External Procedural Justice: Do Just Supervisors Shape Officer Trust and Willingness to Take the Initiative With the Public?

Decades of empirical research have shaped our understanding of organizational justice in the workplace and public assessments of police procedures on the street, but only recently has a nascent wave of research sought to better understand the role that officer perceptions of supervisory procedural justice play in shaping their (un)fair interactions with the public. The nascent research testing this relationship has focused on the evidence that officer perceptions of trust in the public is a pathway between internal procedural justice and external procedural justice. This article tests the role of trust and a parallel pathway that incorporates the concepts of work engagement and personal initiative in the procedural justice literature. Relying on a survey of 638 Croatian police officers, this study finds that the effect of supervisory procedural justice on officers’ external procedural justice is positive but indirect through a measure of trust in the public and the proposed engagement/initiative mechanism. The implications of these findings for research and police practice are discussed.

continue to refine our understanding on how fair and just procedures shape the public's attitudes and behavior toward the police. Over the last decade, policing scholars and practitioners seeking to promote procedural justice on the street have built on the findings of past organizational studies to better understand the antecedents that influence police officers' commitment to procedural justice Myhill & Bradford, 2013;Schafer, 2013;Tankebe, 2014;Trinkner et al., 2016;Wolfe & Piquero, 2011).
Progress on procedural justice research depends on gaining greater insights into the role that officer perceptions and judgments have in shaping the actions of those officers in the field. As part of this effort, Van Craen (2016b) proposed a work relations framework for police agencies that starts with officer perceptions of the internal procedural justice (IPJ) employed by their supervisors that affect the officer's behavior on the street including the officer's external procedural justice (EPJ). A nascent wave of studies has supported either a direct or indirect IPJ and EPJ relationship in Van Craen's conceptual model (C. M. Donner & Olson, 2019;Kutnjak Ivković et al., 2019;Sun et al., 2018;Sun et al., 2019;Sun et al., 2020;Van Craen & Skogan, 2017;Wu et al., 2017;Wu et al., 2019). The work relations framework has served as a platform for testing a number of mechanisms linking supervisory and officer procedural fairness, such as the role of officer trust in the public, positive and negative emotions, job satisfaction, commitment to community policing, and moral alignment with the public. In addition, other police researchers have identified supplementary mediators, such as compliance with supervisors and policies of the organization, organizational identification, and officers' self-legitimacy Haas et al., 2015;Tankebe & Meško, 2015;Trinkner et al., 2016;Tyler et al., 2007), indicating that the relationship between fair supervision and fair policing is truly multifaceted.
This article introduces the concepts of work engagement and personal initiative into the model of IPJ and EPJ. It tests whether trust in the public and work engagement and personal initiative is part of the antecedent paths leading to just police procedures in Croatia. Past procedural justice studies in politically transitioning or economically developing countries have generally resulted in a weaker role for EPJ in the classic Tylerian legitimacy models (see Jackson et al., 2014;Peacock, 2018;Tankebe, 2008). While Croatia has stood out in the Western Balkans for taking European Union-integration reform steps, it still remains challenged by key indicators of good governance (e.g., the recent decline in World Bank rankings for corruption controls and the legal framework for rule of law; Pere, 2015). A Structural Equation Modeling (SEM) analysis of officer survey data will test whether, in Croatia, procedurally fair policing is linked to procedurally fair supervision and examine whether officers' trust in the public, work engagement, and personal initiative take indirect roles in the association of organizational IPJ with officer EPJ on the street.

Literature Review
Most scholarship on just police procedures begin their literature review with Tom Tyler and his adoption of Thibaut and Walker's (1975) concept of procedural justice. In developing his framework, Van Craen (2016b) went farther back in the organizational justice canon to Thibaut and Kelley's (1959) social exchange theory that views fair treatment by supervisors as part of a larger social exchange in which beneficial personal behavior is an exchangeable resource.
Drawing on contemporary social exchange studies (see Colquitt et al., 2013), Van Craen (2016a, 2016b) developed a work relations framework that models the role that IPJ has on stimulating EPJ. Specifically, Van Craen's framework is based on the expectation that officer attitudes and behavior are influenced, and thus partly shaped, by their supervisors. The model expects that, to some extent, officers will treat the public in a manner they learn from superiors in the organization. It predicts a direct effect of fair supervision on fair policing and indirect influences through officers' trust in citizens, anger and frustration, and job satisfaction and morale (see Figure 1). Prior tests of the framework by policing scholars have demonstrated direct and/or indirect paths from IPJ to EPJ (C. M. Donner & Olson, 2019;Kutnjak Ivković et al., 2019;Sun et al., 2018;Sun et al., 2019;Van Craen & Skogan, 2017;Wu et al., 2017;Wu et al., 2019).

