The role of stigma and uncertainty in moderating the effect of procedural justice on cooperation and resistance in police encounters with persons with mental illnesses

Abstract

Within social psychology, procedural justice theory has been used to understand variations in compliance with legal authorities such as police. Thus, it may help explain variation in cooperation and compliance in encounters between police officers and people with serious mental illness (SMI), which are often fraught with difficulty and risk. In this paper, we examine the extent to which perceptions of procedural justice among persons with SMI are associated with self-reported levels of cooperation and resistance in encounters with police. We also examine stigma and encounter type as potential moderators of the procedural justice effect. 154 persons with serious mental illness who reported a police contact within the past year were interviewed using the newly developed Police Contact Experience Survey (PCES), which includes questions about the characteristics of the contact, perceived procedural justice (PPJ) and degree of cooperation and resistance. Participants also completed the Link Perceived Devaluation and Discrimination Scale (PDS). Findings suggest that greater PPJ is associated with more cooperation and less resistance. The effect on cooperation, however, is moderated by both perceived stigma and the type of encounter. The direct effect of perceived stigma (PDS) on cooperation was unexpected, with higher perceived stigma associated with greater cooperation. Findings underline the importance of both procedurally just treatment in police interactions with vulnerable individuals and further efforts to reduce the stigma of mental illness.

Police officers, as often the only 24/7 resource for responding to mental health crisis, have been dubbed “streetcorner psychiatrists” (Teplin & Pruett, 1992). Studies estimate that 10% of all police calls involve people with mental illnesses (Deane, Steadman, Borum, Veysey & Morrissey, 1999)-as complainants, criminal offenders, or persons needing emergency psychiatric evaluation-yet standard police training prepares officers poorly for how to manage these encounters. Additionally, estimates of the prevalence of serious mental illness in jails and prisons range from six to 17% (Council of State Governments, 2002; Ditton, 1999; Teplin, 1990). Compared to the five to seven percent prevalence estimates for the general adult population in the United States (Kessler et al., 1996; Kessler et al., 2004), this suggests that persons with serious mental illnesses are over-represented in these settings. Thus, improving how police respond to this population is a priority, due to the gatekeeper role they play to both the criminal justice and mental health systems.

From the perspective of police officers, these encounters are problematic. Officers report persons with mental illnesses do not respond well to traditional policing tactics (Engel, Sobol, & Worden, 2000), and as such, encounters with persons with mental illnesses are more time-consuming than other types of calls and may be more likely to result in injuries to both the officer and subject (Cordner, 2006). Officers indicate that they lack adequate skills and resources to effectively respond (Bittner, 1967) and express interest in additional training (Vermette, Pinals & Appelbaum, 2005).

Approaches are being adopted nationwide to educate police officers about mental illness and train them in more effective methods for managing encounters with persons with mental illnesses. The logic underpinning many of these programs is that calm and patient handling of these encounters is believed to prevent unnecessary use of force and reduce injuries to both officers and citizens. The Crisis Intervention Team (CIT) model is the most well known of these approaches. This model includes specialized training of police officers, collaboration with community stakeholders, and organizational shifts in policy and procedures. While evidence of the effectiveness of the CIT model is just beginning to emerge (Compton, Broussard, Munitz, Oliva, & Watson, 2011), it has been widely disseminated-with over 1,000 jurisdictions in the United States implementing some version of the model. CIT and other programs designed to train officers to safely and effectively manage these encounters would benefit from a better understanding of factors that influence cooperation and resistance on the part of persons with mental illnesses.

Within social psychology, procedural justice theory has been used to understand variations in compliance with legal authorities such as police, and we have argued that this line of research offers a fruitful avenue for explaining why specialized police response tactics may further cooperation and compliance among individuals with mental illnesses (Watson & Angell, 2007). Building on earlier research by Thibaut and Walker, Tyler and colleagues developed the group engagement model of procedural justice to explain the importance of just treatment in interactions between authorities and subordinates (e.g., employers and employees, legal authorities and citizens) (Tyler & Blader, 2003). This formulation of procedural justice theory posits that cooperation with authority figures will be maximized when individuals feel that they have been treated fairly, given an opportunity for voice, and have been afforded dignity and respect by the authority figure. When authorities interact with subordinates in a procedurally just manner, it makes less difference to the subordinates whether they receive an equal share of resources or achieve the most favorable outcome. The importance of procedural justice, according to this perspective, is that it communicates identity-relevant information about an individual’s worth and status within the context of a group, which overrides more instrumental concerns such as maximizing one’s self-interested gain. In previous work, we tested this relationship in a sample of people with serious mental illness who reported a recent encounter with a police officer: consistent with procedural justice theory, we found that those who saw the officer as having treated them with fairness and respect were more likely to report greater cooperation and less resistance in their response to the officer (Watson, Angell, Vidalon, & Davis, 2010). Here, we extend this work and test a multivariate model and examine potential moderators of the effect of procedural justice on cooperation and resistance.

