM and its relational model of authority (Tyler and Lind, 1992) as a means of explaining how a perception of procedural fairness subsequently engenders cooperative behaviours in groups, organizations and ‘society’. GEM’s conceptualization of social identity clearly draws from social identity theory (SIT; Tajfel and Turner, 1979) with Blader and Tyler’s (2009) suggestion that an individual’s social identity comprises a cognitive and evaluative component. As with SIT, the extent to which an individual feels a psychological sense of belongingness is tied to the emotional value attached to that group membership. Central to the theoretical contribution of GEM is its ‘social identity mediation hypothesis’: that ‘procedural justice judgments are […] a key antecedent of identity assessments. Identity assessments, in turn, are the key determinant of important psychological and behavioural connections to the group’ (Tyler and Blader, 2003: 357).

The GEM therefore explicitly acknowledges that procedural fairness judgements precede and to some extent determine judgements concerning identity. Moreover, the GEM appears to assume that the actions of authorities in their exercise of power through criminal justice ‘procedures’ have direct consequences for the extent to which individuals will or will not see themselves in terms of a specific and available social identity, and that this social psychological process ultimately mediates behavioural outcomes. Therefore, the GEM moves further beyond the implicitly individualistic models of rational self-interest towards an understanding of the centrality of group processes and social identity in shaping behavioural outcomes. Moreover, the GEM also begins to help us to understand that powerful group authorities can shape social contexts in ways that lead ‘subordinates’ to see themselves as active ‘citizens’ and, when they do, to act in a manner that is consistent with, or ‘conforms’ spontaneously to, the identity-based norms of that social context and identity.

Thus, early models of ‘procedural justice’ emphasized that individuals value having control in how decisions are made since it affords them some power to shape the outcomes that they receive within those procedures. However, these early ‘self-interest’ models were primarily focused at the level of the individual rational actor. By contrast the latter approaches begin to emphasize the importance of social identity and social contexts because they stress that the way in which powerful groups act has important identity-based outcomes. Accordingly, the actions of the ‘powerful’ upon the ‘subjugated’ determine whether the individuals in question see themselves as a ‘respected citizen’ or ‘valued’ member of a superordinate social category, such as the ‘organization’, ‘community’ or ‘nation state’. Nonetheless, there was also in this body of work important early recognition of the situational and cultural contingency of ‘fairness rules’. Indeed, while there are likely to be relatively fixed notions of what constitutes ‘procedural fairness’ across large populations it was nevertheless explicitly recognized that these judgements and their behavioural outcomes will vary with the specific socialized values, social contexts and identities in question.

Turn towards social cognitions

Tom Tyler is credited with being the first researcher to apply the concept of ‘procedural justice’ to the issue of ‘citizen’ encounters and experiences of the police. His seminal book on ‘why people obey the law’ (Tyler, 1990) is widely acknowledged to be the first comprehensive statement of PJT. However, it is important to recognize that PJT draws heavily upon an intellectual heritage and therefore carries with it many of the ideas and assumptions discussed above. In particular, Tyler (1990) sought to overcome the prior distinction between ‘rational self-interest models’ and the GVM’s identity-based approach, through a dual process model of ‘instrumental’ and ‘normative’ compliance (cf. Deutsch and Gerard, 1955). On the one hand, instrumental compliance accepts that people can be individually focused cost–benefit analysers when it comes to deciding whether or not to break the law and thus an external deterrence threat is the primary way to motivate acquiescence. On the other, normative compliance suggests that people also conform with the law because of an internal sense of obligation which is based less on deterrence and more upon judgements of procedural ‘fairness’ and legitimacy.

As with Leventhal’s (1980) JJM, ‘procedural fairness’ or ‘procedural justice’, is defined by PJT in terms of core constituent dimensions or ‘rules’, in this case four: participation; neutrality; dignity and respect; and trustworthy motives (e.g. Meares et al., 2014). Taking each dimension in turn PJT asserts that, first, people value having the opportunity to participate in the police decision-making process by having their say; to tell their side of the story or have ‘voice’. Second, people want the police to come to ‘fair’ or objective decisions by behaving with neutrality rather than in accordance with their own personal biases or stereotypes. Third, people want to be treated with dignity and respect by the police. Finally, in their interactions with police, people want to perceive that officers are acting benevolently or with ‘trustworthy motives’. Thus, people are viewed as critically concerned with the (un)fairness of their interpersonal dealings with police officers and the (un)fairness of the way police officers reach decisions.

