Police Decision-Making in the Gray Zone

Abstract

Research finds mentally ill persons have higher rates of police contacts, arrests, and criminal charges for minor offenses and noncriminal behavior. It remains unclear whether the decision-making process and factors affecting discretion reflect a procedural bias that criminalizes the mentally ill. Using observational data from a Canadian police service, the findings suggest higher odds for criminal charges with serious offenses, males, older citizens, a prior criminal record, being under the influence of alcohol or drugs, and being uncooperative with requests for information and compliance. The odds of a citation are higher for proactive calls, more serious offenses, older citizens, mentally ill persons, those under the influence, or with a disrespectful demeanor. The overall findings suggest an indirect procedural bias exists due to situational constraints, a disjuncture between policy and police culture, and limited mental health resources that lead to response strategies that contribute to criminalization of the mentally ill.

Keywords

police decision-making police discretion mental health policing strategies police practice criminalization of the mentally ill

A growing concern is the fact that persons with a mental illness (PMI) are more likely to be victims than offenders, but they have higher rates of police contact, arrests, and criminal charges for minor offenses when compared with the general population (Heslop, Stitt, & Hoch, 2013). Complicating this further is that we continue to lack consensus on whether police decision-making practices contribute to the criminalization of mentally ill persons’ behavior (Godfredson, Thomas, Ogloff, & Luebbers, 2011; Lamb, Weinberger, & DeCuir, 2002; Novak & Engel, 2005) or if caution should be exercised as this conclusion is not necessarily generalizable (Brown, Novak, & Frank, 2009; Teplin, 2000; Watson, Morabito, Draine, & Ottati, 2008). Health Canada estimates that one in five Canadians will experience a mental health problem in their lifetime (Canadian Mental Health Association, 2014). Yet there is little Canadian research on police interactions with the mental health system, effective approaches, the content and consistency of training initiatives, or any standards for collecting data on police–citizen encounters with PMI (Chandrasekera, 2011; Cotton & Coleman, 2010). As a consequence, the Mental Health Commission of Canada and the Canadian Association of Chiefs of Police recognize this growing social problem and in March, 2014, called for research informing best practices when responding to PMI, the development of national standards, provincial frameworks, and increased multisectorial collaboration (Blue Line, 2014).

In recognition of this pressing social problem, the Canadian Association of Chiefs of Police (2008) released policy guidelines predicated on the following principle:

Each police organization should foster a culture in which mental illness is viewed as a medical disability not a moral failure, and in which people with mental illnesses are treated with the same degree of respect as other members of society. It is incumbent on police leaders to set an appropriate tone by modeling non-derogatory language and ensuring the assignment of police personnel to mental health-related positions in the organization is considered carefully. (p. 25)

However, the way in which this manifests itself in policy and practice varies widely. This incongruence is increasingly an issue for Canadian police services due to a significant increase in calls for service involving PMI while no training policy framework exists to ensure consistent implementation of this principle across police services.

Although order maintenance activities comprise much of police work (Brown et al., 2009; Short, MacDonald, Luebbers, Ogloff, & Thomas, 2014; Smith, Novak, Frank, & Lowenkamp, 2005; Watson et al., 2008), ambiguity and contradictory findings exist for the relationship between a citizen under the influence of alcohol or drugs and other factors identified as influencing the decision to arrest, including responding to a dispatched call, offense seriousness, sex, demeanor, and noncompliance with officer directives for information or behavioral change (Mastrofski, Worden, & Snipes, 1995; Terrill & Paoline, 2007). This analysis is part of an ongoing conversation about the nature of the relationship between these factors, their predictive ability, and the decision-making process leading to a criminal charge or citation, particularly as it is largely unexplored in Canada.

This study seeks to address the question of whether a procedural or decision-making bias exists when police decide to invoke the law to resolve the police–citizen encounter. Thus, the research objective is to explore the ways in which the decision-making process and use of discretion with PMI reflects a systematic procedural bias that can directly or indirectly contribute to criminalization of the mentally ill. By distinguishing between crime and non-crime-related encounters, the purpose is to highlight that criminalization of public order and minor offenses occurs in the context of limits on police decision-making autonomy. Understanding police decision-making in PMI-related calls remains important as there is a greater need for partnerships with community mental health resources and training that equips officers with a framework on which to employ evidence-based practice. After providing an overview of the factors affecting the use of discretion, a review of policing PMI, criminalization, and training on police responses to PMI is provided. The method section outlines the sample, procedures, and analytical strategy used to collect and analyze the quantitative and qualitative data. The results present the findings for the three research questions that investigate encounter dynamics, factors affecting discretion, and procedural challenges that affect police decision-making. Finally, the last two sections consider the implications for policy, practice, and training.

Police Discretion and Decision-Making

Police decision-making involves choices “regarding when and how to intervene in situations” (Roberg, Crank, & Kuykendall, 2000, p. 276). In contrast, police discretion is more specific as it reflects the decision to use informal social control (e.g., verbal warning) even when the circumstances legally allow for an arrest, issuing a citation, or laying a criminal charge (Schulenberg, 2015). This article explores both in greater detail as it relates to police responses in calls involving PMI.

Factors Affecting Police Discretion

Legal factors comprise law-related aspects of the encounter that influence the decision to arrest or charge, such as offense seriousness, prior record, type of offense under investigation, and strong situational evidence (e.g., Ericson, 1982; Kochel, Wilson, & Mastrofski, 2011; Riksheim & Chermak, 1993; Sherman, 1980). Extralegal factors reflect aspects of the encounter that “are prohibited or not explicitly authorized by law as relevant to the arrest decision” (Kochel et al., 2011, p. 476). Empirical support is inconsistent in finding a relationship with citizen characteristics such as age, sex, race, and situational factors such as demeanor and complainant preference (e.g., Ericson, 1982; Kochel et al., 2011).

Of importance are three extralegal factors that have varying degrees of empirical support in prior research. First, there is the question of mental health status, which has been found to affect decision-making but just as often to be a nonsignificant factor (Kochel et al., 2011; Novak & Engel, 2005). Second, it is unclear whether a citizen under the influence of alcohol or drugs increases the likelihood of an arrest (Mastrofski et al., 1995; Terrill & Paoline, 2007). Finally, it is beyond the scope of this article to go into the extensive debate concerning definitions and the influence of citizen demeanor. However, Klinger (1996) raises an interesting point that it may only play a role in cases where the offender is excessively hostile toward the officer and that it encompasses multiple dimensions of behavior. Thus, this study demarcates between the dimensions of disrespect and noncompliance as indicators of demeanor.

Policing the Mentally Ill

Calls for service involving a PMI are common with the social and economic costs increasing exponentially. Canadian surveys find that, on average, 26% of officers have contact with PMI at least once a week, a finding almost identical to those in this study where 25.2% were identified by officers as PMI and 2.3% of the encounters involved an Ontario Mental Health Act (1990) apprehension¹ (Cotton, 2004; Godfredson et al., 2011; Royal Canadian Mounted Police, 2010). Estimates suggest 40% of PMI have been arrested in their lifetime, 28% in Ontario have police contact for criminal activity with 20% for violent offenses, 40% for minor offenses, and 40% for noncriminal conduct such as calls for assistance, public order offenses, or street checks (Brink et al., 2011).