A large body of empirical work testing the group engagement model suggests strong support for its precepts, but also indicates the presence of boundary conditions within and outside of which procedural justice may be more or less influential. Of particular relevance to individuals diagnosed with psychiatric disorders is the boundary condition of belonging to a stigmatized group, which has well-established deleterious effects on the functioning of people diagnosed with mental disorders (Livingston & Boyd, 2010). We argue that stigma may likewise attenuate procedural justice effects because the subject, to the extent that he or she feels alienated from society, is less motivated to attend to treatment by authorities in order validate his or her membership in society. In contrast, people may be more motivated to attend to how they are treated under conditions of uncertainty (Van den Bos, 2001) and fear (De Cremer & van Hiel, 2008). They may feel particularly vulnerable in police encounters in which their personal liberty is at risk, for example when they have committed a crime and may be arrested or when they are experiencing a mental health crisis and may be involuntarily transported to the hospital for evaluation. In this paper, we test the moderating effects of perceived stigma and call type on the relationship between perceived procedural justice and cooperation and resistance in self-reported encounters with police among citizens with mental illnesses.

Procedural Justice and Police Encounters

The group engagement model of procedural justice is an element of a larger theoretical model put forward by Tom R. Tyler (2011) in an effort to account for social motivations for voluntary cooperation by members of society in their interactions with organizations and institutions, such as the law and the workplace. The overarching framework stresses that while incentives and sanctions promote compliant behavior, an even more effective and humane approach would be to shape voluntary cooperation by increasing citizens’ identification and sense of obligation to the social groups to which they belong. The procedural justice strand of this framework posits that citizens will accept the decisions of authorities – even if their desired outcomes are not upheld – and act in ways that depart from self-interest (i.e. cooperate), when they perceive that the authorities have treated them in a fair and respectful manner. In regard to police encounters, Tyler and Huo (2002) conceptualize the application of procedural justice as a strategy of process-based policing in which officers encourage citizens to develop a “‘reservoir of goodwill’ toward police” (p. 132) by behaving respectfully toward them and using fair procedures to render decisions. In turn, citizens will comply with laws without the need for threat of sanction if they trust that the motives of authorities are impartial and unbiased.

A number of studies have examined the effect of procedurally fair or just treatment by police on citizen behaviors, both within the immediate encounter and beyond. Survey research on citizen interactions with police has found that when the police are perceived as exercising their authority in a procedurally just manner, people are more likely to obey, cooperate and accept officers’ decisions (Sunshine & Tyler, 2003; Tyler & Blader, 2000). Findings from a study of recidivism among spouse batterers indicate that perceptions of procedural justice and fair treatment by the police decreased the likelihood of subsequent spousal assaults (Paternoster, Brame, Bachman, & Sherman, 1997) suggesting that how officers treat citizens has both an immediate and long-term effect (Tyler, 2004).

Observational studies of police-citizen encounters have found that non threatening verbal behavior on the part of police reduces citizen defiance, while threatening verbal and physical behavior increases citizen defiance (Piquero & Bouffard, 2003); and disrespectful or procedurally unjust behavior on the part of police reduces compliance with police requests (Mastrofski, Snipes & Supina, 1996; McCluskey, 2003; McCluskey, Mastrofski & Parks, 1999). One study (McCluskey, 2003) found this effect to be stronger when the subject is in some way “irrational” (defined in the study as impaired by substance use, mental illness, etc).

Recently, Dai, Frank and Sun (2011) examined the influence of procedurally fair behavior by police on two important outcomes of policing: citizen disrespect (hostile body language and verbal expressions) and citizen noncompliance (failure to comply or indicate willingness to comply). They found that citizen disrespect and noncompliance were predicted by different elements of the quality of interpersonal treatment and decision-making by police. Police disrespect towards the citizen increased and police use of force decreased disrespectful demeanor on the part of the citizen towards the officer. Allowing the person “voice” reduced noncompliance with officer requests.

Boundary conditions influencing the impact of procedural justice

While empirical support for the group engagement model of procedural justice in police encounters is strong, research on procedural justice in other settings nonetheless indicates the presence of boundary conditions within and outside of which procedural justice may be more or less influential. These conditions relate to both characteristics of situations and individuals are a likely relevant to police encounters with persons with mental illnesses.

Situational uncertainty

Van den Bos and colleagues (1998, 2001) suggest that people pay the most attention to procedurally just treatment under conditions of uncertainty about whether an authority will exploit them or not. When people have little information regarding whether they can trust an authority, they rely on the fairness heuristic to make judgments (Van den Bos, Wilke, & Lind, 1998). In these situations, the fairness with which they are treated determines people’s affective reactions (van den Bos, 2001). Building on the work van den Bos and colleagues, De Cremer and and Van Hiel (2008) manipulated uncertainty and certainty eliciting emotions (fear and disgust) as a proxies for uncertainty and certainty. They found that the effects of procedural justice on self esteem were more pronounced in the uncertainty (fear) condition. This suggests that procedural justice matters most to people when they are trying to deal with situations that make them feel vulnerable and uncertain.

