Rethinking police procedural justice

While procedural justice theory has become the dominant paradigm in thinking about police legitimacy, it has several important weaknesses. First, procedural justice's conceptually essential distinction between ‘process’ and ‘outcome’ is blurred in reality, which is visible both in empirical operationalizations and in researchers’ understanding of police work. Second, procedural justice theory views society through an implicit consensus lens, making it poorly equipped to address police–citizen conflicts and structural societal inequalities. This is evident in the theory's inability to unpack the dynamics of police–citizen interactions and its reluctance to problematize the police role in contemporary plural societies. To advance our understanding of police legitimacy and police–citizen relations, particularly among marginalized groups, we strongly recommend working toward theoretical renewal and empirical diversification.


Introduction
Recently, we have witnessed various waves of urban unrest centered around cases of police violence against members of ethnic minorities. In the United States, pivotal moments were the shooting of Michael Brown in Ferguson, Missouri in 2014, and the killing of George Floyd in Minneapolis in 2020. While particularly salient in the USA, clashes, protests and confrontations between the police and the public have recurred in different shapes and with different backgrounds in many other western countries over the past decade. In most of these cases, the tensions have had distinct ethnic dimensions and are strongly connected to society-wide debates surrounding discrimination and structural inequality.
These events have given rise to new efforts to improve police legitimacy and public trust in the police (Bayley, 2018;Saarikkomäki et al., 2021;Worden and McLean, 2017). However, the question of how to improve the relationship between the police and the public has occupied scholars, policy makers and police chiefs for many decades (Bottoms and Tankebe, 2012;Bowling et al., 2019;Schaap, 2021). Answers to this question differ and evolve: changes are clearly visible in the shifting popularity of different types of police trust-building strategies (Goldsmith, 2005). Although different strategies can coexist or recur several times, we have seen several phases in many countries (Schaap, 2018). First, a focus on increasing police professionalism and reducing political or other undue influences on police work. Second, a growing popularity of forms of proximity policing, such as community policing or neighborhood policing (Stone and Travis, 2011). Third, instrumental, effectiveness and efficiency-oriented approaches, emphasizing police service delivery in terms of crime fighting and rapid response (Schaap, 2018).
At present, a fourth wave of police trust-building strategies is in vogue, fundamentally different from (although not necessarily incompatible with) previous schools of thought (Tyler et al., 2015). This new wave, especially salient in the USA, is partially a sign of the times. There, the police have in recent years been facing a crisis that is Not about corruption, political influence, or miscarriages of law. It is not about the effectiveness of the strategies they employ to control crime. It is about the way in which individual police officers deal with the people they encounter daily in the course of their work. (Bayley, 2018: 125) The problem with police-public relationships in contemporary US policing is, by this definition, primarily found in individual interactions between police officers and citizens. Here, a specific theoretical perspective becomes especially attractive: procedural justice theory. Hence, during the protests following the death of George Floyd in June 2020, calls were made again to implement procedurally just policing to help defuse tensions between the police and large segments of the US population (Stone, 2020).
Procedural justice is currently a dominant police trust-building paradigm not just in the USA (Tyler et al., 2015), but across the globe; research is growing rapidly (e.g. Bradford and Jackson, 2016;Oberwittler and Roché, 2018;Wood et al., 2020). The core idea of the theory, supported in these empirical studies, is that citizens' perceptions of legal authorities treating them fairly and in a procedurally just manner are the primary influence on how these authorities are experienced as legitimate and trustworthy. People can accept even unfavorable police decisions if the process to reach them is perceived to be fair, making compliance and cooperation with the police-central concepts in the theorypositive outcomes of procedural justice (Tyler, 1990).
Procedural justice theory has several mildly different interpretations. It is often divided into different elements: first, citizens' perceptions of neutral and nonselective criminal justice processes; second, perceptions of policing agents treating people politely, respectfully and fairly; third, trustworthy motives; fourth, involving citizens in decision-making processes (Tyler, 1990). Another, related interpretation is that procedural justice is concerned with quality of treatment and quality of decision making by authorities (Tyler and Huo, 2002). The model is often contrasted with an instrumental perspective, in which trust is based on perceived police efficacy and effectiveness (Jackson et al., 2012). Increasingly, procedural justice theory has also been expanded to transcend the individual level. Procedurally just behavior by authorities is argued to signal to citizens that they belong to a group, a society, with shared values and moral purpose (Bradford, 2016;Sunshine and Tyler, 2003a;Tyler and Blader, 2003). This has added an element of social identity to procedural justice.