More recently, research in Canada finds PMI are 2 times more likely to be under investigation for a violent offense (Heslop et al., 2013). However, this finding does not negate the trend for PMI to be more likely to be arrested for minor offenses that are seen as violating community norms—disturb the peace, theft, fail to appear, public intoxication, disturbances, and suspicious persons (Hoch, Hartford, Heslop, & Stitt, 2009; Novak & Engel, 2005; Royal Canadian Mounted Police, 2010). Officers interpret nuisance behaviors as more serious because they draw the public’s attention creating a situation where citizens have higher expectations that the behavior will be dealt with punitively or at least steps taken to show something is being done about the behavior (Reuland, Schwarzfeld, & Draper, 2009; Teplin, 2000). At the same time, it remains debatable whether the police decision-making process is similar for PMI and non-PMI despite PMI having 3 to 5 times more interactions with the police than members of the general population (Cotton & Coleman, 2010; Heslop et al., 2013; Hoch et al., 2009).

One response to the increase in number of calls are Crisis Intervention Team (CIT) officers. They are assigned to active patrol duty and trained to serve as specialized responders with the goal of redirecting PMI out of the criminal justice system, when appropriate (Compton, Bahora, & Watson, 2008). This is accomplished by conceptualizing de-escalation techniques as “the ‘active ingredient’ that effects officers’ ability to resolve a call without the use of force, injury or arrest” (Watson et al., 2008, p. 363). This response strategy “is a systemic intervention, not an individual intervention” whereby PMI are a public health, not a safety concern (Watson et al., 2008, p. 366; Watson et al., 2010). Prior research suggests that CIT officers are more likely to divert PMI, but there is no impact on arrest practices, with the exception of substance abuse increasing the odds and female officers less likely to arrest a PMI (Watson et al., 2010).

To summarize, there are three reasons why research on factors affecting the use of discretion for criminal and noncriminal behavior is needed. First, although CIT is the predominant approach adopted in the United States, this is not the case in Canada, where Mobile Crisis Teams (MCTs) consisting of a mental health professional riding with a police officer, are more prevalent, but continue to be restricted in their ability to respond to the higher number of PMI-related calls (Cotton & Coleman, 2010; Watson et al., 2010). This results in patrol officers remaining the primary responders and decision-makers with knowledge based on recruit training and field experience. Second, it is possible that current mental health training is not as effective as we would like to believe. Thus, factors other than training content are influencing the decision-making process (e.g., substance abuse, demeanor). The third reason centers on complications created by an inability to keep pace with the number of officers and resources required to respond to calls for service (Plecas, McCormick, & Cohen, 2012). In 2010 to 2011, calls involving PMI to the London Police Service in Ontario, Canada increased 134%, compared with 28% for non-PMI (Heslop et al., 2013). Excluding calls for service that involve a behavior subject to a potential citation or charge, in 2013, officers at one regional police service experienced a 20% increase that solely involved suicide attempts or persons in mental health crises (Mercer, 2015). In addition, PMI-related calls typically take on average 20 min longer, regardless of the offense under investigation (Engel & Silver, 2001; Hoch et al., 2009; Royal Canadian Mounted Police, 2010; Watson et al., 2008).

Criminalization of the Mentally Ill

The goals of deinstitutionalization were to reduce the number of patients, time spent in psychiatric facilities, and the number of readmissions by maintaining social ties and receiving treatment in the community. Unfortunately, evidence suggests the increase of mentally ill persons in the community did not lead to a greater acceptance of PMI by citizens; rather, a continued stigma associated with mental illness coupled with a decrease in public tolerance for aberrant behavior (Levesque, 2010; Royal Canadian Mounted Police, 2010; Watson & Angell, 2012). This transinstitutionalization changed the role of the police by necessity and the gatekeepers to the justice system are subject to increased scrutiny with the overrepresentation of PMI in the correctional system (Chaimowitz, 2012; Cotton & Coleman, 2010; Mahoney, 2013; Watson et al., 2008).

Stigma and stereotypes may also affect police decision-making. Despite officers taking a variety of factors into account when deciding upon the best course of action, this process may not be neutral due to a potential systematic procedural bias whereby decision-making with PMI results in prejudicial outcomes in comparison with the general population (White & Perrone, 1997). Complicating matters is that individuals from racialized communities are disproportionately represented at all levels of the criminal justice system making “these communities vulnerable to mental disorders and addictions, as well as criminalization” (Chandrasekera, 2011, p. 11).

Criminalizing the mentally ill refers to arresting and charging PMIs for relatively minor offenses and deviant behavior symptomatic of persons requiring mental health care when this may not occur in similar encounters with non-PMI. To date, there is little agreement on whether PMI are more likely to be charged or issued a citation and this gap in knowledge persists despite research examining encounters involving violent offending (Adelman, 2003; Lamb et al., 2002), intoxicated persons (Novak & Engel, 2005; Watson et al., 2010), when citizen actions are offensive to a reasonable person (Hiday & Burns, 2009; Teplin, 2000), and with disrespectful, noncompliant individuals (Engel & Silver, 2001; Kesic, Thomas, & Ogloff, 2013; Novak & Engel, 2005; Watson et al., 2010). Arrests are seen as more expedient and justified when the citizen is declared inadmissible at the hospital (e.g., not a sufficient danger to self or others), no alternative options exist, citizen behavior exceeds the community’s tolerance, or officers feel the citizen continues to pose a problem if not dealt with by way of arrest (Bittner, 1967; Holley & Arboleda-Florez, 1988; Lamb et al., 2002; Royal Canadian Mounted Police, 2010; Teplin, 2000). Officers feel they have less discretion and decision-making ability with dispatched calls and will lean toward an arrest to restore order (Patch & Arrigo, 1999; Slate & Johnson, 2008). In fact, the probability of an arrest can be 67% or higher for citizens endorsing symptoms of a mental illness, particularly when deciding upon dispositional outcomes for minor or public order offenses (Cordner, 2000; Godfredson et al., 2010; Novak & Engel, 2005), yet others find no statistically significant relationship (Bonovitz & Bonovitz, 1981; Engel & Silver, 2001; Teplin, 2000; Watson, Corrigan, & Ottati, 2004).

Police officers typically do not view mentally ill persons in a negative way and are often more empathetic than the general public (Price, 2005; Slate & Johnson, 2008; Watson & Angell, 2012). Negative views enter in respect to the mental health care system, which is part of the decision-making justification used when resorting to arrests and mercy bookings because officers contend with long hospital wait times, low probabilities of civil commitments, and a citizen in need of mental health treatment (Lamb et al., 2002). In this way, a strong argument can be made that officer perspectives on mentally ill persons and subsequent use of discretion is “embedded within the decision making process” (Patch & Arrigo, 1999, p. 33).

