Introduction

Police have a central role in handling situations involving people with mental illness. Attending to mental health crises, responding to people with mental illness who are witnesses and victims of crime, and diverting people with mental illness to social and mental health services are routine duties for police officers (Fry et al. 2002; Steadman et al. 1997). The cornerstone of policing is front line patrol officers, whose ability to carry out their duties effectively are impacted by how they are perceived by the community. The degree to which community members are willing to engage proactively and cooperate voluntarily with the police is also integral to their efficacy. Procedural justice theory provides a framework for understanding how perceptions about authorities, such as the police, are created, maintained, and transformed. The present study examines the influence of selected factors on how people with mental illness perceive their interactions with the police.

For more than three decades, procedural justice theory has provided a theoretical lens for conceptualizing the procedural elements of justice and fairness. According to this theory, a central factor shaping perceptions about authority figures and dispute resolution processes is whether an individual believes they were treated justly and fairly during personal encounters with agents of authority (Tyler and Fagan 2008). Although this may be seemingly intuitive, the notable aspect of this process is that positive appraisals occur regardless of whether the actual outcome was perceived as being favorable. Conversely, unfair and unjust treatment by agents of authority is likely to produce negative evaluations. The theory further postulates that people who believe they are treated with procedural fairness by authorities will be motivated to comply with subsequent decisions (Tyler 1990). Collectively, these aspects of the larger theory are referred to as ‘perceived procedural justice’ (PPJ). When applied to the criminal justice system, the theory stipulates that PPJ regarding agents of the law (e.g., police officers, judges, correctional officers) is important for shaping beliefs regarding whether legal authorities can be entrusted with the responsibilities of performing their duties, which, in turn, determines levels of compliance, cooperation, and law-abiding behavior. Numerous studies have found that members of the general public who feel treated in a procedurally just manner by the police are more likely to: (a) evaluate the encounter positively, (b) view the police as legitimate agents of social intervention, and (c) cooperate with the police (McCluskey et al. 1999; Reisig et al. 2007; Sunshine and Tyler 2003; Tyler 1990; Tyler and Fagan 2008; Wells 2007).

Studies on PPJ have also demonstrated its relevance for a range of processes involving people with mental illness. Research among people with mental illness in the context of receiving mental health services (e.g., civil commitment procedures) indicates that PPJ is associated with greater cooperation and adherence with treatment, enhanced acceptance of involuntary status, and lower perceptions of coercion (Cascardi et al. 2000; Hiday et al. 1997; Lidz et al. 1995; McKenna et al. 2001, 2006). To date, relatively few studies have assessed PPJ among people with mental illness who are involved in the criminal justice system.

Amongst the extant literature, two studies have examined PPJ in the context of mental health courts (Poythress et al. 2002; Wales et al. 2010) and a further two in the context of policing. In the first in-depth study to examine PPJ among people with mental illness regarding their police interactions, qualitative interviews were conducted with 26 individuals with mental illness who had had at least one contact with police in the past year (Watson et al. 2008). The findings indicated that participants’ evaluations of police interactions were greatly influenced by how they perceived police officers’ behavior on aspects of fairness, respectfulness, and kindness. Following this study, a quantitative measure was developed, the Police Contact Experience Scale (PCES), and administered to 152 people with mental illness who had contact with the police in the past year (Watson et al. 2010). The study found that PPJ was associated with participants’ appraisals of police behavior (e.g., perceived coercion, negative pressure), their beliefs about the police (e.g., legitimacy), their feelings about police encounters (e.g., emotional impact), and their willingness to comply with police officers’ requests (e.g., decision acceptance, resistance, and cooperation). Although the researchers did not examine the effects of contextual factors on PPJ, they suggest this as an area that would greatly benefit from future research. Data from the same study was further analysed, which revealed that perceived stigma and encouter type moderated the relationship between PPJ and cooperation with the police (Watson and Beth 2012). Interestingly, the association between PPJ and cooperation was not present among individuals with high levels of perceived stigma related to their mental illness.

In this paper, we describe findings from a mixed methods study that examined how people with mental illness perceived and interacted with the police. The growing interest in PPJ parallels the expansion of efforts to reduce the negative consequences to people with mental illness resulting from contact with the criminal justice system. Building upon the descriptive results presented elsewhere (Brink et al. 2011), the data presented here explores the influence of selected individual, contextual, and interactional factors on PPJ regarding police interactions among people with mental illness.