According to early theoretical accounts of PJT, judgements of procedural ‘fairness’ then play a role in determining perceptions of police ‘legitimacy’ (Tyler, 1990). However, there is little consensus within the PJT literature regarding the precise definition and operationalization of police legitimacy. Broadly, delineations tend to coalesce around the notion that legitimacy entails a belief that the police are an appropriate, proper and just authority (Tyler, 2006). As such, Beetham’s (1991) threefold typology is often used to operationalize the concept such that for the police to be defined as ‘legitimate’ in the eyes of the policed they must: (1) act within prescribed laws; (2) embody shared norms and values, that is act in ‘normatively justifiable’ ways; and (3) there must be evidence of expressed consent from ‘the policed’.

For researchers who accept the above definitions of police legitimacy, ‘procedural fairness’ is therefore viewed as an antecedent factor to ‘police legitimacy’. Accordingly, Jackson et al. (2011) suggest that perceptions of police legitimacy can be measured empirically by three inter-related constructs, which do not in themselves contain any measures of ‘procedural fairness’. Acting within prescribed laws (or the ‘legality’ of the police action) can be assessed by exploring people’s perceptions of the extent to which police behaviour conforms to established legal principles. The degree to which the police embody shared values can be measured by asking people about their perceived ‘moral alignment’ with the police. Finally, expressed consent can be assessed by statements designed to measure the extent to which people feel an obligation to obey police directives. This conceptual and operational approach has recently been challenged by the ‘police legitimacy model’ (PLM) (Tankebe et al., 2016) that asserts perceptions of procedural fairness, along with distributive fairness, the lawfulness of police action and police effectiveness (e.g. at reducing crime) are all constituent parts of a broader overarching concept of ‘police legitimacy’ rather than potentially ‘legitimating factors’. Moreover, the PLM proposes that a felt obligation to obey the police is not an integral component of legitimacy but an outcome of the legitimate exercise of power.

However, it is our contention that this important debate runs the risk of reifying legitimacy and ‘fairness’ by overlooking the dynamic nature of these judgements and the inter-relationships with identity and social context. It is contended that this danger arises partly because the literature has yet to address empirically the processes through which fairness rules and perceptions of legitimacy are bounded, socially determined and rendered situationally and culturally contingent. The lack of focus on social process is arguably due, at least in part, to the foundational impact of Thibaut and Walker’s work on the development of PJT.

For example, Tyler and Folger (1980) applied Thibaut and Walker’s key hypothesis to police ‘citizen’ encounters; that is, they tested the notion that there is an important distinction between ‘procedures’ and ‘outcomes’. They focused on two types of police–‘citizen’ encounters: those where ‘citizens’ initiated a ‘call for assistance’ to the police and those where ‘citizens’ were actually apprehended as a potential suspect. They sought to explore whether judgements of police ‘procedural fairness’ affected ‘citizen’ ratings of satisfaction with the police independently of the favourability of the outcomes they received (i.e. whether the police solved their problem or gave them a ticket for a traffic or motor violation). They noted that while it was relatively easy for Thibaut and Walker to experimentally manipulate the more formal and ostensibly objective ‘procedures’ of the courtroom, police ‘procedures’ in their encounters with ‘citizens’ are less formalized and so are less amenable to controlled experimentation.

Thus, instead of ‘manipulating’ aspects of police actions to show the affect that they had on ‘citizen’ ratings of police satisfaction, they instead utilized survey methodology with the assumption that: ‘citizens will differentially perceive the fairness of the manner in which they have been treated, and that these perceptions will affect satisfaction independently of the outcome of an encounter’ (Tyler and Folger, 1980: 282). In so doing Tyler and Folger essentially took a turn inwards, treating a measure of the subjective perception of ‘procedural fairness’ as a quasi-independent variable and sought to test its effect on ratings of satisfaction with the police.