Although arrests of PMI are lawful, it is not always the best course of action because it can “exacerbate the difficult life circumstances facing the mentally ill” (Heslop et al., 2013, p. 6). However, the criminal justice system is viewed by the police as the “de facto mental health system [in our] country, [despite the] widespread consensus that incarcerating these individuals is . . . inappropriate and counterproductive” (Hafemeister, Garner, & Bath, 2012, p. 155; see also Cotton & Coleman, 2010). A slippery slope is created as arrests are viewed as justifiable when citizens are perceived as dangerous to themselves or others, under the influence, or disrespectful, which are often behaviors symptomatic, or at the very least commonly seen in association with, mental illness (Cooper, Mclearen, & Zapf, 2004). This is particularly acute with additional challenges faced by officers when PMI have a dual diagnosis of substance abuse. Being released on a promise to appear with a condition not to consume alcohol or drugs perpetuates a revolving door in and out of the criminal justice system. Frustration is expressed as these PMI are “not welcome at hospitals due to their substance use or at drug and alcohol treatment facilities due to their mental illness” (Watson et al., 2010, p. 303; see also Teplin & Pruett, 1992). The above reasons underscore the need to gain a better picture of what occurs during encounters with PMI.

Training

It is increasingly apparent that patrol officers feel they are inadequately trained to respond to calls involving PMI and persons in crisis, which arguably is a contributing factor to criminalization of the mentally ill (Borum, 2000; Cooper et al., 2004; Godfredson et al., 2011; Iacobucci, 2014; Lamb et al., 2002; Ruiz & Miller, 2004; Slate & Johnson, 2008; Teplin, 2000). Gaining knowledge on recognizing the signs of mental illness, how it can affect thought processes and behavior, de-escalation, suicide warning signs, when to involve specialized officers and community resources, and those PMI most appropriate for diversion out of the criminal justice system, could all potentially increase confidence and feelings of possessing the requisite qualifications (Godfredson et al., 2011; Lamb, Weinberger, & Gross, 2004; Lurigio, 2000). It continues to be the case that the training recruits receive is too general, and it must start to directly address stereotypes and stigmas embedded in the police culture (Iacobucci, 2014; Kappeler & Gaines, 2011).

The police culture consists of values and attitudes held by officers, which is made visible through interaction and the direct or indirect pressures to conform to norms of behavior. An investigation into the use of force with PMI by the Toronto Police Service led to a conclusion that “culture eats training” (Iacobucci, 2014, p. 117). To be effective, mental health training must work toward developing new attitudes in the local police culture or conflicts in the field between policy and practice are inevitable. Due to the strong cultural influence, general training for patrol officers has similar effectiveness as deploying CIT officers when it concerns changing knowledge, attitudes, and behavior, particularly if training is decentralized at the platoon level (Hatfield, 2014; Iacobucci, 2014). Furthermore, it is understood that even when all officers receive the same training, this does not necessarily mean the same ability to interact with PMI and de-escalate crisis or violent situations. By extension, it becomes important to examine what citizen and encounter characteristics do affect decision-making so the results can inform training curricula to better address this cultural divide.

The Current Study

This study addresses the call by Novak and Engel (2005) for a dual approach by adopting a sequential mixed methods research design to analyze quantitative data on factors affecting dispositional outcomes and subsequently qualitative data to elucidate the decision-making process and constraints leading up to these outcomes. By adopting this approach, the current study is well positioned to contribute to ongoing discourse because the static outcomes of a police–citizen encounter are contextualized within the dynamic decision-making process. The three research questions investigated are as follows:

Research Question 1: How do the characteristics and behavioral dynamics of police–citizen encounters differ for PMI?

Research Question 2: What factors affect the use of discretion when responding to criminal and noncriminal behavior?

Research Question 3: How is the decision-making process affected by procedural challenges when responding to PMI-related calls for service?

Method

For the first time in Canada, this project collected systematic social observation (SSO) data on police decision-making and the use of discretion. The only other Canadian SSO project focused exclusively on the frequency and proportion of time officers spent engaged in work and non-work-related activities (Plecas et al., 2012). In this sequential mixed methods research design, the quantitative data strand is analyzed first while the qualitative SSO data provide the context for the quantitative findings.

Participants

At the time of data collection, the regional police service had three urban and two rural divisions² serving a population of 536,793 in 2012 with a crime rate of 4,524 per 100,000 population, which is lower than the national average of 5,588 (Perreault, 2013). The sample consisted of 74 ride alongs, 402 encounters, and data on 606 citizens.³ Three quarters of the citizens were male (76%) and just under half have a criminal record (45%). Police officers in the sample were predominantly male (81%), with an average of 6.9 years of service (min. = 1 month, max. = 288 months, SD = 71.06). To adjust for a skewed distribution, the interquartile range indicates just over half had 4.5 years of service or less. The analysis includes only those citizens identified as a suspect where the police are in a position to issue a citation or lay a criminal charge (N = 577). Demarcating by the most serious offense under investigation, 317 cases were criminal, and 260 were citation-related offenses. The sample consisted of individuals 18 years of age and older. Thus, 24 encounters involving 29 citizens were excluded from the analysis as they were juveniles between 12 and 17 years of age. Of note, persons under investigation for noncriminal behavior were included as they can be subject to an arrest, and when it comes to mentally ill persons, the dispositional outcomes can be similar (e.g., less discretion).

Materials

Three data collection instruments from the Project on Policing Neighborhoods (Interuniversity Consortium for Political and Social Research Study No. 3160) were modified for the Canadian context (see Schulenberg, 2014, 2015).⁴ First, the ride along form recorded data on the officer and shift (20 variables; for example, duration, sex, years of service, division, weather). Second, the encounter form collected data on the police–citizen interaction recording the context of decision-making (54 variables; background information, for example, type of offense; observation characteristics, for example, bystanders; police actions, for example, supervisory presence, arrest). Finally, a citizen form collected data on personal background (e.g., age, sex, criminal history), behavior (e.g., substance use, demeanor), actions (e.g., requesting information), and police actions (e.g., restraint; 104 variables). Narratives captured all events and behaviors associated with each person and encounter. The field notes included remarks on reflexivity, setting, nonencounter interactions, debriefing, and contextual detail on interactional processes during the encounter.

Procedure

The author conducted 637 hr of ride alongs with patrol officers in all divisions from November 2011 to June 2012. Observational hours did not include time in roll call, cell block, or other activities while not on active patrol. Ride along data were collected on the early (1:30-11:30 p.m.) or late afternoon shifts (5:00 p.m.-3:00 a.m.) from Thursday to Sunday. Although ride alongs 4 days a week on busier shifts increases exposure to PMI-related calls for service, caution is required as the results cannot be generalized to police work overall at this agency because purposive temporal sampling was used to select each ride along day and shift. Although divisions and platoons were randomly selected, this was not possible for patrol zones and officers for operational reasons. With whom the researcher rode was at the discretion of the staff sergeant who took into consideration previous zone assignments at that division and officers voluntarily agreeing to participate in the research study.

The ride along, encounter, and citizen forms were printed in advance and filled out in the car after each call while the officer completed his or her paperwork. To establish trustworthiness of the findings, steps were taken to ensure rigor and consistency by assessing dependability in the same way quantitative researchers seek to ensure reliability (Shenton, 2004). Although the data analysis of the field notes and narratives used the constant comparison method, the intrarater reliability test for dependability was employed. Qualitative coding occurred over a period of 5 weeks. For every 50 encounters coded, random field notes from earlier periods of time were recoded achieving an intrarater reliability level of agreement ranging from 91% to 97%. Quantitative reliability was assessed in a similar fashion by recoding prior encounter and citizen forms with agreement ranging from 96% to 99%.