IPJ
In a process that Van Craen (2016a) labeled fair policing from the inside out, his model's antecedent variable IPJ is a measure of the key principles of procedural justice: neutrality (lacking bias/equal treatment), voice (an opportunity to tell their side/view and participate in the solution), respect (treating persons with dignity/acting politely), and accountability (providing explanations for decisions). After 3 decades of studies on how perceptions of fair treatment by police officers tend to have a positive influence on the citizens' evaluation of the police and their behavior toward officers, scholars are now testing whether fair treatment by supervisors could be related to police officers' attitudes and behavior toward supervisors, the organization, and the public. Previous organizational justice research found that fair supervision is directly related to employee trust in their employer and compliance with that organization's rules (e.g., Cohen-Charash & Spector, 2001;Colquitt, 2001;Colquitt et al., 2013;Tyler & Degoey, 1996). Over the last decade, a growing field of policing literature has linked perceptions of fair treatment by supervisors with a variety of desirable goals, including increased job satisfaction, greater trust in the public, lowered job turnover, officer self-legitimacy, and greater compliance with their agency's rules (C. Donner et al., 2015;Haas et al., 2015;Nix & Wolfe, 2016;Tankebe, 2010;Wolfe & Piquero, 2011). At the same time, policing scholars began reporting a direct association between IPJ and EPJ (C. M. Donner & Olson, 2019;Myhill & Bradford, 2013;Tankebe, 2014;Van Craen & Skogan, 2017).

EPJ
The work relations framework views EPJ as the dependent variable in a model of the antecedent set of relationships predicting officer behavior with the public. This differs from the classic Tylerian theory of police legitimacy (Tyler, 1990(Tyler, , 2004) that theorizes that citizen perceptions of (external) procedural justice act as antecedents to public judgments on the police agency and citizens' behavior toward the police. As in the Tylerian model, the dimensions of EPJ in Van Craen's approach include neutrality, voice, respect, and accountability (Van Craen, 2016b;Van Craen & Skogan, 2017). Both theoretical frameworks thus complement each other by dealing with the same subject from two different perspectives (the citizen and officer perspectives, respectively).
Though initially the wave of procedural justice studies suggested that IPJ directly predicts EPJ, several recent studies indicate that the more dominant path for IPJ to influence EPJ is through indirect mechanisms. Trinkner and colleagues (2016) found in the United States that a procedurally just organizational climate did not have a direct effect on officer support for democratic policing. Moreover, Sun and colleagues (2018), Kutnjak Ivković and colleagues (2019), and Wu and colleagues (2019) found a lack of a direct relationship between IPJ and EPJ in tests of the work relations framework in China, Croatia, and Taiwan, respectively. Instead, these studies suggest that police attitudes and emotions, shaped in part by officer perceptions of supervisor fairness, are strong predictors of officer EPJ. Yet, other studies over this same period reported both direct and indirect effects of procedurally just supervision on procedurally just policing Van Craen & Skogan, 2017;Wu et al., 2017).