Expectations of what procedurally just policing can achieve are sky high. It appears to have taken the place of community policing as the internationally recommended cure-all for problems with policing and police-citizens relationships, especially regarding trust and legitimacy (see Nagin and Telep, 2020).
In this article, we argue that despite its seemingly rock-solid basis, procedural justice has two fundamental problems that scholars and police organizations ignore at their own peril. First, procedural justice's intuitively elegant distinction between process and outcome will, upon inspection, in many cases prove a mirage. Second, procedural justice theory views society implicitly through a consensus lens, by skirting around issues of conflict and cultural pluriformity. This is particularly problematic when thinking about the police.
Critique of the procedural justice approach is not entirely new. About 15 years ago, one of the pioneers of the approach, MacCoun (2005: 193), wrote of the 'dark side of procedural justice', showing how its principles can be used for ill (see also earlier Cohen, 1989). Over recent years, a steady trickle of critical publications has given more depth to the argument that with procedural justice, not all that glitters is gold (Bell, 2017;MacQueen and Bradford, 2015;Maguire, 2018;Nagin and Telep, 2017;Roché and Roux, 2017;Thacher, 2019;Wästerfors and Burcar Alm, 2020;Worden and McLean, 2017). Yet despite these cautionary tales, procedural justice is still viewed as predominantly unproblematic, including in some of our own work (e.g. Saarikkomäki, 2016;Saarikkomäki et al., 2021;Schaap, 2018). We aim to show the underlying assumptions behind the concepts used and where these fall short theoretically and empirically. Our goal in this article is not to reject the core ideas of procedural justice, but rather to address their weaknesses and open up avenues for renewal. This means it is not only worthwhile to extend and unite the diverse points of criticism into a single contribution, but also to start working toward an alternative framework that avoids the current pitfalls of procedural justice. This new approach may provide different leads for improving trust in and legitimacy of the police, while taking into account citizens' rights and well-being. The aim of this theoretical contribution is therefore twofold: first, to examine the problems of the procedural justice perspective and their consequences; second, to help construct an alternative perspective on how police-citizen relations and encounters could be improved.
We briefly outline the case for procedural justice, its origins and promise. Then, we will examine the limitations of the procedural justice approach along two main lines of argument. Finally, we will draw conclusions and sketch an alternative approach that may provide both scholars and practitioners with new leads on how to proceed.

The promise of procedural justice
There are convincing reasons why procedural justice became the dominant approach toward gaining public trust and police legitimacy. It filled an existing hole in police strategies. As a school of thought, procedural justice originates in legal psychology. Early studies attempted to find antecedents of acceptance and legitimacy of courtroom decisions (Lind, 1982;Thibaut and Walker, 1975). Their central assertion was that, where distribution of outcomes (verdicts and sentences) could not be changed, procedures could be adjusted to improve perceived fairness of the steps taken to achieve these outcomes. Procedural elements were found to be more important for perceived fairness and overall satisfaction of citizens involved in court cases than distributive ones.
Soon, this approach was applied to other types of conflict resolution, including the interaction between citizens and the police (Tyler and Folger, 1980). The expectation was that adjusting police behavior to align with what citizens think is fair, would improve citizens' belief that police motives are trustworthy. That perceived (police) fairness, as shown in interactions between officials and citizens, affects legitimacy in the eyes of citizens more than outcome or efficiency became the core argument of the procedural justice school of thought. However, it had broader implications for social science, as it placed legitimacy at the center of the question of why people obey the law and comply with authorities without the need for coercion (Sunshine and Tyler, 2003b;Tyler, 1990;Tyler and Huo, 2002).
Over recent decades, procedural justice theory increasingly focuses on the police. A good fit, considering the high degree of discretionary autonomy of police officers (Bittner, 1980): they have many opportunities to improve the process of interacting with citizens. Yet there are several other reasons why procedural justice theory has become so dominant in contemporary police-citizen research.