Dependent Variable

In Canada, there are conceptual and legal reasons for operationalizing discretion as the decision to lay a criminal charge or issue a citation for municipal or provincial offenses.⁵ A person who is stopped and questioned by the police might be under arrest during the questioning, but may or may not be charged at the conclusion of the encounter. Furthermore, encounters can involve offenses where arrest is not a dispositional option, but charges are laid (see Section 495 of the Criminal Code [1985]).⁶ In practical terms, this means it is possible for a citizen to be placed under arrest for noncriminal behavior (i.e., not contrary to the Criminal Code, 1985, such as intoxication in a public place). An officer has the discretion as to whether a citation is issued and whether the citizen is arrested. The inverse holds true as an arrest may not be lawful for some criminal offenses. Thus, consistent with prior Canadian research, the dependent variable is whether an officer lays a criminal charge or issues a citation when reasonable grounds exist to do so (Carrington & Schulenberg, 2003; Perreault, 2013; Schulenberg, 2015).

Independent Variables

Encounter Characteristics

An argument can be made that with more supervisory and record keeping scrutiny, officers are using less discretion on reactive dispatched calls than those encounters that are proactive (officer-initiated). Thus, source is included in the model and coded 1 for dispatched calls. Offense seriousness is operationalized using Statistics Canada’s Crime Severity Index. If more than one offense was under investigation, the most serious offense as defined by the Uniform Crime Reporting Survey is coded. In this index, a weight is assigned to each offense based on its relative seriousness in comparison with other types of crime.⁷ A log transformation is used to induce normality due to the presence of influential outliers. Some calls for service involve serious violent offenses while the majority are within a lower crime severity weight range (min. = 1, max. = 583, $\bar{x}$ = 54.06, SD = 76.30).⁸ Consistent with Brown et al. (2009), the common predictor of situational evidence (quantity and quality) is excluded as all cases had at least one piece of evidence supporting reasonable grounds (physical, witness, confession) to qualify for inclusion in the sample. The presence of bystanders is also omitted because only 9.2% of encounters involved a bystander, and there are no statistically significant correlation coefficients with the dependent or independent variables.

Citizen Characteristics

Sex is a dichotomous variable with males coded as 1. Citizen age is measured in years (min. = 18, max. = 88, $\bar{x}$ = 32.96, SD = 13.48). A log transformation is performed to induce a normal distribution. Mental health status is based on information from the officer in debriefing after the police–citizen interaction, the dispatch narrative indicating mental health status, a notation in the records management system (RMS), or a Flag M for mental illness listed on the Canadian Police Information Centre (CPIC).⁹ Although concerns can be raised on the ability of an officer to determine whether a person is mentally ill, “if the goal is to understand officers’ decision-making, then officers’ perceptions of mental disorder are more relevant than are classifications based on clinical criteria” (Engel & Silver, 2001, p. 236). Officer identification of mental health status without corroboration from other sources rarely occurred. In these few cases, their assessments were based on strong situational evidence (e.g., confirmation by family members).¹⁰ Data were captured as continuous variables for the number of prior violent, property, drug, administrative offenses, weapons, and other convictions (min. = 0, max. = 57, $\bar{x}$ = 3.21, SD = 7.45). One in 10 citizens is an influential outlier (9.7%), resulting in the creation of a dichotomous variable with 1 designating at least one prior conviction for a criminal offense.

Situational Factors

Under the influence of alcohol or drugs is coded as 1 if there are behavioral indications of intoxication as confirmed by the officer. Disrespect toward the officer is coded as 1 when at any time during the encounter citizens describe the officer using profane language, engage in deliberate actions (e.g., spitting in the direction of the officer), overtly display disregard for officer authority, or make rude remarks when responding to an officer’s questions (Engel, Tillyer, Klahm, & Frank, 2012; Klinger, 1996; Worden & Shepard, 1996). Coding was confirmed with the officer during debriefing. A citizen is coded as 1 if uncooperative. This concept is defined as any time a citizen is noncompliant with officer directives for behavioral change or requests for information (Engel et al., 2012; Worden, Shepard, & Mastrofski, 1996).¹¹

Analytic Plan

To address Research Question 1, frequencies for binary variables were calculated and the chi-square statistic used to assess whether the differences between PMI and non-PMI were statistically significant. For continuous variables, independent sample means t tests were used to determine if the differences in means were statistically significant.

For Research Question 2, the factors affecting the use of discretion were analyzed using three binary logistic regression models: all offenses, criminal, and citations.¹² Due to the small sample size, a nonparametric bootstrap procedure was performed to produce more robust statistical inferences for all three models. With this procedure, observations were resampled from the data with replacement for 1,000 replications to account for distortions in standard errors and confidence intervals.

Finally, to address Research Question 3, a thematic analysis was conducted of the ride along field notes. In first cycle coding, text segments were assigned to descriptive codes providing important contextual information on key concepts, what officers perceived as important, and how these fit into the encounter context. Process codes were created for data relevant to decision-making. This consisted of information on how an officer decided upon a course of action and how he or she goes about executing that action. Finally, emotion codes describe the nature, scope, and circumstances of officers’ feelings in respect to statements or reactions to events, actions, situations, or people. For example, text segments assigned to the emotion code “frustration” include an officer talking about being disrespected by suspects or calls for service affecting them emotionally (e.g., suicide, child abuse). Segments were assigned to only one code. If thematic overlap occurred, this was documented through the use of annotations and memos. These were treated as data during second cycle coding where codes were grouped into categories to identify patterns in the data. Comparisons were made within and across 34 codes with attention directed to causes, explanations, frequency, and substantive significance in relation to the research questions.

Results

The logistic regression analyses find that mental illness is only a predictor for issuing citations, and, regardless of mental health status, a citizen who is under the influence of alcohol or drugs increases the likelihood of criminal charges and citations. A thematic analysis reveals unique patterns in police response strategies, including behavioral indicators officers use as criteria for determining mental health status, procedural challenges with limited resources or information, and complications endemic to PMI encounters that culminate in constraints on officers’ decision-making autonomy.

Comparison of Factors by Mental Health Status

To address Research Question 1, Table 1 presents the results for the analysis of factors affecting officer behavior to ascertain whether statistically significant differences exist between PMI and non-PMI-related calls for service. Notwithstanding the type of offense, 76% of the calls for service involving a PMI were dispatched in comparison with 52% for non-PMI, but there were no differences in the presence of bystanders. Not surprising is that encounters with PMI were on average 20 min longer. More than one third of PMI were under investigation for a violent offense (38% vs. 15%) with no statistically significant differences for other types of offenses. Finally, consistent with the criminalization of the mentally ill hypothesis, PMI-related encounters involve less serious offenses (2.70 vs. 3.14).