Method

Procedure

Participants were recruited from mental health centers throughout Metro Vancouver, Canada. Peer researchers (persons with lived experience of severe mental illness and who had been trained by the investigators) obtained written informed consent and administered a semi-structured interview to the participants. Interviews were conducted in private and lasted approximately 60 min. Participants were paid $10 for taking part in an interview. The research protocol was approved by the research ethics boards of the University of British Columbia, Simon Fraser University, and several relevant health agencies.

Participants

The inclusion criteria for the study were: (a) current formal diagnosis of schizophrenia, schizoaffective disorder, delusional disorder, or bipolar disorder (self-reported); (b) previous direct contact with the police in Canada (self-reported); (c) age 19 years or older; (d) able to speak and understand English; and (e) cognitively capable of providing research consent. We focused on these diagnostic categories specifically, as opposed to others (e.g., depression, anxiety), because the signs and symptoms associated with these disorders are more likely to be observed by police officers during an interaction. Previous contact with the police included interactions with the police under any circumstances (e.g., as a suspect, as a victim, requesting other assistance) and at any time in the participants’ lives.

Materials

The semi-structured interview guide contained detailed questions regarding participants’ lifetime and most recent police encounters, and included several scales from the Police Contact Experience Scale (PCES) (Watson et al. 2010). The PCES is a measure for assessing situational, procedural, and interpersonal aspects of interactions between people with mental illness and the police. It includes a 10-item perceived procedural justice (PPJ) scale that assesses the extent to which someone believes they have been treated in a procedurally just manner during a police interaction. The PCES is also comprised of scales that assess perceived coercion (5 items), negative pressure (5 items), positive pressure (2 items), and subject demeanor (7 items). All PCES items are rated on a 4-point scale. Participants in the present study were asked to think about their most recent interaction with the police for the purpose of rating the PCES items. Higher scores on the PCES scales indicate higher levels of the corresponding construct (e.g., procedural justice, coercion).

Statistical Analyses

Multiple linear hierarchical regression analyses were performed to assess the relationship between PPJ and selected individual, contextual, and interactional variables (Tabacknick and Fidell 2007). For multiple regression analysis, a sample size of 52 is the minimum required for detecting a large effect size (r 2 = 0.35) given a 0.80 statistical power level with 8 predictors in the model and a probability level of 0.05 (Soper 2012).

Dependent Variable

PPJ was measured with the 10-item subscale from the PCES. Regression models were performed on the total average PPJ subscale score, which demonstrated a high degree of internal consistency (α = 0.94). All analyses were performed using the Statistical Package for Social Sciences (SPSS).

Independent Variables

Explanatory variables potentially linked, empirically or theoretically, to PPJ were identified and separated into three categories: individual, contextual, and interactional. Individual variables comprised several socio-demographic, clinical, and experiential factors that are summarized in Table 1. For the ‘police use of excessive force’ variable, participants were asked to indicate whether or not they believed that they had ever been the recipient of excessive physical force by a police officer. Similarly, for the ‘injury by police’ variable, participants were asked to indicate whether or not they had ever suffered minor or serious injury during an interaction with a police officer.

Table 1 Summary of individual factors
Full size table

Contextual variables consisted of factors surrounding the most recent police interaction that are summarized in Table 2. In addition to the interactions that were initiated by the police (35.1 %, n = 21) and the participants themselves (13.3 %, n = 8), the participants indicated that their most recent interaction with the police was also initiated by other citizens (25.0 %, n = 15), staff members (e.g., mental health) (13.3 %, n = 8), and family members (5.0 %, n = 3). Data pertaining to rates of poverty and serious crime in the local neighborhoods where police interactions occurred was obtained from government statistics (www.bcstats.gov.bc.ca). Local neighborhood poverty was defined as the percentage of the population with income less than $25,000 CAD. Local neighborhood serious crime was defined as the number of serious drug, violent, and property crimes per 1,000 local population.

Table 2 Summary of contextual factors
Full size table

Interactional variables consisted of processes that had transpired during the course of the most recent police interaction and are summarized in Table 3. Over one-quarter of the participants (28.3 %, n = 17) came into contact with the police in the context of needing help for a mental health crisis. Participants also sought assistance from police as victims (18.3 %, n = 11) and witnesses of crimes (5.0 %, n = 3). Additionally, several of the interactions involved criminal justice-related incidents, such as engaging in criminal acts (18.3 %, n = 11), public disturbances (11.7 %, n = 7), domestic disputes (10.0 %, n = 6), and motor vehicle infractions (5.0 %, n = 3). Four PCES scales were analysed as interactional variables. Negative pressure refers to the use of deception, threats, or force by police officers in order to influence an individual’s behavior. In contrast, positive pressure refers to the use of persuasion or inducements by police officers (Gardner et al. 1999).