Trust in the Public
Historically, policing studies have viewed a lack of police trust in the public as a feature of an institutional culture shaped by the hierarchical and authoritarian nature of police organizations (e.g., Paoline, 2003;Walker, 1998), but organizational research increasingly has challenged the inelasticity of this trait. Most notably, scholars have begun to directly measure officer trust in the public (see Carr & Maxwell, 2017;Van Craen, 2016a). Recent quantitative studies have found that officer trust in the public is an influential antecedent to positive policing outcomes (Carr & Maxwell, 2017;Mourtgos et al., 2019). Trusting other persons involves having positive expectations about the words, actions, and decisions of trustees (Colquitt et al., 2013). More specific definitions of trust stress people's belief in the good intentions or goodwill of others (Tyler & Huo, 2002;Uslaner, 2004).
Based on a survey using the same test items for trust in the public, Van Craen and Skogan (2017) found that trust in the public was significantly associated with IPJ and EPJ in the work relations framework. Their study revealed that in the United States officer trust in citizens mediated the IPJ-EPJ relationship. In their test of the work relations framework, colleagues (2018, 2019) found the same latent variable of police trust in the public predicted EPJ in China and Taiwan, but in the case of Taiwan trust in the public was indirectly associated with IPJ. Croatia's communist past and recent progress in developing democratic institutions offer a new test of the role that trust in the public has in predicting EPJ.
Though direct tests of police views on the public are relatively rare, complementary research demonstrates that a procedurally fair organizational climate protects officers against developing cynical and distrusting attitudes toward citizens (Trinkner et al., 2016), and such negative attitudes have been found to decrease officers' inclination to treat citizens in a procedurally fair manner (C. M. Donner & Olson, 2019). With regard to the specific dimension of respect, Porter and Alpert (2017) showed that police recruits who exhibit more cynical attitudes toward the public are more likely to endorse reacting disrespectfully to the public.
The link between procedurally fair supervision and officer trust in citizens is based on the idea that supervisors' behaviors function as important signals to officers about the moral standards of the society in which they work (Van Craen, 2016b). As representatives of law and government, police supervisors are expected to play an exemplary role. If they are not fair and cannot be trusted, it may be interpreted as a cue that nobody can be trusted. Likewise, daily positive experiences with fair and lawful behavior of supervisors contribute to the belief that this is a common type of behavior in society and, consequently, leads officers to develop more trust in citizens. In addition, Van Craen (2016b) has posited that the level of officers' trust in citizens influences officers' level of external procedural fairness. According to him, officers should be inclined to listen to citizens' views and treat them with respect when they have positive expectations about their words and actions. Building on these findings and theoretical explanations, this study tests the assumption that IPJ in a Croatian police department also predicts officer trust in the public which then shapes officer EPJ.

Work Engagement and Personal Initiative
In parallel with trust in the public, this study proposes two constructs that reflect the officer's attitudes toward their police agency and their role on the street. Work engagement is hypothesized as an intermediary variable between IPJ and personal initiative which serves as the precursor to EPJ. In other words, fair treatment by supervisors shapes officer engagement with their work, which then predicts their willingness to take the initiative on the job that includes adopting fair procedures in their interactions with the public.
Work engagement is a positive, work-related state of mind that is characterized by enthusiasm and energy (Salanova & Schaufeli, 2008). Barely a subject of academic study 2 decades ago, Maslach and Leiter's (1997) observation that work engagement represents the antithesis of employee burnout gave rise to thousands of published articles seeking to understand the role of employee engagement in organizations (see Schaufeli & Bakker, 2010). Another construct from organizational behavior studies is personal initiative, which has been defined as conduct characterized as self-starting or proactive in the approach to work (Frese & Fay, 2001).
Though police research on work engagement has been limited to officer well-being studies (Rothmann, 2008), a recent wave of organizational behavior research has examined the IPJ relationship with work engagement. The studies have demonstrated that the four core components of IPJ (neutrality, voice, respect, and accountability) serve as precursors to work engagement across dozens of professions ranging from accountants to public administrators (Al-Shbiel et al., 2018;Karatepe, 2011;Ledimo & Hlongwane, 2014;Ozer et al., 2017). Policing studies have not specifically tested personal initiative as part of procedural justice studies, but organizational behavior scholars have demonstrated that work engagement can predict employee's personal initiative (see Salanova & Schaufeli, 2008;Sonnentag, 2003). Most notably, findings from these studies support work engagement as a mediating variable between IPJ and innovation at work (Agarwal, 2014;Kim & Park, 2017).
Criminal justice researchers have identified officer initiative as a precursor to the organization's fundamental goals (Bazemore & Senjo, 1997;Famega et al., 2005;Mawby & Yarwood, 2011). A condition for such a positive relationship seems to be that officer initiative is a direct result of their degree of engagement with police work. Positive and desired initiatives conveying the values of the organization can contribute to the realization of the agency's goals, including the implementation of a procedural justice approach to policing.
As a country with a former socialist/communist legacy, Croatia is still undertaking the transition toward a more service-oriented police institution that permits considerable autonomy and initiative for officers on the street. In this context, it is particularly relevant to examine and monitor the role that work engagement and personal initiative play in the application of procedural justice principles in officer-citizen encounters. The conceptual model in Figure 2 will serve to test whether these two factors and officer trust in the public mediate the relationship between IPJ and EPJ in Croatia.