Table 1:

Comparison of Factors Affecting Police Discretion by Mental Health Status

	Non-PMIM (SD)	PMIM (SD)	t
Encounter duration	46.25 (42.88)	65.63 (63.96)**	t = −3.21, p = .006
Offense seriousness	3.14 (1.55)	2.70 (1.56)*	t = 2.37, p = .019
Age	31.78 (12.97)	38.80 (14.11)***	t = 4.36, p < .001
	Non-PMI%	PMI%	χ²
Encounter characteristics
One or more bystanders	49.2	38.9	χ² = 2.84, p = .092
Offense under investigation
Violent	15.3	37.5***	χ² = 19.98, p < .001
Property	17.8	20.8	χ² = 0.41, p = .520
Public order	23.7	24.0	χ² = 0.01, p = .964
Citizen characteristics
Sex	76.3	72.9	χ² = 0.41, p = .521
Prior criminal record	38.9	75.5***	χ² = 36.02, p < .001
Violent conviction(s)	24.0	51.1***	χ² = 22.56, p < .001
Property conviction(s)	26.1	54.3***	χ² = 23.69, p < .001
Administration of justice	18.4	43.6***	χ² = 22.43, p < .001
Citations	28.2	44.7**	χ² = 8.24, p = .004
Flags listed on CPIC	22.2	55.3***	χ² = 33.92, p < .001
Violent/assaultive	18.2	50.0***	χ² = 34.64, p < .001
Situational factors
Under the influence	29.7	52.1***	χ² = 14.87, p < .001
Disrespect	25.0	30.2	χ² = 0.95, p = .330
Uncooperative	23.7	34.4*	χ² = 3.94, p = .047
Officer actions
Conversational request	79.7	70.8*	χ² = 3.30, p = .046
Provide information/advice	64.4	63.5	χ² = 0.02, p = .882
Check citizen well-being	11.0	35.4***	χ² = 27.44, p < .001
Provide assistance	29.2	61.5***	χ² = 29.91, p < .001
Referral to agency/citizen	19.5	35.4**	χ² = 9.46, p = .002
Search	17.4	32.3**	χ² = 8.94, p = .003
Investigative detention	28.8	41.7*	χ² = 5.14, p = .023
Arrest	9.3	21.9**	χ² = 9.54, p = .002
N = 577	n = 432	n = 145

Note. PMI = persons with a mental illness; CPIC = Canadian Police Information Centre.

p < .05. **p < .01. ***p < .001.

Although there are no differences by sex, PMI are on average 7 years older, over half have CPIC flags (55% vs. 22%), and more than twice as many have a flag for violent/assaultive behavior. The majority of PMI in contact with the police have a prior record (76% vs. 39%) with slightly over half for violent offenses and just under half for citations. To provide an example of a prior record, a homeless PMI with a substance abuse problem had 66 prior citations for intoxication in a public place and trespassing of which most involved an arrest.

Examining situational factors known to affect the use of discretion, more PMI were under the influence of alcohol or drugs than non-PMI (52% vs. 30%). The findings support the need to unpack the concept of demeanor because there was no statistically significant difference for disrespect shown toward the officer. However, a statistically significant difference exists as PMI are more often noncompliant with officer requests and directives (34% vs. 24%).

With non-PMI citizens, officers are slightly more inclined to make conversational requests for information (80% vs. 71%) but no more likely to offer advice. In addition to handling the offenses under investigation, 25% of PMI-related encounters (vs. 11%) had the officer checking the citizen’s well-being and making referrals (35% vs. 20%). Despite inconsistent patterns for situational factors, greater proportions of PMI receive assistance (62% vs. 29%) but are at the same time subject to actions more invasive of personal liberty (e.g., investigative detention, search, arrest). Most interesting is the lack of a statistically significant difference for the use of discretion.¹³ The results for Research Question 2 shed some light on this finding once the analysis controls for criminal charges versus issuing a citation.

Factors Affecting the Use of Discretion

To address Research Question 2, Table 2 presents the results for factors affecting the use of discretion when investigating all types of offenses, χ²(9, N = 577) = 99.97, p < .001, Nagelkerke R² = .21. Officers are more likely to use their discretion at dispatched calls. When looking at citizen characteristics, officers are more likely to formally sanction males but less likely to do so with younger citizens. Prior to demarcating by type of offense, mental illness and a prior record are not statistically significant predictors. However, officers are 2.4 times more likely to charge or issue a citation if the citizen is under the influence of alcohol or drugs and 1.7 times if disrespectful of the officer’s authority. Although in the expected direction, when citizens are noncomplaint with officer directives, this does not affect the likelihood of using discretion.

Table 2:

Logistic Regression for Using Police Discretion With All Offenses

	β	SE	Wald	Exp(β)	95% CI for Exp(β)
	β	SE	Wald	Exp(β)	Lower	Upper
Encounter characteristics
Source	−.79***	0.21	14.54	0.45	−1.27	−0.35
Offense seriousness	.27***	0.06	20.79	1.32	0.16	0.41
Citizen characteristics
Sex	.85***	0.20	17.18	2.33	0.44	1.25
Age	−.41*	0.21	3.93	0.67	−0.81	−0.01
Mental illness	.32	0.23	1.94	1.38	−0.15	0.86
Prior criminal record	.06	0.21	0.07	1.06	−0.35	0.46
Situational factors
Under the influence	.89***	0.23	15.57	2.44	0.46	1.42
Disrespect	.51*	0.25	4.09	1.67	0.03	1.06
Uncooperative	.41	0.23	3.18	1.51	−0.04	0.93

Note. Bootstrap results are based on 1,000 replications. CI = confidence interval.

p < .05. **p < .01. ***p < .001.

Table 3 presents the results for those cases where the most serious offense under investigation is for criminal behavior, χ²(9, n = 317) = 64.64, p < .001, Nagelkerke R² = .25. Dispatched calls are no longer a significant predictor. As expected, officers are 2.1 times more likely to criminally charge a citizen the more serious the alleged offense. The relationship between discretion, males, and age continues to be significant when responding to criminal behavior. As expected, officers are 2 times more likely to lay a charge if the citizen has a prior record. Being under the influence of alcohol and drugs continues to be a strong predictor (2.6 times more likely to charge). Disrespect is not statistically significant along with mental health status. However, uncooperative citizens are 2.2 times more likely to be charged.

Table 3:

Logistic Regression for Using Police Discretion With Criminal Offenses

	β	SE	Wald	Exp(β)	95% CI for Exp(β)
	β	SE	Wald	Exp(β)	Lower	Upper
Encounter characteristics
Source	.27	0.36	0.56	1.30	−0.54	1.05
Offense seriousness	.63*	0.13	4.61	2.14	−0.40	0.76
Citizen characteristics
Sex	.79**	0.28	7.72	2.20	0.27	1.41
Age	−.84**	0.30	8.10	0.43	−1.50	−0.25
Mental illness	.42	0.99	1.97	1.52	−0.25	1.12
Prior criminal record	.98**	0.26	6.73	1.98	0.18	1.27
Situational factors
Under the influence	.95**	0.30	9.90	2.59	0.27	1.60
Disrespect	−.06	0.34	0.03	0.94	−0.86	0.67
Uncooperative	.80**	0.31	6.79	2.23	0.21	1.48

Note. Bootstrap results are based on 1,000 replications. CI = confidence interval.

p < .05. **p < .01. ***p < .001.