Table 3 Summary of interactional factors
Full size table

In order to reduce the number of variables added to the final regression model, independent stepwise regression analyses were performed to examine the association of variables within each category (individual, contextual, and interactional) with PPJ. Variables significant at a 0.01 level were entered into the hierarchical linear regression model. Individual variables were forced into the first block of the model, followed by contextual and interactional variables in the second and third blocks, respectively. Variables were entered in this order so that the effects of interactional variables on PPJ could be ascertained after controlling for individual and contextual factors. To confirm the final model, all significant and non-significant variables were entered into a stepwise regression analysis.

Results

The majority of the participants’ most recent interactions occurred within the past year (n = 33, 55.0 %), 21.7 % (n = 13) occurred within 1–3 years, and 23.3 % (n = 14) occurred more than 3 years prior to the interview. Overall, recent police interactions were viewed positively by the participants. The average rating was 2.93 (SD = 0.59, min = 1.00, max = 4.00), with 81.7 % (n = 49) of participants having an average score above the mid-point of the scale (2.50). Only two participants had mean total scores at either the high or low extreme ends of the scale, indicating that the PPJ measure did not suffer from floor or ceiling effects. There was no significant association between PPJ and how recent the participants’ had last interacted with police (r = −0.09, P > 0.05).

Individual Factors and PPJ

Among the individual variables entered into the stepwise regression analysis, gender (B = 0.57, SE = 0.19, p < 0.01), history of incarceration (B = −0.79, SE = 0.19, p < 0.001), and country of birth (B = 0.49, SE = 0.17, p < 0.01) were significantly associated with PPJ. Higher levels of PPJ were related to being male and born in Canada (i.e., non-immigrant). Lower levels of PPJ were related to a history of being incarcerated. Together, these three variables accounted for 26.3 % of the variance in PPJ (R 2 adj  = 0.26, SE = 0.50, F = 7.88, p < 0.001).

Contextual Factors and PPJ

Among the contextual variables, quality of previous police experiences was the only variable significantly associated with PPJ (B = 0.22, SE = 0.04, p < .001). Participants with positive views of their previous interactions with police were more likely to perceive their most recent contact as being procedurally just. This variable accounted for 29.7 % of the variance in PPJ (R 2 adj  = 0.30, SE = 0.49, F = 25.49, p < 0.001).

Interactional Factors and PPJ

Among the interactional variables, perceived coercion (B = −0.43, SE = 0.06, p < 0.001), negative pressure (B = −0.44, SE = 0.08, p < 0.001), experiencing mental health crisis (B = 0.28, SE = 0.08, p < 0.01), and apprehension or arrest (B = 0.33, SE = 0.10, p < 0.01) were significantly associated with PPJ. Higher levels of PPJ were related to the police officer using less coercion and less negative pressure (e.g., threatening, physical force). PPJ was higher for interactions that involved a mental health crisis and for those that did not result in apprehension or arrest. These four variables together explained 78.3 % of the variance in PPJ (R 2 adj  = 0.78, SE = 0.27, F = 53.14, p < 0.001).

PPJ and all Factors

The first block of the hierarchical multiple linear regression analysis contained the three individual factors (gender, history of incarceration, born in Canada), which were all significant, and together, accounted for 26.3 % of the variance in PPJ, R 2 adj  = 0.26, SE = 0.50, p < 0.001. In the second block, the individual factors remained significant after adding the single contextual variable retained (quality of prior police interactions), and the explanatory power of the overall model improved, R 2 adj  = 0.47, SE = 0.43, p < 0.001. In the third iteration of the regression model, the individual and contextual factors were no longer significant after the introduction of the four interactional factors (perceived coercion, negative pressure, mental health crisis, and apprehension/arrest), which together accounted for 78.2 % of the variance in PPJ, R 2 adj  = 0.78, SE = 0.28, p < 0.001. Therefore, PPJ was most strongly influenced by what took place during the police interactions (i.e., interactional factors), as opposed to the individual characteristics of the participants (i.e., individual factors) or extrinsic factors surrounding the encounter such as neighborhood or prior experience (i.e., contextual factors). This was confirmed by performing a multiple regression analysis using all significant and non-significant variables; perceived coercion and negative pressure were the only variables associated with PPJ at a 0.01 significance level.