Policing in Croatia
Applying the work relations framework on police in a transitioning European country provides an alternative social setting from past tests of the model conducted in North  Though Croatia broke from the former socialist Yugoslavia nearly 30 years ago, the country remains a transitional democracy still reforming the institutions and practices of democratic governance. Freedom House (2017) and World Bank (2017) quality of governance indicators suggest that Croatia made rapid progress in rule of law and institutional effectiveness over the first 2 decades of independence, but that progress has stalled over the last decade.
As a result of the war in the early 1990s, the country quickly replaced the former socialist militia with the Croatian national police agency. As the public perceived that the police were part of the country's defense in that decade's war, the level of support for the police in Croatia in the 1990s was much higher than in a typical transitioning country in Eastern Europe at the time (Kutnjak Ivković, 2000. Over the last 2 decades, the level of support for the police has decreased. For example, in 1995, over 60% of the Croatian respondents on the World Values Survey expressed a lot of confidence in the police, while only about 50% did so on the European Values Survey in 1999. In the 2013 Transparency International Global Corruption Barometer, about one half of the Croatians believed that the police are either "corrupt" or "extremely corrupt" (T. I., 2013), although the police are not perceived to be even in the top five most corrupt public institutions/groups in the country.
In 2003, community policing was officially introduced throughout the national police (Cajner Mraović et al., 2003, p. 11). Community-policing reform included the idea that the police should become a public service  and that an increased emphasis should be put on proactive policing and crime prevention (K. Borovec & Kutnjak Ivković, 2013;Cajner Mraović et al., 2003). The reform also required the organization of crime prevention in local communities, changes to the police-public communication, and development of community and police partnerships (K. Borovec & Kutnjak Ivković, 2013;Cajner Mraović et al., 2003).
In 2011, the Ministry of the Interior (K. Borovec et al., 2011) published the results about their study on internal communication within the Ministry. The vertical communication (i.e., communication with the supervisors) was evaluated in positive terms: Over two thirds of the respondents in a stratified representative sample stated that they were satisfied with the supervisor availability, their familiarity with, and understanding of, the subordinates' work-related problems, and the supervisors' recognition of the subordinates' potential. In a 2016 survey of a representative sample of the Croatian police officers (L. Borovec et al., 2018;Cajner Mraović & Faber, 2016), a substantial proportion of police officers-about one third-expressed low levels of confidence in the supervisors. The respondents' level of confidence seemed to be negatively related to the supervisors' position in the hierarchy: The respondents had the most confidence in the supervisors at their local police station and the least confidence in the supervisors at the Ministry level (L. Borovec et al., 2018). Furthermore, the majority of the respondents also expressed their dissatisfaction with the (lack of) opportunities to participate in the selection of work tasks and in the analyses of police effectiveness, thus signaling serious issues with voice, one of the four key principles of IPJ (e.g., Van Craen, 2016a).

Sample
This study is based on a survey collected in different Zagreb sites in 2016, when members of the research team distributed questionnaires to police officers from the Zagreb Police Administration at the beginning of their shifts. Of the 700 questionnaires distributed to police officers, a total of 638 were completely filled out, leading to a response rate of 91%. This high response rate is in line with previous studies of the Croatian police (see, e.g. Kutnjak Ivković, 2015) as well as a number of officer surveys on procedural justice conducted in other countries (e.g., Haas et al., 2015;Jonathan-Zamir & Harpaz, 2018;Wu et al., 2017;Wu et al., 2019).
From a population of more than 3,130 police officers assigned to the Zagreb Police Administration, 638 officers completed the study's questionnaire. The sampled officers represented six of eight of the city of Zagreb's police stations and seven of nine police stations in Zagreb County. As the distribution of sworn police officers across different regions/cities is not publicly available, we have no means of comparing the characteristics of our sample with the demographics of the population of police officers in Croatia.
Most of our respondents are male (81%), between the ages of 25 and 40 (52%), and Croats by ethnicity (98%). About two thirds (68%) had a high-school diploma, while the rest had a college degree or more advanced degrees. The majority of the respondents (52%) are quite experienced police officers with 15 or more years of service. The respondents were mostly assigned to patrol (28.7%), investigations (20.4%), or border protection (18.5%). Those with supervisory responsibilities composed roughly one fifth (19.0%) of the sample.