Individual characteristics

Studies examining relational models of procedural justice also suggest that characteristics of the individuals involved are important. This literature indicates that the effects of procedural justice are strongest when the authority is a member of the same group (Tyler, Boekmann, Smith, & Huo, 1997) and when the person feels highly identified with the group (Huo, Smith, Tyler, & Lind, 1996). Here, the relevant group may vary by context and include racial or ethnic group, socio-economic class, occupational group, organization membership, or mainstream society. Building further on this work, Van Prooijen, van den Bos, and Wilke (2004) suggest that the level of group inclusion/exclusion by peers has important consequences to how people respond to procedural fairness. They conducted a series of studies that suggest that people who experience high levels of social inclusion (both generally and as manipulated by investigators) are more sensitive to the fairness of procedures compared to those who experience low levels of social inclusion. As discussed below, the stigma related to mental illness may create barriers to social inclusion for persons with mental illnesses and thus influence their response to procedurally just or unjust treatment.

Stigma as a Moderator of Procedural Justice Effects

We suggest that stigma may act as a unique boundary condition for procedural justice effects among people with mental illnesses. Stigma has been widely studied in relation to mental illness, primarily within sociology and social psychology (Phelan, Link, & Dovidio, 2008; Yang et al., 2007), undergirded by Goffman’s (1963) conceptualization of stigma as an attribute that becomes discrediting to its bearer through a social process of labeling and symbolic representation of human differences. Through social interaction, averred Goffman, certain characteristics come to convey a devalued social identity which reduces the standing of the affected individual.

While one strand of stigma research focuses upon the agents who stigmatize, the other focuses upon the reactions of stigmatized persons themselves. Corrigan and Watson (2002) refer to these strands, respectively, as public and self stigma. Survey research on public stigma related to mental illnesses suggests that despite increases in public knowledge about the types and likely causes of these disorders over the past 50 years, there is continued endorsement of negative public stereotypes, such as dangerousness and incompetence (Phelan, Link, Stueve, & Pescosolido, 2000; Pescosolido, Monahan, Link, Stueve, & Kikuzawa, 1999; Pescosolido, Martin, Long, Medina, Phelan, & Link, 2010). These negative attitudes about mental illness are in turn related to fear and preferences for social distance (Link, Phelan, Bresnahan, Stueve, & Pescosolido, 1999; Pescosolido, et al., 2010), perpetuating social exclusion of persons with mental illnesses.

Research on self-stigma, also referred to as internalized stigma, focuses upon how negative images and stereotypes about mental illness are applied by the bearer to him-or herself, resulting in problematic outcomes for the individual (Corrigan & Watson, 2002; Livingston & Boyd, 2010). When powerful others engage in the process of stigmatization, they possess the capacity to deny opportunities for the stigmatized person to participate fully in community life and citizenship (Link & Phelan, 2001). Concomitantly, the stigmatized individual may respond to the shame induced by the stigma process by withdrawing from social participation. This withdrawal may simply serve to protect the stigmatized person from social rejection. However, those with more fully internalized negative conceptions of mental illnesses may withdraw because they feel they are not capable or worthy of social opportunities. Some research suggests that these individuals also find comfort in limiting their association to social others who know of and accept their conditions – categorized by Goffman (1963) as the “own” and the “wise” -- such as their families and other people who have mental illnesses (Estroff, 1985; Herman, 1987; Angell, 2003; Mandiberg, 1999). This strategy confers protective effects on self-esteem but nonetheless may continue the vicious cycle of “otherness” or separation from society and further the public’s desire to avoid members of the stigmatized group.

Internalized stigma has been measured in a variety of ways, ranging from an awareness or consciousness of negative public attitudes to more direct measures of endorsement of negative attitudes toward the self (Livingston & Boyd, 2010). Regardless of the measure used, the construct of internalized stigma includes the degree to which the person believes that the public devalues people with mental illness and desires to minimize contact with him or her. We argue that this stigma awareness, even if it is not accompanied by agreement or endorsement of the negative self-content, signals the perception of the individual as feeling like an outsider to the social mainstream. In this sense, stigma may modify the process of social cooperation envisioned by procedural justice theory. That is, if the effectiveness of procedurally just treatment rests upon the individual’s sense of solidarity with the social group (e.g., of law-abiding citizens), persons who feel highly stigmatized may be less sensitive to its effects. Such an effect would have important implications, for to the extent that officers cannot influence cooperation by treating such individuals fairly and respectfully, the potential for dehumanizing or forceful treatment could increase.

Using data from a study in which people with mental illnesses reported on an encounter with police officers, we test the proposition that the effects of procedurally just treatment on reactions to police officers (cooperation and resistance) will be weakened when the individual possesses a stronger sense of stigma awareness.