As seen in Table 4, the factors affecting the use of discretion differ for the decision to issue a citation, χ²(9, n = 260) = 35.58, p < .001, Nagelkerke R² = .26. In contrast to criminal behavior, dispatched calls decrease the odds of a citation. Consistent with prior findings, the more serious the behavior under investigation, the more likely formal social control is used by officers to restore order. Although younger citizens continue to be less likely to be issued a citation, a significant departure is the fact PMI are 3.3 times more likely to be given a citation regardless of prior record. This finding supports the criminalization of the mentally ill hypothesis as they are more likely to be subject to formal social control for minor, noncriminal behavior that can nonetheless result in an arrest and custody. The relationship is even stronger for under the influence of alcohol or drugs where officers are 3.7 times more likely to issue a citation. If citizens are disrespectful of the officer, they are 2.8 times more likely to be given the citation, whereas this was not a statistically significant predictor for criminal charges. As one officer said when using his discretion, “He didn’t fail the attitude test, lives a few doors down from my uncle, so he’s close to home and not drunk, therefore, not my problem.” Finally, the second indicator of demeanor, uncooperative citizens, is not significant.

Table 4:

Logistic Regression for Using Police Discretion With Citations

	β	SE	Wald	Exp(β)	95% CI for Exp(β)
	β	SE	Wald	Exp(β)	Lower	Upper
Encounter characteristics
Source	−1.36**	0.45	9.34	0.26	−2.69	−0.47
Offense seriousness	0.31*	0.14	4.78	1.36	−0.01	0.83
Citizen characteristics
Sex	0.37	0.36	0.65	1.45	−0.77	1.45
Age	−1.10*	0.38	5.30	0.33	−2.41	−0.14
Mental illness	1.19*	0.37	2.98	3.29	−0.12	2.40
Prior criminal record	0.11	0.35	0.04	1.11	−0.97	1.61
Situational factors
Under the influence	1.32*	0.36	4.90	3.75	−0.08	3.38
Disrespect	1.03*	0.38	2.61	2.81	−0.42	3.50
Uncooperative	0.80	0.37	1.34	1.23	−0.36	4.58

Note. Bootstrap results are based on 1,000 replications. CI = confidence interval.

p < .05. **p < .01. ***p < .001.

Decision-Making Process in PMI-Related Encounters

The findings for Research Question 3 unpack the decision-making process, and the results fall into three larger themes of procedural challenges, encounter dynamics, and complications based on characteristics or situations unique to PMIs. As will be made clear, officers not only acknowledge the unique circumstances of PMI-related calls but also experience constraints on their decision-making autonomy as a result of external factors beyond their control.

Procedural Challenges

The procedural challenge theme consists of three dimensions: determining mental health status, coping with limited community resources, and the growing disjuncture between departmental policy and the police culture. The first procedural hurdle is that officers contend with a critical systemic problem. In the sample, 23% of PMI were not flagged mentally ill on CPIC or in the RMS despite possessing a prior criminal record or extensive police contacts. For instance, a 32-year-old male, paranoid schizophrenic with delusions at a disturbance call did not have a Flag “M” despite many prior police contacts. The disturbance was determined to be unfounded as were the last 27 calls, which were changed to mental health calls, the six prior noise complaints, and two voluntary Ontario Mental Health Act (MHA) apprehensions.

These flags are important during the police–citizen encounter. Officers verbally noted the flag and then used this information to adjust forms of engagement from conversational to confrontational. For instance, one call involved a 21-year-old, with a substance abuse problem, six flags (contagious, escape risk, family violence, mentally ill, suicide risk, and violent), and a criminal record for assault, breach of probation, and failure to comply with a disposition. The officer said, “The last time she was in custody she attempted suicide in the cell block.” The officers characterize her as a known drug user, and she is often seen intoxicated in the downtown core. In this case, the officer used the flags strategically and engaged communication in a very nonthreatening, conversational manner. He used his discretion and did not arrest or issue a citation for trespassing.

One might assume that the informant or complainant would be a reliable source of information; however, officer requests can end in one of two ways. On one hand, they can be cooperative and provide background information, context, warn the officer about the PMI’s history of behavior, or express concern about the PMI’s well-being and mental stability. On the other hand, they were uncooperative, had nothing positive to say about the PMI, and requested removal or an arrest. As a result, officers use behavioral cues to determine a citizen’s mental health status. Indicative characteristics include the citizen stopped taking his or her medication or is a known drug user. Officers were alerted with previous contacts for intoxicated person, disturbances, prior threats or attempts of suicide, and if they had prior interactions with the citizen. The latter is more common with citizens under the influence of alcohol or drugs, the homeless, and sex trade workers. Finally, negative behavior is considered including being uncooperative when asked questions, refusing to identify themselves or provide information on their psychiatric history, failing the “attitude test,” and when disagreements arise on whether the police should be present.

Second, it became clear there are limited community mental health resources available to officers, particularly after-hours. Nine encounters involved calls to the Mobile Crisis Unit and the phone call went unanswered or it was decided the citizen isn’t in “enough crisis” to warrant an in-person visit. Further investigation revealed inadequate staffing with only two mental health workers on call on any given day despite a population size of over 500,000. This leaves officers with few options other than taking no further action or at the other end of the continuum citations and charges.

Finally, there is the issue of conflicting messages from departmental policy and culture. There is evidence of indirect and direct pressure to resolve PMI-related calls in an expedient manner. Indirect pressure involved messages from the staff sergeant communicated by dispatch on the computer terminal or on rare occasions over the air. Expediency can be a form of self-preservation. A rationale provided for informal dispositions is that arrests require significant paperwork and time lost with 911 calls waiting. However, the direct pressure is due to a conflict arising with the “monthlies” that primarily count arrests, citations, and charges as evidence of productivity. For those officers who stated they “feel sorry” for the mentally ill or expressed a genuine concern and desire to be part of a solution were upset about the incongruence and the corresponding impact on their use of discretion.

The police culture supported informal dispositions; yet departmental policy on evaluating performance is not seen as giving credit for these dispositional outcomes. This is openly reinforced further in the police culture. One of the first roll calls each month starts with the announcement of the platoon’s statistics and how they ranked in comparison with other platoons. Directing comments toward two officers, one staff sergeant commends them for the number of arrests and citations by saying, “your hard work gets noticed.” On a different shift, formal social control is encouraged by a staff sergeant and inspector by instructing officers to have at least 10 citations by the end of their shift on St. Patrick’s Day. An officer laments a very common theme of “I’m tired of hearing about the monthlies which have nothing to do with real police work.”

Police–Citizen Interaction Dynamics

The police–citizen encounter with PMIs has two facets of behavioral dynamics and response strategies. Officers perceive three behavioral dynamics and processes unique to PMIs for both criminal and noncriminal calls for service. First, observable characteristics include being under the influence of alcohol, poor hygiene, indications of drug use (e.g., hyper, physical ticks, rapid speech), disoriented, scattered narrative, repetitive, refusing assistance, or despondent. As an officer with 7 years of service said, “Those are the ones we always seem to be dealing with because they’ll stop taking their medication and start thinking they’re okay not to take it and their behavior escalates.”