This methodological ‘twist’ is a key preoccupation that runs through much of the subsequent PJT work. Accordingly, following this initial study, a large body of survey evidence has been amassed linking the ‘citizen’ experiences of ‘procedural fairness’ to their perceptions of police legitimacy, a willingness to accept police decisions, satisfaction with the police and to behavioural intentions such as compliance with the law and cooperation with the police (e.g. Sunshine and Tyler, 2003a). Correspondingly, the mainstay of the PJT literature is not so much focused on the precise nature and context of an individual’s interactions with the police but rather it centres on perceptions of these encounters. In this respect, PJT has essentially become a social cognitive theory, in all but name. Thus, on the one hand, it stresses theoretically the centrality of the social relationships between a police officer and ‘citizen’. On the other, it is empirically concerned with cognitions or how ‘citizens’ perceive these encounters with police. Put slightly differently, the PJT literature often gives primacy to an analysis of an individual’s cognitions about policing at the expense of an exploration of the actual social encounters, contexts and processes through which such cognitions are ultimately understood to be formed.

Officers as mirrors

Despite the key emphasis on expressive, relational explanations of why procedural fairness matters to people in both the GVM and the GEM, there are only a handful of existing PJT studies that have empirically addressed the impact that social identity processes may have in relation to public perceptions of the police and policing (e.g. Bradford et al., 2014). Within this body of literature, there are two primary ways in which social identity is conceptualized. Both are explicitly based in the GVM and the GEM accounts outlined earlier and tend to assume that police officers, through the way they treat those they encounter, communicate messages concerning inclusion, status and value within the group that the police are assumed to represent. The social categories invoked in this research are usually conceptualized and operationalized in terms of national, community or ‘citizenship’ identities.

In the studies that draw upon the GVM, emphasis is given to the idea that group identification (e.g. national or community identity) precedes interactions with group authorities. For example, Murphy et al. (2015) explored the GVM with a representative sample of Australian residents. They posited that those strongly identifying with Australia will value procedural fairness since the police are important state representatives and so police treatment is especially ‘identity relevant’ to them. However, for those who identify more with their own ‘ethnic subordinate group’ (2015: 723) procedural fairness ‘may matter less’ (2015: 723) in explaining their intentions to cooperate, since the police represent a social category with which they do not identify (i.e. the nation state of Australia). Thus, GVM researchers have essentially suggested that people’s perceptions of and reactions to ‘procedural fairness’ will vary depending on the extent to which people already identify with the superordinate category.

By contrast, applied to policing, the GEM suggests that one reason why police procedural fairness is important to people is that police activity is ‘identity relevant’ and can actively shape and alter their subjective relationship to the categories the police are assumed to represent. Accordingly, people value procedurally fair policing as it indicates that they are included in, and a valued member of, some form of superordinate category. This in turn is assumed to encourage people to internalize and legitimize the values associated with membership of that superordinate category, thus engendering cooperation and compliance with its representatives (the police). Conversely, procedurally unfair policing suggests exclusion from this category. Hence, people will tend to reject the category norms and comply less with the group authority’s directives (Bradford, 2014).

There is evidence supporting both accounts. For example, in line with the predictions of the GVM, Huo (2003) reported that among a sample of Americans identification with the United States moderated the link between perceptions of procedural fairness and decision acceptance. Those who identified strongly with the USA placed a greater emphasis on how they were treated by legal authorities (i.e. procedural fairness) and less emphasis on the outcomes of their interactions. In concordance with the GEM, Bradford et al. (2014) conducted a representative survey of Australians and reported that when people felt the police were ‘procedurally fair’ their strength of identification as an Australian citizen and perceptions of police legitimacy were enhanced. Conversely, perceptions of police unfairness were related to a weakened sense of identification and lower levels of police legitimacy.