Uncertainty as a Moderator of Procedural Justice Effects

We also examine situational uncertainty as a moderator of the effects of procedurally just treatment on cooperation and resistance. In different types of encounters, persons with mental illnesses may feel more or less vulnerable and uncertain. For example, in calls involving mental health crises or suspicion of significant law violations, their personal liberty is at stake. Whereas in calls involving street stops for identification checks or minor nuisance offenses, or when the subject has called the police for assistance, they are in less danger of being taken into custody. If conditions of uncertainty influence peoples’ attentiveness to procedurally just treatment (Van den Bos et al., 1998, 2001) then we would expect that procedural justice would have stronger effects on cooperation and resistance for subjects of encounters involving a mental health crisis or a law violation compared to other call types.

Methods

The analysis reported here is of data from the final phase of a mixed methods study of the experiences of persons with mental illnesses in encounters with police (Watson, Angell, Morabito, & Robinson, 2008; Watson, Angell, Vidalon, & Davis, 2010). The study involved the development of the Police Contact Experience Survey (PCES) and validation for the Perceived Procedural Justice (PPJ) and Perceived Coercion (PC) subscales contained within. The full PCES was designed to cover the following constructs: perceived procedural justice and coercion; positive and negative pressures, emotional impact, cooperation/compliance; decision acceptance; overall satisfaction; legitimacy; outcome satisfaction; long term compliance with the law; and treatment participation. All items were rated by participants on a four point Likert-like scale with higher scores indicating greater agreement.

Participant Recruitment

Participants were recruited from three psychosocial rehabilitation programs operated by Thresholds, Inc., a community mental health center serving multiple neighborhoods in Chicago. Thresholds is Illinois’ largest psychiatric rehabilitation center, serving over 4,000 people with mental illness annually at its twenty two service locations and more than 40 housing developments in the Chicago area. Eligibility criteria were uniform for all phases of the study and included having had an encounter (any type) with the police within the past 12 months; receiving services at Thresholds South, the Dincin Center, or Austin Apartments; having a non-substance use Axis I diagnosis and being over 18 years old.

Participants were recruited via flyers posted in common areas at each site that listed a phone number to call to schedule interview appointments. Interviewers were also regularly approached onsite by individuals who had heard about the study and were interested. All interviews were conducted in person at the program sites. The interview protocol consisted of the PCES, the UCLA Extended Version of the Brief Psychiatric Rating Scale (Ventura, Nuechterlein, Subotinik, Gutkind, & Gilbert, 2000), Insight & Treatment Attitudes Scale (McEvoy, Freter, Everett & Geller, 1989), the Perceived Devaluation and Discrimination Scale (PDS) (Link. 1987; 2001) and questions eliciting demographic information. The order of administration was counterbalanced to control for order effects. Participants were paid $25 for the interview. All procedures were approved by the Behavioral and Social Sciences Institutional Review Board at the University of Illinois at Chicago.

Dependent Variables

Compliance and resistance

In his study of procedural justice and compliance with police officer requests, McCluskey (2003) measured compliance at two time points during the interaction, at the initial request for compliance made by the officer and at the conclusion of the encounter (e.g. whether the individual was in compliance with all of the officers stated demands). While McCluskey used measures of citizen behavior, Tyler and Huo (2002), have used a single self report item (“I did what I was asked to do”) to measure citizen compliance with police officer requests. We use a combination of these measures that uses the Tyler and Huo item in reference to initial compliance (“I did what the officer asked me to do right away”) and two additional items about compliance by the end of the interaction and future compliance with police officers. The Cronbach’s alpha (.75) for the three item scale indicates acceptable scale reliability. Resistance is measured with three items developed for this study in which participants rated the extent to which they argued, struggled, and were agitated during the encounter. The alpha for this scale (.67) indicates minimally adequate scale reliability.

Control Variables

Three control variables were included in all models: participant age (centered at zero by subtracting the mean), participant race (0=white, 1=nonwhite), and participant gender (0=female, 1-male).

Independent Variables

Procedural justice (alpha=.94). The Perceived Procedural Justice (PPJ) subscale of the PCES (Watson, Angell, Vidalon, & Davis, 2010) is a 10 item measure designed specifically to tap perceptions of procedural justice (e.g. fair and respectful treatment, opportunity for voice) in police encounters with persons with mental illnesses. Items are rated on a 4 point agreement scale. Rasch analysis was used to examine the dimensionality of the 10-item measure, illustrating good person and item reliability (.88 and .91) and adequate person separation (2.74). Cronbach’s alpha for the measure was 0.94. Construct validity was also supported with medium to large correlations with theoretically related constructs. Here we use the sum of the 10 items divided by 10 to produce a scale score on the original scale metric, for ease of interpretation. Scale scores were centered by subtracting the mean from the raw score.