Second, officers describe PMI as potentially unpredictable due to irritability, mood shifts, aggressive behavior, or attention seeking behavior. Physical indications include volatility (e.g., ripping off clothes), kicking or spitting at the officer, and throwing belongings. Verbal behavior includes yelling, use of profanity, aggressive language, uttering threats about repercussions for police actions, or deliberately and repetitively lying about history, facts, and events.

Finally, when discussing the decision-making process, the words and behavior that occur during the investigation process are emphasized by the officer more for PMIs. These include denial, evasiveness, excuses for behavior, becoming defensive about the amount of alcohol consumed, or a belief they are smarter than the police. For many officers, these behaviors were associated with a disrespectful demeanor. In these cases, officers typically used less discretion when responding to noncriminal behavior.

When considering police response strategies, officers adopted a restrained and calm approach focusing on rapport and de-escalation while seemingly dismissing disrespectful behavior when investigating criminal behavior. For instance, one call for service involved an Aboriginal, 33-year-old male, flagged mentally ill and violent, of special interest to police, on the sex offender registry, and who had a history of prior sex and mischief convictions. The citizen made a rude gesture in the officer’s direction, and he replied, “You wanted my attention and now you’ve got my attention.” Despite describing the citizen negatively during debrief, the officer used his discretion and he gave the citizen a verbal warning. Verbal tactics when encountering disrespect include officers in a calm and steady voice repeating statements, advice, and offering reassurance with statements such as, “It’s okay. You are not alone.” “This is the same advice I would give anyone.” “No need for that.” “Don’t worry. You’re doing great.” Advice offered were recommendations to seek substance abuse treatment, encourage the person to set goals for themselves, take prescribed medication, quit using drugs, and to leave a volatile relationship.

Within the above context, there is a generic sequence of events in PMI-related encounters. When the label of mentally ill is activated, verbal and physical containment occurs. Containment is not the first action with non-PMI unless there is verbal or physical aggression. This is followed by the use of de-escalation techniques, officers check well-being, ensure the safety of all parties, and use investigative detentions. The complainant or witness’ side of the story is gathered first, and then officers ask the PMI for their account of events. Additional record searches are conducted to verify versions of events and history. Verbal warnings issued to non-PMIs are typically stern and include statements on the ramifications of their actions and the consequences if the behavior continues. In contrast, PMIs receive warnings in the form of an admonishment similar to warnings given to juveniles (e.g., “smarten up”) rather than offense-related statements. Arrests and criminal charges are subject to more directives in the show cause court packet, and receive a higher number and more restrictive conditions for release (e.g., not attending a location, associate with certain people, and abstain from alcohol and drugs) despite a recognition of the “revolving door syndrome.” For both PMI and non-PMI, multiple charges are laid if reasonable grounds exist to do so to ensure “one sticks”; however, officers justified arresting, charging, issuing citations to PMI more often when perceived as “rabble-rousers” out to cause trouble for the sake of trouble, or “problem children” who played up their mental illness as an excuse for their behavior. Finally, officers routinely made recommendations for the type of court intervention, whether mental health court is appropriate, and explained the need for additional release conditions on a promise to appear to avoid a repeat call for service that shift. In the case of PMI, it appears this is viewed as less problematic as they “just might get the help no one else is giving them.”

Unlike the quantitative findings, the decision-making process does not support the use of a zero-tolerance approach to intoxication in a public place, urination in public, trespassing on private property, and similar infractions. Officers are much more likely to work toward a resolution that ceases the behavior rather than criminalization. Police responded to a disturbance call that involved a homeless 25-year-old male, flagged mentally unstable, and released on bail. Upon arrival, he ripped off his jacket, hoodie, and turned his pockets inside out while throwing his wallet and other personal effects all over the sidewalk. He was extremely belligerent, disrespectful, used excessive profanity, and unpredictable. A background check revealed he was the son in a famous case of neglect and abuse after his mother was murdered. The officer used his discretion and did not issue a citation for disturbing the peace, breach of probation, and public intoxication. On average, the circumstances where discretion is less likely to be used is when citizens are perceived as a danger to themselves or others, under the influence, or disrespectful, which are all behaviors that are potentially interpreted as symptomatic of a mental illness. In these cases, officers hold divergent opinions. “Just because we’re cops doesn’t mean we’re better than them. At the same time, today we get along well, but tomorrow he might want to shoot me.” Or take the officer who believes “mental health is an illness, not a choice. He’s in a ‘black hole’ . . . would you be angry with him if he had cancer? No.” At another call, the hospital nurse warns the officer she hates police and will punch him hard because she’s a big girl. The officer comments, “I just assume everybody is going to punch me in the face.”

What are the Complications?

Three themes highlight the challenges faced by officers in PMI-related encounters and the barriers created for using informal dispositions. First, over half of PMI were under the influence of alcohol or drugs (52%). If they are not too intoxicated, willing, and space is available, they are transported to a detoxification facility. If they are too intoxicated, the PMI is transported to the hospital by paramedics. Although rare, if a responsible party is available and willing, custody is transferred. Virtually all of the officers observed made clear that their goal is to use cell block as a last resort for PMI who are issued a citation for intoxication in a public place. The challenge officers often face is a situation where they feel there is no alternative, but to arrest an intoxicated PMI and take him or her into custody for 6 hr until sufficiently sober (67.4% in custody were under the influence; χ² = 20.96, p < .001). As a male officer summarized,

Arresting him for intoxication in a public place for the safety of himself was a last resort. He couldn’t stay at the [homeless shelter], he wasn’t drunk enough for transport to [the hospital], too drunk for [church shelter], and there is no room in detox for him. Detox says they have room if he is stable, able to walk, and not being aggressive. I’m out of luck.

Second, the Ontario Mental Health Act (1990) apprehensions are no simple affair. The police are concerned with two primary issues: (a) the protection, safety, and welfare of the public; and (b) acting in the role of parens patriae to protect members of the community who are unable to care for themselves. Officers talked at length about the impact on their ability to respond to 911 calls when facing extended “babysitting” time at the hospital sometimes for their entire shift ( $\bar{x}$ = 5.27 hr, SD = 2.41). This often led into discussions about PMI typically not meeting the civil commitment criteria. This leaves officers with the options of an informal disposition that does not ensure safety, or criminal charges that contribute to the problem of criminalizing the mentally ill.

Considerable frustration is voiced about what should happen if the citizen is not committed. Several officers expressed how hard it was to deal with their emotions when PMIs are released and successfully end their lives days later. They felt the criteria for a civil commitment are too stringent and put too many PMI at risk. The hospital “passes the buck and responsibility to us.” Officers indicated this is a prime example of when they see arrest and charges as a more practical, expedient, and humane alternative. Officers feel their assessment is ignored by doctors, the PMI “knows exactly how to answer the questions,” and then “the doctor decides it’s just a temper tantrum.” Thus, they lay a criminal charge as a backup “so there are consequences regardless of a person’s mental health status,” whether they are perceived as a danger to themselves or others, or due to a belief that the criminal justice system is the only way the PMI will get the mental health assistance required. Unfortunately, this represents a further contribution to the criminalization of the mentally ill as charges are rationalized to be in the best interests of the PMI. Many officers shared that they were inclined to see referrals as futile and invoke formal social control as the backup.