Discussion

Recent studies have called for further investigation into the factors that influence PPJ among people with mental illness (Wales et al. 2010; Watson et al. 2010). The present study examined PPJ in the context of police interactions among this population, providing empirical data on the degree to which these perceptions were associated with a range of different factors. In relation to their most recent contact with the police, participants reported that the officers acted, for the most part, in a procedurally just manner. The results also suggested that PPJ was strongly related to the way in which police officer(s) handle situations involving people with mental illness. Lower ratings of PPJ were associated with police interactions that involved negative pressure (e.g., threats, deception, or physical force), coercion, and apprehension or arrest. These findings align with results of a comparable study by Watson et al. (2010), who found that PPJ was highly correlated with perceived coercion and negative pressure. Similarly, Sunshine and Tyler (2003) also found that, among a sample of the general public with recent police interactions, the three strongest predictors of PPJ were quality of treatment by police, quality of police decision-making, and police performance.

Among our participants, individual and contextual factors had some influence on PPJ; however, the explanatory power of these variables was substantially reduced once interactional factors were considered. The extant research literature is mixed regarding the relationship between PPJ and individual-level factors. For instance, several studies have found that PPJ among the general public in the context of police encounters does not appear to be influenced by socio-demographic characteristics (e.g., race, sex, class) (Gottfredson et al. 2007; Paternoster et al. 1997; Sunshine and Tyler 2003; Tyler and Fagan 2008; Tyler and Huo 2002; Wells 2007). Similar findings have been reported for people with mental illness. For example, a study by Hiday et al. (1997) determined that PPJ of people with mental illness regarding their hospital admissions was unrelated to several personal characteristics (e.g., age, race, residential status, diagnosis, symptom severity).

On the other hand, a few studies have uncovered significant relationships between PPJ and some individual factors, such as legal status (Lidz et al. 1995; McKenna et al. 2001), relationship status (Hiday et al. 1997), and education level (Sunshine and Tyler 2003). The evidence is perhaps most conflicting as to the role of gender in PPJ, with some studies indicating no relationship (Lidz et al. 1995; Sunshine and Tyler 2003), and others reporting contrary findings (Gottfredson et al. 2007; Hiday et al. 1997). Research also suggests that the nature of the context (e.g., interaction with judge, police, correctional officer) and situation (e.g., interacting with police as a victim versus a perpetrator of crime) may have some bearing on PPJ among the general public (Bradford and Stanko 2009; Tyler 1988; Vermunt et al. 1998; Wells 2007). Among our participants, police interactions that occurred while individuals were in mental health crises, as opposed to other types of interactions (e.g., street stop, suspect of a crime), were associated with more positive ratings of PPJ. Overall, however, the empirical evidence has established neither a strong nor consistent association between individual and contextual factors and PPJ. When combined with the findings of the current study, this suggests that PPJ may be influenced most strongly and reliably by dynamic factors that are potentially modifiable through intervention, such as police training focused on reducing the use of coercion or negative pressure. Future research should examine PPJ in a single type of police interaction, such as a mental health crisis, using a larger sample size in order to produce a more in-depth and nuanced investigation into the dynamic relationships between the person with mental illness, the police officer(s), and the environment.

Several methodological limitations should be considered when interpreting our findings. First, the sample size was modest in relation to the number of comparisons made, which could have increased the risk of spurious findings. The small sample size restricted the types of analysis that could be performed on the data, including limiting subgroup analyses and the number of independent variables entered into the model. For instance, using a conservative significance threshold (p < .01) for entering variables into the hierarchical regression model may have excluded potentially important explanatory variables of PPJ. Second, characteristics of our sample may have introduced further limitations. Younger people and those from ethnically diverse communities were underrepresented in the study. Third, the study may be limited by the subjective, retrospective, self-report nature of the data collection. None of the self-reported information gathered was corroborated; therefore, some information may be inaccurate or biased.

The results of the current study, in conjunction with those of other research, have implications for improving interactions between people with mental illness and the police. The importance of PPJ is underscored by Tyler and Fagan (2008) who state that “Knowing what is experienced by members of the public as fair or unfair is key to developing and maintaining public views that the legal system is legitimate” [emphasis added] (p. 264). Moving the focus of inquiry back a step, the current study examined why interactions with the police are experienced by people with mental illness as fair or unfair. The behavior of police officers appears to be the key as to whether or not these interactions are perceived as procedurally just. An officer who uses a cooperative approach without verbal or physical force is likely to be perceived as procedurally fair. Among our study participants this was irrespective of individual or contextual variables.