Perceived stigma was measured using the 12 item Perceived Devaluation & Discrimination Scale (PDS) developed by Link and colleagues (Link 1987, Link, 2001, Link, Mirotznik, & Cullen, 1991). This widely used measure is based on modified labeling theory, which holds that when people become officially labeled with a mental illness, they are already aware of negative societal beliefs about people in this labeled category. The labeling process leads the individuals to take on the group identity of a person with mental illness, and in the process, negative public attitudes become personally relevant. Applying negative public attitudes to the self, according to the theory, leads to a cycle of self-devaluation and expected rejection by others. In order to capture the respondent’s awareness of how the public might devalue someone with a mental disorder, the scale asks respondents about the extent to which they believe other people consider a person with mental illness unintelligent or a failure. To capture expectations of rejection, the scale also includes items pertaining to anticipation of discrimination in jobs, romantic relationships and friendships. Responses were scored on a four-point Likert-type scale from strongly disagree (1) to strongly agree (4). Six items were reversed scored before summing and dividing by 12 to produce a total score on the original metric with higher scores indicating greater perceived devaluation and discrimination. The scale had good reliability in this sample (Cronbach’s Alpha .82). Scale scores were centered by subtracting the mean from the raw score.

Negative pressures to do what the officer wanted were measured with a five items based on the Positive & Negative Pressures scale in the MacArthur Admission Experience Survey (Gardner et al., 1993; Monahan et al., 1999) (e.g. the officer lied to me, threatened, used physical force). The scale had good reliability (Cronbach’s Alpha .82). Scale scores were centered by subtracting the mean from the raw score.

Call type was coded by the interviewer based on the participant’s description of the nature of the encounter. These codes were collapsed into four categorical variables with one (street stop) as the reference category for the purpose of the analysis below. The categories are: street stop (includes nuisance activity such as panhandling or public drinking, an identification check or because the participant “resembled a suspect”); law violation of any type; request for assistance; and mental health crisis call. When entered into the regression models, street stop was left out as the reference category, based on prior research indicating that procedural justice was less important in street stops for identification than in other types of encounters (McCluskey, 2003).

Moderators

Four interaction terms were calculated to examine PDS and call type as moderators of the PPJ effect on cooperation and resistance.

Sample

In total, 155 individuals were recruited for the study. One participant was unable to complete most of the interview. Sample characteristics described are for the 154 participants: 75 (48.7%) from Thresholds North, 40 (26.0%) from Thresholds South, and 39 (25.3%) from Austin Apartments whose interviews were used in the analysis. One hundred nine (70.78%) of the respondents were male. Ages ranged from 20 to 63 years old, with a mean of 42.5 (SD=10.4) years. Ninety six (61.9%) participants identified as African American, 34 (21.9%) as White, seven (4.5%) as multiracial, four (2.6%) American Indian, one (0.6%) Native Hawaiian/Pacific Islander and 12 (7.7%) as other. Thirteen (8.4%) participants indicated Hispanic/Latino ethnicity. One hundred seventeen (76.0%) participants were single/never married, six (3.9%) were currently married, eight (5.2%) were separated, 20 (13.0%) were divorced and three (1.9%) were widowed. Fifty three (31.4%) participants indicated less than a high school education, 45 (29.2%) had graduated from high school, and 56 (36.3%) reported some college or vocational school. Annual household incomes reported ranged from $0 to $46,000, with a mean of $7,711(SD $6,961)

Analysis

Complete data for variables included in the models was available for 139 participants. Five people had a call type that fell outside of the four categories and 10 people had missing data for one or more items included in scale score calculations. Prior to selecting variables to include in the regression models, we examined bivariate relationships between the outcome variables and participant and situational variables. Theoretically relevant and significantly correlated independent variables were retained for the regression models. Cooperation and resistance were predicted in separate models using linear regression. When predicting each outcome variable, two regression models were tested using an approach that accommodates the recommendations of Aiken and West (1991). Model 1 included the control variables, call type, PPJ, PDS and negative pressures. The hypothesized two-way interaction effects were not entered in this model. Model 2 adds the four two-way interaction terms to the predictive model and enables us to test the two-way interaction effects. Post hoc analysis of simple slopes (Aiken & West, 1991) allows us to further explore the moderation effect of PDS in the cooperation model. We estimate the simple effects of PPJ at high, medium and low levels of PDS.

Results

Frequencies for call type, race and gender variables and means (SD) for uncentered scale scores are listed in the first column of Table 1. The largest category for call type was law violation, with 56 (40.29%) participants, followed by street stop, request for assistance and mental health crisis call with 45 (32.37%), 31 (22.30%) and seven (5.04%) participants respectively. The majority of the sample was male and nonwhite, and on average 43 years old (SD 10.4). In general participants did not endorse experiencing negative pressures but did perceive that the public stigmatizes persons with mental illnesses. In terms of perceived procedural justice, the mean fell between disagree and agree on the original four point scale.

Table 1.