However, despite this empirical support it is suggested that the contrasting set of assumptions about the underlying relationships between police action and identity exposes various limitations to PJT work both in its GEM and GVM guises. One weakness of the current PJT literature is that researchers have tended to explore issues of social identity by focusing almost exclusively on what appear as relatively fixed and abstract superordinate categories. For example, Bradford’s (2014) measure of social identification involved asking Londoners how strongly they felt they belonged to: (a) their local area; (b) London; and (c) Britain. Whereas Sunshine and Tyler (2003b: 158) assessed the extent to which New Yorkers identified with an ‘imagined community’ with statements such as the ‘values of most of the people in my neighbourhood are very similar to my own’. Finally, Bradford et al.’s (2014: 549) measure sought to capture the extent to which Australians viewed themselves as law-abiding citizens of Australia with questions including ‘Do you see yourself first and mainly as a member of the Australian community?’ and ‘Do you see yourself as an honest, law-abiding citizen?’

By focusing predominantly on these kinds of social categories and by assuming police officers act as pre-defined ‘moral arbiters’ of entry to them, PJT research has tended to overlook relational identification (Radburn et al., 2018). More specifically, research has tended to disregard the extent to which people identify with the police as a distinct social group. Indeed, it seems reasonable to assume that people can identify with the police just as they can with other relevant social categories (e.g. one’s work organization; Blader and Tyler, 2009). Moreover, PJT research appears to conceptualize the normative and ideological content of these categories as relatively fixed and subsequently the police are often constituted un-problematically as being prototypical representatives of them, a prototypicality that is assumed but not often measured. An exception is Sunshine and Tyler (2003b) who explored the degree to which the police were prototypical of the ‘community’s’ moral values. They did this by asking New Yorkers the extent to which they agreed with statements such as ‘The police in [my] neighbourhood act in ways that are consistent with [my] own moral values about how people should be treated’ (2003b: 157). Yet with this measure it is noticeable that there are no direct references to specific police actions. Instead, such statements are aimed at assessing the extent to which the police in general represent the moral standards and values of ‘the community’. Moreover, Sunshine and Tyler’s (2003b) underlying theoretical approach explicitly assumes that the police are de facto prototypical group representatives of the community and nation state.

Thus, while the models of underlying psychology in PJT research recognize the centrality of ‘social identity’ processes they are grounded in the identity-based models of the GVM and the GEM, which convey a very specific set of propositions for how these forms of psychology function. First, in theorizing police–‘citizen’ encounters, PJT researchers have focused almost exclusively on the extent to which police activity affects an individual’s sense of relatively fixed and abstract social categories such as national, community or ‘citizenship’ identity. Therefore, PJT is currently somewhat limited in its capacity to understand the processes at work in more complex and dynamic intergroup encounters involving differing forms of identity (e.g. political affiliation, football fandom, etc.) where perceptions of police (il)legitimacy and levels of instrumental and normative compliance with the law can and do change. Second, it is largely taken as self-evident that the police are representative of these superordinate categories (Pehrson et al., 2017). As such, there is a danger of discounting the idea that by engaging in ‘unfair’ or ‘illegitimate’ actions the police may equally render themselves as unrepresentative state or community actors. For example, consider Reicher’s (1984) analysis of the St Pauls riot in Bristol. Far from the police reflecting and being prototypical representatives of the St Pauls community, the police were seen by crowd participants as hostile and racist outsiders attacking the ‘black community’ by raiding the Black and White cafe, an important community hub.

Intergroup contexts, identities and interaction

Despite the historical lineage of the theory being rooted in ‘in efforts to understand and explain riots and rebellion’ (Tyler and Blader, 2003: 351), there has been a paucity of PJT research focusing specifically on the emergence of criminality within or police management of crowds (Stott et al., 2011). Indeed, to date there has been only one study that has formally examined issues of procedural fairness, police legitimacy and social identity specifically in the context of policing crowd events. Stott et al. (2011) sought to utilize PJT and the elaborated social identity model of crowd behaviour (ESIM) to explain the presence and absence of collective conflict during football crowd events involving Cardiff City football fans. Undertaking a three-year longitudinal ethnographic study, their analysis showed that fans’ perceptions of police legitimacy were associated with a policing approach that sought to enhance communication and dialogue with fan groups. In this context, where fans adjudged their intergroup relations with the police as being legitimate, fans were more likely to comply with police directives and even ‘self-regulate’ – to essentially police themselves – psychologically marginalizing potential ‘trouble-makers’.