The first of these reasons lies in the powerful statistical relationships that have been found between procedural justice principles on the one hand, and trust, legitimacy and compliance on the other. Elements of police procedural justice, such as fair and respectful treatment, explain unprecedented amounts of variance-differences between individuals -when it comes to trust and legitimacy (Jackson et al., 2012;Schaap, 2018;Sunshine and Tyler, 2003b;Wolfe et al., 2016).
Second, by concentrating on fair treatment and focusing explicitly on obedience and public acceptance of police decisions, the purpose of the procedural justice paradigm is very clearly to reduce (the need for) police use of force (Tyler and Huo, 2002: 5). Third, procedural justice proves particularly important for victims of crime (Koster, 2017;Wolfe et al., 2016). With the increasing societal attention to victims all over the western world since the 1980s, the appeal of police and court strategies that emphasize treating victims well has grown.
Finally, the procedural justice paradigm concentrates predominantly on the individual (albeit with society-wide implications) and can thus be described as a micro-level strategy. The rise of proximity policing (neighborhood-oriented) strategies fit its place and time, when problems in the USA and UK were often phrased in terms of tensions between police and community (Scarman, 1982;Skogan, 1990). More instrumentalist approaches focus on police effectiveness on the macro level (for instance through national clearance rates), which matches the New Public Management emphasis on global output measurements as it gained dominance throughout the 1990s. Recall Bayley's (2018) assessment that the problem in police-public relationships is now defined to be on the level of individual interactions, and the value of procedural justice theory becomes obvious.

Key problems with procedural justice
We identify two key problems with the vast majority (but not all) of the existing procedural justice studies. The first issue is that procedural justice theory has trouble distinguishing process from outcome both empirically and conceptually, causing researchers to overestimate the importance of procedural justice. The second issue partly follows from the first: by implicitly applying what we consider a 'consensus perspective' (eschewing issues of conflict and pluriformity), procedural justice theory often carries a thoroughly one-dimensional image of society. We show this through two examples: procedural justice's inability to open up the black box of what actually happens in day-to-day experiences and police-citizen interactions, and its reluctance to critically assess the police role and function in plural society.

Distinguishing process from outcome empirically
First, we draw attention to how procedural justice is typically studied, emphasizing problems with correlation and causation between the central concepts. When it comes to the police, procedural justice is usually (although not exclusively) measured through quantitative cross-sectional survey research. Most of that research was initially conducted in the USA by Tyler and colleagues (Sunshine and Tyler, 2003b;Tyler, 1990;Tyler and Huo, 2002), but soon the efforts of Jackson, Bradford and their colleagues in the UK led to the inclusion of procedural justice measurements in the fifth wave of the European Social Survey (Jackson et al., 2011). This worked miracles for the emergence of research on this topic in many European countries. Their work has contributed enormously to our understanding of trust, legitimacy and procedural justice. However, the increasingly complex statistical modeling underlying most contemporary publications also obscures a systematic weakness of this approach.
Present-day quantitative sociology, social psychology and criminology are often characterized by sophisticated statistical models that can be portrayed in complex path diagrams. Arrows between various constructs show the theoretically hypothesized relationships between a range of variables. These relationships are then tested, often through advanced structural equation models, to see if they hold. By now, however, we increasingly see diminishing returns. Young (2011: 15-19) argued that contemporary criminology focuses too much on data and methods and too little on theory, causing a loss of balance. The risk is that scholarly efforts are more directed toward applying and finetuning research methods, rather than understanding the problem in question.
When it comes to quantitative procedural justice research, one crucial issue is that the vast majority, no matter how elegantly modeled, remains cross-sectional in nature: the various constructs were all measured at the same time. What these models do empirically, is merely to establish correlations (or covariances) between variables: any causality in these models is purely theoretically founded. This means that the process of the police-citizen interaction cannot be empirically distinguished from what happened before and after. This is not much of a problem when we are trying to explain attitudes through demographic characteristics. When older people have more trust in the police than youngsters, common sense reliably tells us that it is age that determines trust, not the other way around. However, this becomes increasingly blurred when we explain attitudes with other attitudes, as is common in procedural justice literature. The underlying theoretical relationships are contested. This means that causality may be reversed or that (unmeasured) third common causes may explain both variables (Nagin and Telep, 2017).
An example of this problem can be seen in a discussion on the precise relationship between perceived procedural justice and legitimacy. In a 2013 contribution, Tankebe found that perceived procedural justice is a dimension, an integral component, of legitimacy. This finding was repeated by Sun et al. (2018), yet challenged in another article by Jackson and Bradford (2019), who argued that the evidence points to perceived procedural justice being a determinant of police legitimacy.