Predicting cooperation and resistance in encounters between police and persons with mental illness (n=139)

VARIABLE	Cooperation Model 1		Cooperation Model 2		Resistance Model 1		Resistance Model 2
	B (SE)	(B)	B (SE)	(B)	B (SE)	(B)	B (SE)	(B)
Constant	3.286 (.188)**		3.167 (.186)**		1.976 (.204)**		1.990 (.209)**
Type of call
MH crisis (n=7)	−.133 (.258)	−.044	−.099(.249)	−.032	.237 (.280)	.066	.239 (.281)	.066
Law violation (n=56)	−.294(.128)*	−.216	−.248(.124)*	−.182	.287 (.139)*	.041	.278 (.140)*	.173
Req assist (n=31)	−.382(.166)*	−.235	−.296 (.164)	−.182	.259 (.179)	.135	.253 (.185)	.132
Street stop (n=45)
Age (43.0, SD 10.4)^	.011(.005)*	.168	.011 (.005)	.169	−.009 (.006)	−.125	−.011 (.006)	−.147
Male (n=99)	.026 (.127)	.018	.059 (.124)	.040	−.200 (.138)	−.114	−.190 (.140)	−.108
Nonwhite (n=108)	.009 (.139)	.005	.039 (.136)	.023	−.103 (.149)	−.053	−.107 (.151)	−.055
PPJ (2.4, SD 0.8)^	.224(.083)*	.268	−.002 (.118)	−.019	−.337(.090)**	−.346	−.289 (.133)*	−.297
Neg Pressures (2.0, SD 0.6)	−.129(.104)	−.126	−.154 (.104)	−.150	.311(.114)*	.256	.366 (.117)*	.302
PDD (2.8, SD 0.5)^	.297(.115)*	.222	.267 (.114)*	.199	−.149 (.126)	−.094	−.116 (.129)	−.174
PPJxPDD			−.341(.124)*	−.221			.135 (.140)	.074
PPJxMH Crisis			.680 (.342)*	.166			.538 (.385)	.111
PPJxLaw Violation			.347(.153)*	.240			−.074 (.173)	−.044
PPJ x Request Assistance			.288 (.174)	.170			−.078 (.193)	−.040
R2	.136		.196		.266		.265
F	3.434**		3.609**		6.589**		4.849**

^{^}mean prior to centering listed for ease of interpretation. Centered variable used in regression model.

^*p.<.05,

^**p.<.01

Separate analyses were performed to predict cooperation and resistance. In each of these instances, the Model 1 analysis examined the direct effects of the predictor variables on the outcome variable (see Table 1, Model 1). In Model 2 (Table 1, Model 2), the four interaction terms were added. We also examine the simple effects of PPJ on cooperation when PPD is low, centered at the mean and high.

Predicting Cooperation

Direct effects

When predicting cooperation, several significant findings emerged. Two of the call type categories, law violation and request for assistance, were associated with decreased cooperation B = −.294, S.E. = .128, t(130) = −2.302, p < .05 and B = −.382, S.E. = .166, t(130) = −.2304, p < .05 respectively compared to the street stop reference category. Both PPJ [B = .224, S.E. = .083, t(130) = 2.685, p < .05] and PDS [B = .297, S.E. = .115, t(130) = 2.576, p < .05] were associated with increased cooperation, as did Age [B = .001, S.E. = .005, t(130) = 2.084, p < .05]. Negative pressures and participant gender and race did not show significant effects.

Moderation

We hypothesized that PDS and call type would moderate the relationship between PPJ and the outcomes (cooperation and resistance). In order to test this, four two-way interaction terms were computed (i.e., PDS xPPJ; MH Crisis xPPJ; Law Violation x PPJ and Request Assistance x PPJ). These four two-way interaction terms were simultaneously entered as additional predictors in the regression models. Thus, the regression models used to test these four interaction terms included all of the predictors contained in the original “direct effect model” plus the four additional two-way interaction terms. The column labeled “Model 2” in Table 1 summarizes the results of these analyses. Several significant patterns of moderation emerged. First, call type appeared to moderate the PPJ effect on Cooperation. Both the MH Crisis x PPJ and Law Violation by PPJ interactions were significant B = .680, S.E. = .342, t(126) = 1.992, p < .05 and B = .347, S.E. = .153, t(126) =2.270, p < .05 respectively, indicating that the positive PPJ effect on cooperation is enhanced in encounters involving mental health crisis and law violations. The Request for Assistance x PPJ interaction was not significant. The PDS x PPJ interaction was also significant B = −.341, S.E. = .124, t(126) = −2.756, p < .05.

Analysis of simple slopes (Aiken & West, 1991) indicates that among participants who were “low” on PDS (one standard deviation below the mean), the PPJ effect on cooperation was strongly positive B = .355, S.E. = .099, t(129) = 3.583, p <.001. Likewise, for participants that were at the mean of PPD, the effect of PPJ was significant and positive B = .208, S.E. = .082, t(129) =2.531, p <.05. However, for participants who were “high” on PDS (one standard deviation above the mean), PPJ was not significantly related to cooperation B = .062, S.E. = .107, t(129) = .577, p =.565.

Predicting Resistance

Direct effects

When predicting resistance, the analysis showed several significant findings. One call type, law violation, predicted greater resistance B = .287, S.E. = .139, t(130) =2.064, p <.05. Both PPJ and negative pressures were significant predictors of resistance, with PPJ reducing resistance B = −.337, S.E. = .090, t(130) = −3.729, p <.001, and negative pressures increasing B = .311, S.E. = .114, t(130) = 2.735, p <.01. None of the other variables showed significant relationships with resistance.

Patterns of moderation

No patterns of moderation of the PPJ – resistance relationship were found.