Questionnaire and Measures
The questionnaire was developed as part of an international comparative project intended to assess the role of IPJ and EPJ across the globe (see: Haas et al., 2015;Kutnjak Ivković et al., 2019;Sun et al., 2018;Sun et al., 2019;Van Craen & Skogan, 2017;Wu et al., 2017;Wu et al., 2019). Scholars working in different countries made minor changes in the survey's test items, particularly following pretests, to accommodate distinct cultural understandings of the questions, but the overall survey provided standardized measures across the different countries. This also took place in Croatia, where the questions were translated into Croatian and then translated back into English by the authors. After the questionnaire was pretested with the Zagreb police, it was distributed with minor adjustments to collect the full sample (see Kutnjak Ivković et al., 2019).
The first antecedent variable in the study is IPJ. This latent variable is inferred from four items (neutrality: my colleagues and I have been treated equally by our supervisors; respect: my supervisors have no respect for their officers; voice: my supervisors show interest in what we have to say; and accountability: when implementing changes, my supervisors explain why these changes are necessary). The respondents selected one answer for each of the four questions on a 6-point scale from strongly agree to strongly disagree. The answers for these items, with the exception of respect, were reverse-coded.
The second antecedent variable in the study is work engagement. This latent variable is based on two questions (Enthusiasm 1: I go to work with enthusiasm; Enthusiasm 2: I am motivated when I go to work). The respondents selected one answer for each question from a 6-point scale from strongly agree to strongly disagree. The answers for the two items were reverse-coded in the multivariate analyses.
A hypothesized predictor of EPJ is personal initiative. This latent variable is formed based on three items (Initiative 1: I take the initiative to work with my supervisors; Initiative 2: I help my supervisors without being asked to do so; and Initiative 3: I give information to my supervisors without being asked to do so). The answers to the three questions were given on a 6-point scale from never to very often.
Another hypothesized predictor of EPJ is trust in public. The latent variable is inferred from the answers to four questions (Tr1: most citizens have good intentions; Tr2: I trust that most citizens will behave; Tr3: citizens can mostly be trusted to do the right thing; and Tr4: generally speaking, most citizens are on the side of the law in terms of what is good and what is bad). The respondents selected one answer on all four questions from a 6-point scale from strongly agree to strongly disagree. The answers for the first three items were reverse-coded in the multivariate analyses.
The outcome or dependent variable, EPJ, is based on four pairs of items. Two questions form a neutrality index (ext_neutrality: I reprimand a person for a particular behavior, while another for the same behavior I look the other way [reverse-coded]; it makes little sense that police seek to be impartial because that is impossible). Two questions form the respect index (ext_respect: I behave toward citizens roughly [reverse-coded]; police officers must treat everyone respectfully, regardless of their behavior). Voice is measured with additional items (ext_voice: I am making efforts to listen to citizen suggestions about the issues that need to be resolved; I find time to listen to citizen suggestions about improvements in safety and living conditions). Finally, accountability is measured as the fourth dimension (ext_account: I explain to citizens why the police see to it that he or she abides by the laws and rules; I explain to citizens why some actions and measures are needed).
The second item for neutrality and second item for respect provided answers on a 6-point scale from strongly agree to strongly disagree, while the other six items were on a 6-point scale from never to very often.