The problem here is that structural equation modeling, as these researchers applied it, is unable to do more than test the strength and measure of fit of a pattern of covariations between a large number of variables. Fit measures, factor loadings or effect sizes say nothing about causality if the underlying data were measured at a single point in time.
Hence, a statistical model in which perceived procedural justice is specified as an integral part of legitimacy may fit the data just as well as one where it is specified as a powerful determinant of legitimacy. They may statistically be equivalent, meaning the nature of the relationship depends on one's theoretical logic. In the case of Sun et al. (2018) and Tankebe (2013), this is the theoretical work of Bottoms and Tankebe (2012); in the case of Jackson and Bradford (2019), the approach of Tyler (1990) and Tyler and Huo (2002).
So, is perceived procedural justice a determinant of legitimacy or a component of legitimacy? Can we distinguish process ('fair police treatment') from outcome (defined here as 'police legitimacy')? The difference between the two is essential: if a citizen has positive perceptions of police procedural justice, will that lead to more police legitimacy, or does it reflect pre-existing legitimacy? There is some empirical evidence that the latter may, in fact, be the case. Worden and McLean (2017) found in a combination of observation and survey research that there is only a weak relationship between actual observed police procedural justice and citizens' perceptions of police procedural justice in those same instances. The authors hypothesized that perceptions reflected the attitudes citizens already brought to the interaction, formulated through previous experiences or (social) media images, rather than based on what happened during the encounter (see also Bell, 2017;Roché and Roux, 2017;Wästerfors and Burcar Alm, 2020;Wilson et al., 2021). Trinkner et al. (2019), examining this hypothesis, reported inconclusive results based on a (longitudinal) vignette experiment.
It seems likely that both the actual interaction as well as previous experiences and preformed attitudes shape police legitimacy, but evidence is limited. Generally speaking, longitudinal or experimental studies with procedural justice, especially when rooted in real-life situations rather than hypothetical scenarios, tend to show far more ambiguous results than cross-sectional ones. As the latter dominate the field, the strong relationships between perceived procedural justice and other variables such as acceptance of police authority, police legitimacy and trust, are probably in need of careful nuancing.
As a result, while they may not fully solve causality issues, there is a need for using more longitudinal data. In addition, more attention should be paid to how people's perceptions of trust and legitimacy are constructed and formulated over time. This is a more complex process than current research indicates: the bulk of existing research applies survey methods, while qualitative research on procedural justice remains rare. However, qualitative approaches using observations and in-depth interviews can play an important role. Such studies could contribute not only to (re)defining the link between process and outcome, but also to (re)conceptualizing the central terms of procedural justice; that is, legitimacy, trust and compliance: what do they mean for different people and groups in society? What actually happens during interactions? What nuances are there? This chimes with arguments we will return to in the later sections on 'Day-to-day experiences and interactions' and 'The police role and function in society'.

Distinguishing process from outcome conceptually
The distinction between process and outcome is crucial in procedural justice literature (Tyler, 1990;Tyler and Huo, 2002). In the previous sections, we understood outcome broadly ('legitimacy', 'compliance'). In this section, we will add more depth to this perspective by framing outcomes as the result of specific interactions, which may include various actions or sanctions. The procedural justice approach has provided vital insights into how police officers act. It has emphasized the importance of fair, respectful and unbiased police treatment of citizens, and giving them a say, rather than focusing on the outcome of the interaction in terms of solved problems, fines, arrests and so on. Yet when we talk about the work of policing, it is unclear that processes can in fact be separated from outcomes.
Procedural justice theory emerged in legal psychology because courts and judges simply have to hand down negative outcomes (sentences, punishments, sanctions). The only way to improve satisfaction, obedience and legitimacy is then through improved fairness of the process. In transplanting the procedural justice paradigm to the police, it was argued that the police are in many ways similar to the courts: as links in the criminal justice chain, they need to fine, threaten and incarcerate people by the nature of their function (Tyler and Huo, 2002). As a result, attention has turned to the process through which these outcomes are achieved. Additionally, compliance is even more in question when it comes to the police than to the courts.