Discussion

Building on the group engagement model of procedural justice, we assessed to what degree the procedural justice effect – that is, the relationship between feeling as though one is treated fairly and respectfully by police and subsequent behavior in the encounter (cooperation and resistance) -- is moderated by the level of stigma perceived by the person with mental illness and by situational uncertainty (call type). Our findings suggest that in general, individuals who perceived that a police officer treated them with greater procedural justice were more likely to report being cooperative and less likely to be resistant. The association between procedural justice and cooperation, however, is moderated by both perceived stigma and the type of encounter. Specifically, the association between fair and respectful treatment and cooperation operated in accordance with the group engagement model only for participants who reported a lower level of perceived stigma. In contrast, the association did not hold for participants who reported high levels of perceived stigma. With regard to uncertainty, we found that the relationship between perceived procedural justice and cooperation was largest in the two encounter types that may entail the most uncertainty and vulnerability-encounters in which the participant was accused of a crime and encounters in which the participant was experiencing a mental health crisis. Finally, the results also showed a direct effect of perceived stigma, which was somewhat surprising. While greater perceived stigma reduced or eliminated the effect of perceived procedural justice on cooperation, its direct effect was to increase cooperation.

Our findings on stigma and uncertainty as boundary conditions of the PPJ effect make sense in the context of the broader literature on procedural justice. We can consider perceived stigma to be an indicator of objective and subjective processes of identifying, at some level, with a marginalized social group and dis-identifying with the dominant majority (in this case, “normals,” or people who do not have mental illnesses). Even if people do not agree with a negative stereotype about their group, their knowledge that the general public holds the negative belief leads them to expect unfair or discriminatory treatment. Someone who feels completely excluded may no longer bother to try to affirm his or her status as part of the dominant group, and thus will be less attentive to the quality of treatment received from authorities for this purpose. This does not mean the person will not cooperate, as our findings about higher stigma being associated with greater cooperation suggest. The procedural justice framework assumes cooperation is based on normative commitments to and shared values with authorities -- in this case, the police. For people who feel wholly excluded from the “system,” these normative commitments and shared values may be absent. Discussing findings from a study of public cooperation with police in Ghana, Tankebe (2009) notes however, that in sociopolitical environments in which the price of noncompliance is abuse and torture, normative consent is not an option. In such contexts, compliance may be otherwise motivated, such as out of fear, survival, or complete demoralization.

Participants in our study were public mental health system clients. In addition to their mental illnesses, they were poor, and predominantly African American. As a group, they are among the most marginalized and vulnerable. In qualitative interviews conducted in an earlier phase of the study, many participants expressed feeling extremely vulnerable in encounters with police officers. One participant stated “You just have to speak to them the best way you can and hope that you aren’t going to say something wrong. I know some police, they got the attitude that they would rather kill you… And it’s been done around here (Watson, Angell, Morabito, & Robinson, 2008, p. 452).” Thus, for some participants, cooperation in the encounter may have been purely a survival strategy not impacted by the quality of treatment by the police officer. With respect to the direct effect of stigma on cooperation, we suggest that individuals who highly endorse the existence negative public attitudes toward people in their labeled category and anticipate discrimination may react in a similarly fear-motivated manner in their dealings with the police. Interestingly, in a recent study of Muslim Americans’ willingness to cooperate with police in anti-terror work, perceived social discrimination against Muslims was negatively associated with cooperation, defined as willingness to engage in cooperative anti-terrorism actions if asked to by police and willingness to report suspicious activity to police (Tyler, Schulhofer, & Huq, 2010). Procedural justice was positively associated with cooperation. Perhaps the impact of perceived stigma and discrimination on cooperation varies depending on the immediacy of the encounter with the authority, as well as the type of cooperation considered.

In previous studies, the level of perceived stigma (using the PDS) has been shown to be associated with a variety of negative outcomes, including lowered self-esteem, self-efficacy, psychological demoralization, and disempowerment. It is possible that the moderating effect observed in this study may arise from another mechanism, such as the effect of self-esteem on procedural justice processes. While some research has found that procedural justice reactions are stronger for people with high compared to low self-esteem (De Cremer, Knippenberg, Dijke, & Bos, 2004), other research has found a positive relationship between procedural justice and organizational commitment among people with high self-esteem, but no such relationship among people with low self-esteem (Wiesenfield, Swann, Brockner, & Bartel, 2007). Unfortunately, our study did not include a measure of self-esteem that would permit us to test this alternative possibility.

While our findings must be viewed as exploratory and subject to several important limitations, we suggest, tentatively, that they have implications for the growing literature on how to maximize successful encounters between people with mental illnesses and police. Obviously, we would not condone promoting further marginalization and abusive treatment of persons with mental illnesses as a method of increasing cooperation in police encounters, which would be unethical and unjust. It would also be unwise given our finding that forceful and abusive treatment (negative pressures) in the immediate encounter increased resistance. Rather, we suggest that our findings support policing strategies consistent with procedural justice AND the need for continued efforts to reduce stigma and promote full inclusion of persons with mental illnesses.