Univariate Statistics Analysis
To gain a better understanding of Croatian officer perceptions and judgments in relation to the factors used in the model, we first examine the distributions for each of the items. In terms of the antecedent variable of IPJ, most Zagreb officers appear to view their supervisors as demonstrating fair procedures, but they question the neutrality of their superiors. Specifically, more than 70% disagreed with the statement that supervisors do not respect their officers (respect); Nearly 60% thought supervisors show an interest in what officers have to say (voice); and more than 64% agree that, when implementing changes, supervisors explain why these changes are necessary (accountability). This expectation of procedural fairness was offset by the more limited number (less than 42%) of respondents who thought that officers were treated equally by supervisors (neutrality).
In regard to officer engagement to the organization, the majority of Zagreb police officers reported that they go to work with enthusiasm (60%) and were motivated to work (55%). A larger portion of officers in Zagreb were positive in their judgments on the public. More than 72% thought most citizens had good intentions (tr1) and that they could trust citizens to behave (tr2). Moreover, 85% of the officers viewed most citizens on the right side of the law (tr4). The least supported statement was for item tr3, in which only 60% of officers agreed that citizens can mostly be trusted to do the right thing.
Despite a majority of officers suggesting they go to work with enthusiasm, only modest portions of them were willing to take the initiative in their workplace. About 51% responded that they almost never take the initiative in working with their supervisors, and more than 41% never help their supervisors or keep them informed on issues unless specifically tasked. Less than 3.5% of the officers reported often taking the initiative in their work across the latter two items.
The police in Zagreb do not appear to excel at treating the public procedurally fair as a significant minority of the officers admit to being biased. Specifically, 43% reported that they reprimand one person for a particular behavior and looked the other way for others; while more than 38% agreed with the statement that it makes little sense for the police seek to be impartial, because that is impossible. In terms of respect, more than 40% admitted to sometimes treating citizens roughly, and 27% disagreed with the statement that police officers must treat everyone respectfully regardless of their behavior.
The officers were more supportive of the last two components of EPJ. Even as most officers were favorable to allowing citizens to have a voice in interactions, just over 10% reported that they rarely make an effort or find time to listen to citizen suggestions. In the accountability index, 19% of officers stated that they rarely explain to citizens why the police see to it that they abide by the laws and rules, while a similar 20% of the officers suggested that they rarely explain to citizens why some actions and measures are needed.
The last step in the univariate analysis was to examine for normality the distribution of each item used in the model's five scales. As summarized in Table 1, the skewness results across the variables were within the range expected for univariate normality, but the Kurtosis coefficient for a majority of variables (11 of 21) in the model was outside the range (+3) generally viewed as adequate (Onwuegbuzie & Daniel, 2002). In order to address this normality concern with the data, the study's factor analysis uses the Satorra-Bentler estimator that does not require normality across the data.

Multivariate Statistical Analysis
In support of the newly hypothesized engagement/initiative role in the model, an exploratory factor analysis (EFA) was run on the Croatian sample (see Table 2). The EFA extracted, with the assistance of an oblique (Promax) rotation, five factors from the 21 items which correspond to the five hypothesized variables (IPJ, trust in public, work engagement, personal initiative, and EPJ) with no indication of cross-loadings across these five latent variables. The EFA did demonstrate a few items loaded weakly: the test item Respect on the IPJ factor and the test item Neutrality on the EPJ factor. The study pursued an EFA because of the new relationships being tested in the model, but the two weak test items were part of the two conventional procedural justice variables. After a full SEM analysis found that the absence of either factor did not change the SEM results, the decision was made to apply the rule of thumb of keeping factors loading above 0.40 and dropping those two test items from the analysis (see Cabrera-Nguyen, 2010). In testing for measurement reliability, Cronbach's a scores were calculated for each of the latent variables. The satisfactory Cronbach's a scores for each of the latent variables and the discriminant validity demonstrated in extracting a matching set of factors suggest that the five-factor model fits the data.
After finding an EFA supportive of the five separate factors hypothesized for the model, a confirmatory factor analysis (CFA) was conducted to further validate the measures. As demonstrated by the measurement model in Figure 3, the CFA goodness-of-fit tests indicate that a fivefactor model fits the data well (model fit statistics: w 2 _SB ¼ 159.25, df ¼ 83, p < .001; RMSEA_SB ¼ .032; CFI_SB ¼ .976; and TLI_SB ¼ .970).
Finally, we used path analysis to test the hypothesized direct and indirect relationships between IPJ, trust in public, work engagement, personal initiative, and the outcome variable EPJ. STATA Version 15 SEM Builder was used to develop the path model (Figure 4). The goodness-of-fit measures for this path analysis provide a strong fit of the sample to the model (model w 2 _SB ¼ 186.30, df ¼ 113, p < .001; RMSEA_SB ¼ .026; CFI_SB ¼ .984; and TLI ¼ .980).
The SEM analysis of the Croatian data does not support the expectation that officer perceptions of the fairness of their supervisors' procedures directly predict EPJ. In the SEM model, IPJ is not significantly correlated with EPJ. Instead, our analysis supports the mediating role of trust in the public that was found to be significantly associated with IPJ and EPJ. Specifically, the latent variable IPJ had a moderate positive association with trust in public (b ¼ .21, p < .01) that had a moderate positive association with EPJ (b ¼ .18, p < .01). These results suggest that fair supervision influences fair policing indirectly, by fostering officers' trust in citizens. The SEM model also supports the hypothesized work engagement and personal initiative path. First, IPJ had a strong positive association with work engagement (b ¼ .53, p < .001). Second, work engagement had a moderate positive association with personal initiative (b ¼ .19, p < .01). The final segment of the path was the moderate positive association between personal initiative and EPJ (b ¼ .24, p < .01). Fair supervision thus also seems to affect fair policing by increasing officers' engagement and initiative. The study also tested whether work engagement directly predicted EPJ, but the lack of a significant association supported the theoretical role of work engagement as an intermediary variable.