This work in many ways validated PJT’s central ideas and findings within the context of policing crowds, particularly regarding the importance of police treatment and perceptions of police legitimacy in encouraging ‘compliance’. Thus, Stott et al. (2011: 395) suggested, that ‘the processes we have identified here are consistent with the theoretical principles of both ESIM and PJT’. While agreeing with this broad assertion, it is argued here that there are conceptual limitations that emerge when applying PJT to the policing of crowds that merit exploration. Perhaps most significantly, as we have argued above existing PJT work is predominantly individualistic and interpersonal in its orientation, in the sense that the model of police–‘public’ interactions often assumed is that of a dyadic relationship between a ‘citizen’ and a police officer. While categories such as ‘citizen’ themselves come loaded with assumptions, Tom Tyler and colleagues argue that the ‘model of legitimacy we offer reflects the reality that interactions with the police are interpersonal experiences’ (Meares et al., 2014: 114).

Yet the social identity approach in social psychology arose, in part, out of dissatisfaction with reductionist individualistic approaches to group processes and as such draws a qualitative distinction between individual and group level psychology. Moreover, the approach also conceptualizes crowd events as fundamentally intergroup encounters. Therefore, to gain a fuller theoretical understanding of crowd–police relations there is a requirement to recognize that PJT must also be explicit that inter-relationships between police and ‘citizens’ can be intergroup rather than merely interpersonal in nature. In this sense, Taylor and Brown’s (1979) influential criticism of social psychology in the 1970s − that the study of interpersonal phenomena tended to be divorced from the wider social and historical context – rings true for the PJT literature. For instance, Armaline et al. (2014: 376) argue that PJT work does not tend to consider ‘the long and burdensome history of aggressive policing practices embedded into the social fabric of urban communities of color’. Indeed, Tyler and Lind (1992: 143) make clear that: ‘[u]nlike Tajfel and Turner […] we focus on the search for information about one’s position within one’s group, rather than on the position of one’s in-group vis-a-vis other groups’. We contend that this form of ‘de-contextualization’ has important implications for the explanatory power of PJT, particularly when applying it to crowd policing.

As suggested above, these conceptual issues appear to be inter-related to methodological ones. For example, most PJT research within a policing context comprises large population surveys that are utilized to identify national trends between people’s judgements towards their encounters with police officers and to their behavioural intentions such as their propensity to cooperate with the police or towards conforming to the law. As such there is a reliance on individuals’ self-reported views of policing independent of an in-depth investigation of the actual nature and context of these interactions. Thus, PJT researchers, through their methodological choices, often ‘have no way of knowing what the police are doing […] the beginning point of our analysis is the self-reports of community residents – policing as they experience it’ (Sunshine and Tyler, 2003a: 528).

These issues also have implications for the study of police legitimacy. As Smith (2007) argues, the vast bulk of PJT work has been focused on three concerns: (1) the extent to which an individual perceives the police’s actions to be fair; (2) how these judgements are related to the individual’s perception of whether the police are legitimate; (3) how these perceptions of the police relate to an individual’s behavioural intention to comply with the law and/or cooperate with the police. In other words, the focus has been on seeking an explanation for inter-individual differences in perceptions of police legitimacy and to establish its psychological antecedents (e.g. procedural fairness) and consequences (e.g. compliance with the law or cooperation with the police). Such research therefore tells us little if anything about the processes through which people come to perceive policing as illegitimate, do not comply with the law and otherwise go on to engage in violent confrontation with the police or other forms of collective criminality, such as the looting witnessed during the 2011 English riots (see Reicher and Stott, 2011; Stott et al., 2018).

The trajectory of research development in the field of PJT is of course completely understandable given that the primary theoretical concern to date has been on the validity of some of PJT’s central contentions. There can be little dispute that this work has done a good deal to validate some of the theory’s core hypotheses. Our critique is therefore to focus debate on how the approach needs to develop in order to expand on its explanatory power. In this sense, it is contended that by ignoring the immediate contexts of people’s interactions with police it is impossible to explore and ultimately explain the dynamics of change. For example, actors can enter a crowd event with a sense of police legitimacy but then develop very strong perceptions of police unfairness and illegitimacy during the event, perceptions that may last well beyond this specific encounter (Vestergren et al., 2017). Through the paucity of studies exploring the group level dynamics of actual encounters between ‘citizens’ and police it remains unclear how PJT currently helps understand these fundamental and enduring transitions.