Our finding that call type moderated the impact of procedural justice also makes sense in the context of research on uncertainty as a boundary condition of procedural justice (van den Bos et al., 1998; van den Bos, 2001). This work suggests that people will pay more attention to the quality of treatment by an authority under conditions of uncertainty about whether they can trust the authority not to exploit them. The effect of PPJ on cooperation was enhanced when the subject’s encounter involved either a significant law violation or a mental health crisis. These are both situations in which people are at risk of loss of liberty. They may be taken into police custody and taken to jail or transported for emergency psychiatric evaluation. Either way, they may feel vulnerable and uncertain and be more inclined to look to how the officer (s) treats them for reassurance. Given our sample size, we did not have the power to examine the presence of a three way interaction of call type, perceived stigma and procedural justice.

Our findings of moderator effects on cooperation, but not resistance, warrant comment as well as acknowledgement of the only minimally adequate internal consistency of the three item scale used to measure resistance. At face value, resistance may seem like the inverse of cooperation. Yet, while procedural justice had the expected effect of reducing resistance in the encounter, the pattern of prediction was somewhat different from that of the cooperation model. Perceived stigma had no effect and we found no patterns of moderation. Rather, PPJ was associated with reduced resistance and negative pressures (or more coercive and forceful behavior on behalf of the police officer) was associated with increased resistance. Additionally, resistance was greater in encounters involving law violations than street stops for identification checks or nuisance activities.

In their observational study of police encounters, Dai, Frank and Sun (2011) also found that disrespectful demeanor (similar to our resistance construct) and noncompliance (likewise similar to cooperation) were differently predicted. They suggest that while citizen disrespect and noncompliance are related and important considerations in terms of maintaining social order in police –citizen interactions, they are distinguishable. Citizen disrespect “reflects a broader range of citizen responses to police behaviors, whereas noncompliance captures citizen responses to specific requests made by the police (Dai, Frank, & Sun, 2011, p.161).”

Our findings suggest that increasing officers’ use of strategies consistent with procedurally just treatment (empathy and respect, allowing voice, fairness) and reducing use of negative pressures (coercion & force) may improve their ability to safely and effectively resolve calls involving persons with mental illnesses. While not framing their approach specifically in procedural justice terms, promising models of specialized police response to persons with mental illnesses incorporate training in de-escalation skills. Some, such as CIT, include experiential components designed to reduce stigma and promote empathy (Compton et al., 2011). CIT, which features both specialized officer training and service system alteration, is particularly popular, with over 1,000 programs currently operating in over 39 states. The emerging body of research suggests CIT is effective for improving safety and linkage to mental health services (Compton et al, 2011; Morabito, Kerr, Watson, Draine, Ottati, & Angell, 2012; Ritter, Teller, Munetz, & Bonfine, 2010; Watson, 2010). While it is important to examine the effects of these programs on arrest rates, diversion to treatment, use of force, and frequency of officer injury, additional questions regarding the mechanism of effects likewise remain unanswered. In particular, it is important to understand in what ways training police officers to forego traditional policing methods and to take a more patient, respectful and empathic approach to handling crisis calls involving subjects with mental illnesses actually changes officer attitudes and behavior, and in what way this in turn alters the subjective experience and cooperativeness of the individual with mental illness. Procedural justice theory offers a useful lens for considering how specialized police training may transform both police behavior and citizen responsiveness

Limitations

While our findings have important implications for theory, research and practice, they must be considered within the context of several important limitations. First, our relatively small sample consisted of individuals with serious mental illnesses receiving public mental health services in Chicago. Their experiences in police encounters may not be generalizable to individuals with mental illnesses receiving private services, those not receiving services, or those who come in contact with police in nonurban settings. Second, we did not have collateral reports of the encounters from police officers themselves or observational data. Independent and dependent variables were measured based on participants’ perceptions of the encounters. Hence, these measures are subject to both recall biases and problems of shared method variance. While the associations found in the data are compelling and consistent with theory, we caution against causal inference. Future studies that include observational data and or police reports of encounter characteristics are needed to confirm our findings. Third, we assessed experiences specific to a particular police encounter, which could potentially overlook other important interactions with police that influence cooperation and resistance in the immediate encounter. Finally, our sample size, while adequately powered for our main analysis, limited our ability to more fully explore complex interactions.

Conclusions

In conclusion, we reiterate the importance of furthering research to better understand both how procedural fairness benefits interactions between citizens and authorities and the ways that stigma, perceived and internalized by affected individuals, may alter their response to psychological processes in social interactions with authorities. If people with mental illnesses who are highly aware of stigma and who feel disempowered in their relations with authorities simply submit to authority, their rights may be subverted. In addition, the subjective experiences of marginalization that prevent them from attending to fair treatment has troubling implications for other aspects of citizenship and community integration. On the other hand, our findings reinforce the importance of procedurally fair treatment for less marginalized individuals and may provide a theoretical basis for interventions like CIT that teach police to treat people with mental illness in a more humane manner.