However, procedural justice thinking has insufficiently adapted to the changing context of the police as opposed to the courts; the police are a fundamentally different organization. Courts have time to consider their decisions, but are relatively limited in the range of possible outcomes. Police officers, as street-level bureaucrats, have larger discretionary autonomy, operate in a more dynamic environment, deal with a greater variety of people in different circumstances and hence have a wider range of ways to go about their dealings with citizens (Bittner, 1980;Skolnick, 1966). Considerable discretion allows the police to choose where to go, who to stop and whether or not to engage in interactions; they also have a variety of skills and legal possibilities (Bowling et al., 2019;Muir, 1977). Police officers often get to choose the type of outcome of the interaction (sanctions, warnings, forms of aid or use of force) on the spot (Thacher, 2019).
All of this opens avenues for illegitimate police actions and decisions as much as legitimate ones, raising the important question to what extent compliance and trusting the police is, under all circumstances, desirable (MacCoun, 2005). Meanwhile, the majority of police work concerns keeping the peace in a broad and general way, rather than law enforcement (Banton, 1964). Keeping the peace is, by definition, a process without end point; policing is always ongoing. While many police-public interactions do have clear outcomes, for the police, unlike for the courts, the process of interacting itself is often the outcome. This is a shared characteristic for 'people processing organizations' (Hasenfeld, 1972). This can also be observed from the side of those who are policed: are being stopped, being asked for one's ID, being taken to a police station or being asked to move on outcomes or processes? Because process and outcome are, in many police-citizen interactions, so intertwined, the result is that our understanding of what procedural justice research findings mean may be incomplete. We will show how this is the case in the following sections.

Day-to-day experiences and interactions
At the heart of building procedurally just policing are interactions between citizens and the police. As a result, it is essential to pay attention to citizens' subjective understanding (perceptions) of the police. However, this comes with limitations, as most of the research in this field still relies on survey methods. These may not always be the most suitable method to study interactions. Furthermore, current procedural justice research is often more occupied with measuring compliance and cooperation with the police than with aiming to understand people's judgments and experiences. Yet one could argue that it is inherently important that people are treated fairly and with respect: this should not merely be considered a tool to improve compliance or police legitimacy (Watson, 2020).
It is remarkable to see that citizens' (adversarial) experiences are rarely included in procedural justice research. While quantitative studies often include a general 'evaluation of police contact' variable, which certainly has some use, that is usually as deep as the analysis goes. This is problematic, since to build better police-citizen relations and to increase our understanding of trust and legitimacy, we require information on day-to-day experiences and interactions between the police and citizens.
Furthermore, we particularly need to understand the experiences of those groups that are the most common targets of police interventions. Such often quantitatively small, hard to reach, but essential sub-groups are where perceptions of police legitimacy are probably worst and with whom the police often spend much time (Lee, 1981). These groups include for instance people in low-income urban areas, ethnic minorities, homeless people, asylum seekers, youths, people with mental health issues and people in isolated rural communities. Only a limited number of studies focuses on how procedural justice affects these marginalized groups (see recently Kyprianides et al., 2021). Some surveys have focused on ethnic minorities and found that procedurally just policing increases views of police legitimacy among these groups (e.g. Bradford, 2016;Bradford et al., 2017;Madon et al., 2017;Murphy and Mazerolle, 2018;Tyler, 2005). Yet even here, while profoundly understanding ethnic minority experiences is key to good police work, many existing studies include ethnicity as a single background variable in a survey design to explain compliance and legitimacy. More versatile approaches to understanding the variety of experiences with the police in diverse societies are lacking.
With some exceptions using qualitative approaches (e.g. Novich and Hunt, 2017;Saarikkomäki, 2016;Saarikkomäki et al., 2021;Wästerfors and Burcar Alm, 2020;Worden and McLean, 2017), procedural justice research hence pays little attention to what is actually going on in the interaction between police officer and citizen. A new research agenda with a focus on micro-level interaction situations is needed, as are fresh methods. First, in-depth, time-consuming research, including ethnographic observation, would be valuable. In police research, such studies were more common in the 1960s through the 1980s (see, including many others, Banton, 1964;Bittner, 1980;Muir, 1977;Skolnick, 1966;Van Maanen, 1975), but could be considerably strengthened were they supplemented with present-day theoretical, methodological and empirical insights. Second, more research on citizen perceptions of these interaction situations is required. Police-citizen encounters are sometimes very intense events, at worst coercive and violent. These situations are emotionally profound for both parties in terms of feelings, expectations and details of verbal and non-verbal communication. These include the tone of a police officer's voice or the way they look at people, meaning that the structure and setting of the interaction are key (Saarikkomäki, 2016;Saarikkomäki et al., 2021). Such factors, which certainly influence police legitimacy and trust, cannot be measured easily in surveys.