Discussion
In this article, we assessed links between fair supervision and fair policing in Croatia, a European country that is still developing practices of good governance. The findings of this study can be helpful for other transitioning democracies and advance cumulative scientific knowledge. Our results do not support the hypothesis that officer perceptions of the fairness of their supervisors' procedures directly predict EPJ. This finding is in line with previous studies in China, Taiwan, and the United States where researchers found the indirect mechanisms were either the sole or dominant path between IPJ and EPJ (Sun et al., 2018;Trinkner et al., 2016;Wu et al., 2019). Although there are studies that have found a substantial direct effect of fair supervision on fair policing (C. M. Donner & Olson, 2019;Myhill & Bradford, 2013;Tankebe, 2014;Van Craen & Skogan, 2017), in Croatia, the indirect links seem more important. In parallel with recent tests of the work relations model demonstrating the role of mediating factors in the IPJ-EPJ relationship (such as moral alignment, job satisfaction, and anger), the Zagreb study supported the role of indirect paths for supervisory justice to impact officer behavior on the street.
The role of trust in the public as a significant path between IPJ and EPJ in Croatia supports prior research. Fair treatment by supervisors leads officers to have more positive expectations about people in general, which in turn makes them more inclined to act procedurally just in encounters with citizens. This finding draws our attention to the importance of positive contacts to counter the more inimical interactions with criminal suspects and others not interested in engaging with the police.
Police culture research has shown that frequent contacts with criminals and other law violators may induce police officers to develop a pessimistic view of their social world (Loftus, 2010; , 1974). Such a view is often characterized by distrust, cynicism, and suspicion toward the public. When officers' behavior is guided by these attitudes, it is less likely that they will endeavor to apply procedural justice principles in their dealings with citizens. Yet, positive experiences and relationships, like procedurally fair supervision, lead officers to have a more optimistic view of their social environment and thus seem to have the potential to mitigate the negative effects of encounters with disobeying citizens. Our study suggests that IPJ can be a useful part of a broader organizational strategy to foster trust and EPJ in the ranks.

Maanen
The findings in Croatia not only supported past research on trust-based variables between IPJ and EPJ, our SEM analysis also found an alternative path between IPJ and EPJ through officer work engagement and personal initiative. Though still a nascent path in the organizational behavior literature (Lopez-Cabarcos et al., 2015;Moon et al., 2008), the engagement/initiative mechanism may have a particular resonance in law enforcement work.
Internal procedural fairness, for instance, can enhance external procedural fairness by increasing officers' support for and compliance with organizational policies, including use of force policies (Haas et al., 2015;Van Craen and Skogan, 2017). Our study shows that procedurally just supervisory practices also encourage fair policing by increasing officers' work engagement and personal initiative. In line with Haas et al.'s findings in Argentina, the results of our research in Croatia indicate that for countries that are transitioning to community and procedurally fair policing, fair supervision is an approach that can facilitate this reform process.
A similar mechanism may also be at work in countries that have already made more progress in this respect. Surveys in countries like United States and UK have shown that the level of organizations'/supervisors' procedural justice determines the degree to which officers are satisfied with their job/organization as a place to work (Myhill & Bradford, 2013;Wolfe et al., 2018), the level of their motivation at work (Bullock & Garland, 2019;Nix & Wolfe, 2016), and the extent to which they engage in extra-role activities . Moreover, her colleagues (2017, 2019) have demonstrated that in both democratic and communist Asian countries, officers' job satisfaction and morale mediate the relationship between IPJ and EPJ.
When officers' morale and job satisfaction are low because of unfair supervision, they will not be inclined to make many efforts. They will only do the minimum (Reynolds et al., 2018). It is thus unlikely that demoralized officers will spend much time listening to citizens' views or be interested in people's suggestions to tackle disorder and crime problems. Nor will they endeavor to be accountable. We note that our work engagement variable is closely related to the motivation and morale variables that have been included in these previous studies and that our concept of personal initiative shows similarities with the concept of extra-role behavior. This suggests that, although researchers have adopted different terms, a similar set of mechanisms have been identified that partly explains the link between IPJ and EPJ in many countries around the world.
As part of the transition to democratic policing, Croatian officers have received training in the principles of community policing (see Kutnjak Ivković et al., 2019). Nevertheless, officers socialized in interacting with the public through the traditional tools of command and coercion need more than training alone to change to a paradigm built around respectful dialogue. This study's findings of a work engagement and personal initiative mechanism between IPJ and EPJ suggest that fair supervision may also play a critical role in forming committed officers, willing to stick their necks out via personal initiative, to pursue procedural justice in their interactions with the public.
In another analysis of this study's Croatian police data, Kutnjak Ivković and colleagues (2019) demonstrated that a variable measuring officer community policing values is a strong mediating mechanism between IPJ and EPJ in a test of Van Craen's model. This antecedent path to EPJ is in line with the current study's finding that EPJ is influenced by officer attitudes toward their work in the street. As with community policing values, the work engagement and personal initiative mechanism appear to reflect officer willingness to go beyond simply doing their job but also making special efforts to serve the public and meet the agency goals in their interactions with the public.