Legitimacy, relational identification and the dialectics of ‘procedure’

Research on the policing of crowds at international football tournaments has demonstrated that reductions in ‘public disorder’ were associated with policing approaches designed to facilitate the expressions of fan identity (Stott et al., 2007, 2008). Moreover, Stott et al. (2008) showed how England fans who were planning to attend the 2004 UEFA European Championships in Portugal displayed strong negative correlations between measures of in-group identification (as an England fan) and a measure of identification with the police. In other words, prior to the tournament seeing yourself as an England fan meant that you tended to see the police as ‘nothing like me’. Given their long history of antagonistic relations with police at that time, this was perhaps unsurprising. However, during the tournament England fans collectively experienced their intergroup relations with the police as broadly legitimate, both directly and vicariously. In a post-tournament survey, taken in the weeks immediately following the event, the pattern of this correlation reversed, so measures of in-group identification among England fans now showed a strong positive correlation with the measure of identification with the police.

What this suggests is that the collective experience of legitimacy with police during the tournament related directly to contextually and historically derived identity norms and values. Put slightly differently, England fans perceived policing as legitimate because it facilitated what ‘we’ want to do and was different to the ‘heavy-handed’ policing ‘we’ have experienced elsewhere. The data also suggest that such identity-based interactions during these crowd events then transformed previously antagonistic intergroup relationships. Thus, when the police acted towards these crowds to facilitate the expression of their group identity (across time and events), the emerging perceptions and experiences of police legitimacy appear to have shifted the boundaries and normative content of the identities involved. The experience of police legitimacy may have in turn functioned to lead fans to see the police as ‘us’, and act collectively in terms of facilitating the carnival of football by ‘self-regulating’, to prevent disruption by anyone seeking to create ‘disorder’ (‘them’). ² To draw on Peelian cliches, by acting with perceived ‘legitimacy’, at that moment in that context for that time, the police were ‘the public’ and ‘the public’ were the police.

This finding does suggest that identity, legitimacy and intergroup interaction are intimately intertwined and as such it is important to understand how perceptions of police legitimacy and identity change through and within interaction – an opportunity that is concordant with Bottoms and Tankebe’s (2012) call for criminologists to explore the ‘dialogic’ nature of legitimacy dynamics between ‘power-holders’ and ‘audiences’. Drawing on the work of the sociologist Max Weber and moral/legal philosopher Joseph Raz, they argue that essentially all political regimes lay ‘claims’ to being legitimate, that ‘power-holders’ always attempt ‘to establish and to cultivate the belief in [their] legitimacy’ (Weber, 1978: 213). Bottoms and Tankebe (2012) suggest that there are two important implications to be derived from these assertions. First, that to ‘claim’ legitimacy is suggestive of the fact that authorities such as the police are often dealing with different social groups who may have opposing interests and perspectives; what may be legitimate for one group may be illegitimate for the other. Second, that the notion of ‘cultivating’ legitimacy is suggestive of an active and ongoing relationship between police officers and ‘citizens’. Thus, far from being an objective – even reified – given, fairness and legitimacy judgements are perhaps fundamentally interwoven elements of an ongoing, dynamic and historical process.

Finally, there is the question of power. Within the interpersonal relationships between police officers and ‘citizen’ PJT suggests that the power resides with the police officer by virtue of their position. As Bottoms and Tankebe (2012) put it the police officer is the ‘power-holder’ and the citizen the ‘audience’. Yet, the power relationship between police and crowds is not so simple. Crowds are places where such routine architectures of power can be reversed. According to the ESIM, collective conflict with the police is only possible when there is a shared sense of empowerment experienced by crowd members by virtue of their situationally determined shared self-definition that is demarcated, at least in part, by their united opposition against the police. The challenge that ESIM’s perspective on crowds and social identity processes confronts PJT work with is to articulate ‘how power emerges from and functions within social relationships with a definite social, ideological and historical content rather than reifying it as an abstract external force producing generic psychological effects’ (Turner, 2005: 1).