However, police-citizen relations are not only about direct contacts. They also involve interpretations of other situations, hearsay, shared, collective experiences and images of police practices-family and friends play important roles here (e.g. Wästerfors and Burcar Alm, 2020;Wilson et al., 2021). Evaluating trust in the police is a social, historical, cultural and political matter-a matter of people's identifications, references and accumulated experiences (Bell, 2017;Loader and Mulcahy, 2003;Roché and Roux, 2017;Wästerfors and Burcar Alm, 2020). While some literature in procedural justice theory has attempted to integrate these notions within its larger body of work (e.g. Tyler and Blader, 2003), many researchers' reluctance to study them in their day-to-day settings, especially in neighborhoods exposed to targeted police practices, is to their detriment.

The police role and function in society
Although police behavior in interaction with citizens comes with its shares of problems, it is often not just unfairly or unjustly perceived behavior during the process that causes tensions, but the presence of the interaction in the first place. This is very much highlighted when it comes to the question of 'over-policing'. We called more attention to face-to-face (micro-level) interactions earlier. However, this should not come at the expense of attention to context and structural inequalities in society, and how these connect with the police role and function in our pluralistic societies (see also Bell, 2017;Roché and Roux, 2017).
In the consensus model on which procedural justice theory is in our view implicitly based, it is important for the police to show citizens that they share a common collective identity and align in core moral questions (Jackson et al., 2012). In this model, voluntary compliance is achieved through police legitimacy, instead of coercive power. Here, the state, and the police as its representative, is a benign actor working for the common good and all strata within society accept this as a fact-or should be convinced to do so.
In practice, in many societies, there are at least substantial minorities contesting this view of the state, policing and societal order, for instance due to over-policing or because the police are not seen to protect their group (Fassin, 2013). That police resources and attention are unequally distributed across society is well known, their activities being especially focused on groups sometimes labeled 'police property' (Lee, 1981): the poor, youth, migrants, ethnic minorities and other suspected and labeled communities (see Aliverti, 2020;MacCoun, 2005;Schclarek Mulinari and Keskinen, 2020;Solhjell et al., 2019). These groups and communities are often well aware of police scrutiny, which contributes to poor police-public relationships (Engel, 2005;Lee, 1981).
Procedural justice theory typically offers no solutions here, as it focuses on the process of interacting without asking whether the interaction itself is legitimate (Epp et al., 2014). The overrepresentation of ethnic minorities or marginal groups in being stopped without just cause, 'ethnic profiling', can even be concealed by applying procedurally just treatment (see Cohen, 1989). MacCoun (2005: 189) argued that in these cases procedural justice practices in fact have the potential to 'coopt citizens and distract them from outcomes that by normative criteria might be considered substantively unfair or biased'. Indeed, Saarikkomäki et al. (2021) found that ethnic minority youths' high trust in the police in the Nordic context made some of them accept unfair or even discriminatory interventions by police officers.
We return to the problem of distinguishing process from outcome. The problem, as phrased by Thacher (2019), is that the procedural justice paradigm, by nearly exclusively relying on empirical research, ignores the more fundamental moral questions that surround the emphasis on procedures rather than outcomes. Crawford et al. (2017: 23) noted that, in England, inhabitants of social housing are 'the subject of differential justice in ways that mere improvements in procedure will effect little change'. For many marginalized groups, it is hence argued that the law and the police are not there for their protection. The police function, from the perspective of these groups, is merely to control them, leading to what can be called anomie about the law (Bell, 2017;Brogden, 1982;Fassin, 2013).