Policy Implications
Our research indicates that procedurally fair supervision may be a useful vehicle for former socialist countries and other transitional states across the globe to transform the police into a more service-oriented and public-friendly government agency. Though lacking a direct association between IPJ and EPJ, internal procedural fairness can facilitate the change toward more fairness in policing by fostering attitudes that in many transitioning countries are not self-evident. Just supervisory practices are likely to increase officers' levels of work engagement, personal initiative, and trust in the public. Therefore, fair supervision helps to improve strained relationships between officers and police leaders and between the police and communities. That is precisely the policy goals that many transitional countries are seeking to achieve.
Certainly, just supervision starts at the top with leadership that treats mid-management with analogous procedures that reflect the principles of neutrality, respect, voice, and accountability. For countries that are in a process of democratization, a fair policing from the inside out approach thus will require a radical change of the organizational culture and model. Police chiefs will have to adopt participative and transactional leadership styles instead of operating as an extension of authoritarian political leaders. They will also have to cultivate values like respect and equality in their management practices. By practicing what they preach, police chiefs can play an important role in the reshaping of the set of assumptions, values, and norms that guide supervisors' behaviors.

Limitations
This study uses the broad definition of IPJ common in criminal justice studies and at the heart of the "fair policing from inside out" framework. Future studies will benefit from testing the same model by using multidimensional measures of IPJ which reflect the strongest of these subindexes.
The study is based on a convenience sample contributing cross-sectional data. The analysis assumes that fairness of supervision precedes the other mediating variables in time, but crosssectional data do not offer the prospect of testing the direction of causality. We argued, for instance, that IPJ fosters officer engagement, but work engagement also may precede officers getting treated with more respect and receiving more opportunities to give input. We further note that the study's findings cannot be generalized from the Zagreb district to the entire country of Croatia.
The study did not observe officers' behavior toward citizens but measured officers' self-reported treatment of the public. This means that a social desirability bias may have influenced some answers. Yet, by examining officers' behavior, this study advances cumulative scientific knowledge, as most studies on this topic have focused on officers' support for fair policing. Certainly, future studies should incorporate mixed methods to facilitate observed behavior and longitudinal data to assess the key relationships over time (see Mastrofski et al., 2010).
As a society that is still transitioning to a fully democratic and rule-of-law state, the use of a questionnaire designed for Western democracies in Croatia could be questioned. A wave of studies of public perceptions of procedural justice in developing countries has demonstrated a weaker role for EPJ in the classic Tylerian model of police legitimacy (see Peacock, 2018). Though few current Croatian police officers served in either the socialist era or the war period in the 1990s, a legacy of prioritizing state security over public service likely still influences the institution's organizational culture. Organizational justice models, such as the work relations framework, should serve as a means of tracking changes in officer judgments on their supervisors and the impact that has on levels of trust and the values they express on the street.