The bulk of procedural justice research currently neither addresses well enough the problem of over-policing of certain groups, nor the profound cynicism of those groups of citizens toward the police and the criminal justice system (Bell, 2017), nor does it acknowledge sufficiently that police officers are more than individual characters: that they also represent a system, and that this system can be skewed or perceived as such (Schclarek Mulinari and Keskinen, 2020). After all, in many countries, there is a long history of conflicts between ethnic minorities and the police, and the police themselves are not operating in a vacuum: if societal structures are unequal and discriminatory, it is not a surprise if the police are as well. Police researchers have long drawn attention to the question of what the police are for and what kind of order they represent: that of the common, public good, or the status quo in which well-off groups or classes maintain their dominant position (Bowling et al., 2019).
Then there is the police organization itself to consider. The police identity is known to be shaped by specific socialization processes (Van Maanen, 1975) and the multifaceted construct that has come to be called police culture (Bowling et al., 2019). Among other things, police culture is said to be characterized by us-versus-them attitudes, strong internal solidarity and stereotypes associated with certain classes and (ethnic) groups (see also Holmberg, 2003;Terpstra and Schaap, 2013).
Summarizing, due to its consensus perspective, procedural justice theory and research currently do not pay sufficient attention to societal fault lines and problematic aspects of the police function and socialization. As a result, the paradigm tends to underestimate the scope of the challenge in police-public relationships. Especially problematic is its focus on ensuring citizen compliance with the police. A core question is whether compliance and cooperation can or should be expected in cases where structural exclusion and discrimination exists. Non-compliance can under some circumstances be defensible or even desirable, challenging unequal societal structures (Bell, 2017). Indeed, public noncompliance and protest may be among the most powerful drivers of improvements in policing (Schaap, 2021).

Discussion and conclusion
The past decades have seen changes and various cycles in the ways the police try to improve legitimacy and how scholars study this phenomenon. Currently, the dominant strategy and field of study is that of procedural justice. In this article, we aimed to explore the limits of the procedural justice school of thought in order to develop theoretical, empirical and practical leads for future research.
Our two main lines of critique involved, first, the challenge of distinguishing processes from outcomes, shedding light on larger empirical and conceptual issues within the theory. Second, we examined procedural justice's implicit assumption of societal consensus and its weakness in dealing with conflict and societal pluralization. Together, these issues reflect procedural justice theory's relative shortsightedness in terms of drawing from insights from prior research (see also Rock, 2005), its lack of variety of research methods and its limited sensitivity to context and structures (see also Young, 2011). In our view, we need rethinking both within the procedural justice paradigm and beyond. Recently, there have been promising contributions and novel approaches within procedural justice research using a variety of methods and focusing on novel groups (e.g. Bell, 2017;Kyprianides et al., 2021;Murphy and Mazerolle, 2018;Worden and McLean, 2017). More, however, is needed.
If we are to make further progress in research, both theoretically and empirically, as well as in our practical advice to police organizations, it is our firm conviction that a new, more holistic approach toward the study of the police is essential. Accordingly, we shall attempt to outline the beginnings of an alternative approach toward understanding and fostering police-public relationships. Throughout this next section, we will show what such a new approach may look like on two main strands. Obviously, if only for reasons of brevity, our endeavor will be limited to drawing a rough sketch of such an agenda. Our suggestion is to view this agenda as the starting point for further discussion.
The first strand of this agenda is the matter of focus. As we have indicated, how people's perceptions of trust and legitimacy are constructed and formulated is a far more complex process than current research indicates and there are problems when aiming to separate processes from outcomes. Furthermore, concepts used in procedural justice theory, such as compliance, obeying the law, cooperation or increasing police legitimacy, all carry an implied consensus idea. Meanwhile, procedural justice theory has been less focused on citizens' well-being, inclusion and substantive justice. We have highlighted, too, that blind trust in the police is undesirable (see MacCoun, 2005). Accordingly, we would encourage new research agendas to move beyond this consensus model and beyond focusing only on how police legitimacy is increased. New research should focus on how to build better police-citizen relationships, while acknowledging the tensions and risks inherent to the police function.
As the second strand of our proposed agenda, a new research approach ought to reconcile some of the most promising notions of different trust-building strategies and perspectives on different strategic levels. It could balance analysis of structural features in society with organization-level policies and with individual-level police discretion and actions. New insights should then have implications on the level of the state, of that of police organizations and on that of individuals. This would be demanding but realistic: the aim is to build on various intellectual and methodological traditions, requiring versatility of the researcher.
All in all, a future research strategy, then, should meet several conditions: it should study police-public relationships holistically and critically, meaning methodologically versatile and with attention to process, outcome, and how the two are intertwined; then, it should analyze actual police work and real-world interactions at the micro level; finally, it should take societal tensions and fault lines into account at a more structural level. We suggest five potential directions of inquiry that meet these criteria.
First, to regain an understanding of what daily police work looks like, who police officers are, what they do and what they do not do. This builds on a long, but in recent decades largely abandoned police research tradition (including among others the works of Banton, 1964;Bittner, 1980;Cain, 1973;Manning, 1977;Muir, 1977;Skolnick, 1966;Van Maanen, 1975).
Second, to address not only what the police do, but also how-how the police deal with crime, crime victims and suspects, how they secure public order and interact with citizens under different circumstances. How the police collaborate with other actors and partners (e.g. public-private partnerships), how they help and care for those in need, consider citizens' well-being and try to foster the public good. These two directions would help us understand the contemporary police profession, too often treated as a black box in procedural justice theory.
Third, not to focus only on citizen perceptions of police legitimacy, but to study how and why the police affect their surroundings more broadly, how they (re)shape citizens' identity and feelings of safety, inclusion and belonging, as well as the reverse-in such a way that it connects to legitimacy (see also Loader and Mulcahy, 2003;Saarikkomäki et al., 2021;Solhjell et al., 2019). Policing is about intense interactions, about vulnerability and emotion. Particularly important is to consider structural inequalities and minority groups' experiences of policing and how they constitute, at worst, not only (dis)trust and legitimacy of the police, issues on which procedural justice focuses (e.g. Bradford et al., 2017;Epp et al., 2014;Madon et al., 2017;Tyler, 2005), but also exclusion and marginalization of these groups (Bell, 2020).
Fourth, to take into account context and environment where police work takes placecross-national differences are of interest, but so are changing institutional environments, societal pressures and surrounding actors' contradictory wishes and expectations (see also Schaap, 2021). This links to a need to rigorously evaluate, using diverse approaches, reforms and changes in the police organization and their consequences for police-public relationships.
Fifth, to refocus police research on the explicitly normative notion of 'good police work', which is not necessarily the same as efficient or even trust-building police work (following Bittner, 1980;Bottoms and Tankebe, 2012;Muir, 1977). The craft of the police profession should take center stage, with its antecedents and consequences, as should the question of what good police work is under what conditions. At the same time, we should observe the limits of police ability to make a difference for good (Bowling et al., 2019;Manning, 1977).
Together, these suggested directions widen the scope of analysis beyond the focus of procedural justice theory. They could both expand our understanding of police-public relationships and temper overly high expectations when it comes to improving these relationships in terms of practical perspectives. For police leaders and policy makers, procedural justice has offered tempting leads. Procedural justice training is now common in many police organizations, with promising results (Dai, 2020;Wood et al., 2020) and the use of body-worn cameras, too, fits procedural justice's emphasis on improving individual interactions (Demir et al., 2020). We, however, without rejecting the above practices and many insightful findings of prior procedural justice studies, advocate a broader approach toward improving police-public relationships. The central goal in this approach would be to refocus police efforts on public well-being, guaranteeing that the police are a neutral actor working for the public good. Elements from procedural justice theory may contribute to these aims, but we should be careful to note that neither they nor aspects derived from other approaches provide a magic bullet (Roché and Roux, 2017).
The approach taken throughout this article obviously comes with a number of limitations that should be considered. First, we have focused here on public police organizations, omitting reference to pluralization and transformation of policing. Furthermore, we have focused on police organizations and legitimacy in western countries; there is a need to broaden this perspective. A final consideration is that the way police organizations function is intertwined with deep societal, structural characteristics. Changing this is bound to be extraordinarily challenging for a range of well-documented reasons. We understand that not all will agree with our call for more attention to structures and contexts of policing, preferring the 'solid ground' of more limited questions. We do not reject these. However, given that policing agents occupy a pivotal position in our rapidly changing societies and represent the authority of the state, we are obliged to address such fundamental issues.
It is our position that improvement in the relationship between the police and the public can be attained step by step. Insights from procedural justice could foster this goal, but should not be seen in isolation and must certainly not be considered the holy grail of police legitimacy. We hope our analysis will contribute to a reappraisal of procedural justice not as a solution to every policing problem, but rather as one of many potentially useful yet imperfect leads for both understanding and improving policepublic relationships.