Finally, there is an omnipresent challenge of determining the best way to resolve encounters with persons who have no fixed address. Encounters observed with a homeless person were primarily for criminal offenses related to survival (e.g., theft) or offenses officers articulated were frequent with mentally ill homeless persons (e.g., possession of narcotics, public intoxication). Two thirds of homeless persons have a history of mental illness, the majority a substance abuse problem, and have a higher rate of police contact than other PMI. This challenge became particularly acute when deciding on whether to arrest and detain for intoxication. An address is required to be released after a 6-hr hold. Several of the homeless shelters allow their address to be used, but this was not always a guarantee.

Discussion

On one hand, the findings suggest officers adopt a more social services approach because they are more likely to check a PMI’s well-being, provide assistance, and make referrals regardless of the type of behavior under investigation. On the other hand, evidence exists to support the criminalization of the mentally ill hypothesis as PMI are more likely to be under investigation for less serious offenses and subject to commands for information, searches, investigative detention, and an arrest. Criminalization occurs when issuing citations or placing citizens under arrest for social welfare reasons, but officers acknowledged that in an ideal world, this is not the most appropriate outcome to the encounter.

As expected, offense seriousness, age, and being under the influence of alcohol and drugs were strong predictors of the use of discretion. Where the findings depart from prior research is when examining the two dimensions of demeanor. There is no statistically significant difference in the proportion of PMI who were disrespectful, but a significant difference for those seen as uncooperative. When controlling for other factors, disrespect and noncompliance were not significant predictors of the decision to lay a criminal charge, but disrespectful behavior and mental health status were strong predictors on the likelihood of a citation.

An indirect systematic procedural bias offers a preliminary picture of decision-making practices and constraints that can lead to prejudicial outcomes for PMI. Officers are often forced to rely on behavioral cues to determine mental health status as CPIC flags are inconsistent as is information provided by the informant or complainant. Although the quantitative data provide evidence for criminalization of minor offenses, the qualitative data suggest officers are aware of this trend and do not want to adopt a zero-tolerance approach with citations but feel restrained in their ability to respond in other ways. Exacerbating the situation is the conflict between departmental policy and police culture where pressure exists to resolve incidents quickly using informal means while receiving little credit for doing so. Alternatively, there is encouragement for charges and citations to achieve productivity expectations. This suggests reexamining the performance evaluation criteria used by police services to better reflect the nature and activities of a growing proportion of police work.

Evidence also points to the importance of developing effective partnerships, collaborative agreements, and additional mental health community resources to allow for more diversion to reduce the criminalization of PMI and to improve the civil commitment process. Additional research is needed on nonarrest dispositions and the circumstances favorable to their use in PMI-related encounters. The ways in which specialized units affect dispositional outcomes is important to better understand these situational contingencies facing officers and assist in the development of response and training frameworks. These frameworks should include tactics other than arrest that can be used when citizens refuse mental health treatment but do not represent a danger to self or others thereby negating the option of a civil commitment.

Finally, we continue to know little on how mental health training influences the ways in which situational and behavioral factors affect officer decision-making processes. Observational data suggest the police, mental health practitioners, and social services operate in their own silos. There is little understanding of each other’s perspectives and respective challenges when responding to PMI. Recruit training could involve exposure to video simulations that include other system actor perspectives and current information on community resources. Recognizing a disjuncture between the training content and police cultural norms and expectations will continue to pose a challenge for law enforcement. However, placing greater emphasis on reorienting views of mental illness, aligning cultural norms with training principles, and increasing the number of specialized CIT and MCT officers can help shift the balance.

Limitations

The community context was not considered in this study due to insufficient data to demarcate by division and patrol zone. There is the possibility that the decision-making process and use of discretion is affected by perceptions of the neighborhood (e.g., whether officers use more discretion in high crime rate areas as the public is more tolerant and the behavior is minor in comparison with systemic problems such as drugs or prostitution). Second, citizens with no prior calls for service or who did not endorse obvious behavioral indicators are not included. Third, due to the small sample size, the findings are not generalizable to police work in general at this police service and the age distribution suggests younger PMIs are underrepresented. There is also the inability to conduct separate subpopulation analyses for PMI versus non-PMI. For instance, the sample could not be subdivided by mental health status and category of offense. Finally, separating traffic and public order citations could produce different results as could demarcating by violent or property crime. Furthermore, generalizability is limited because shifts, zones, and officers were not selected using random procedures. However, the findings do contribute to the discourse on policing practice with potential transferability of findings to other jurisdictions due to the identification of generic processes associated with patrol duties.

Conclusion

From a policy and practice perspective, officers face significant challenges when responding to PMI, which needs to be addressed to reduce the criminalization of the mentally ill, particularly with noncriminal behavior that may be more symptomatic of the mental illness. Who benefits from the use of discretion differs due to the decision-making context, procedural constraints, and for PMI under the influence of alcohol or drugs. Given these findings, understanding the situational contingencies officers’ face can assist in the development of a cross-jurisdictional response and training framework geared to implementing evidence-based practice.

What does need to be emphasized is officers’ passion about being part of the solution, ensuring citizens’ well-being, and advocating for keeping PMI out of the criminal justice system if at all possible. However, they also made clear that, in too many circumstances, they are faced with the decision to arrest PMI for public order behavior and lay criminal charges for minor offenses due to limits on their decision-making autonomy. Officers face complex situations, and their decision-making operates in a gray zone despite their professional judgment that criminalization of PMI exacerbates their life circumstances and well-being.

Footnotes

This research was supported in part by a grant from the Social Sciences and Humanities Research Council of Canada.

Notes

Jennifer L. Schulenberg is an associate professor in the Department of Sociology and Legal Studies at the University of Waterloo, Ontario, Canada. Her research centers on police decision-making, policing vulnerable populations, economics of policing, misconduct, culture, research methods, and youth crime.

References

Adelman

(2003). Study in the blue and grey. Police interventions with people with mental illness: A review of challenges and responses. Vancouver, British Columbia: Canadian Mental Health Association.

Bittner

(1967). Police discretion in emergency apprehension of mentally ill persons. Social Problems, 14, 278-292. doi:10.2307/799150

Blue Line. (2014, March 26).National standards being developed for calls involving mentally ill (Conference studies standards). Blue Line, 26(5), 24. Retrieved from http://blueline.ca/articles/national_standards_being_developed_for_calls_involving_mentally_ill

Bonovitz

J. C.

Bonovitz

J. S.

(1981). Diversion of the mentally ill into the criminal justice system: The police intervention perspective. American Journal of Psychiatry, 138, 973-976. doi:10.1176/ajp.138.7.973

Borum

(2000). Improving high risk encounters between people with mental illness and the police. The Journal of the American Academy of Psychiatry and the Law, 28, 332-337. doi:10.1007/s10488-014-0594-9

Brink

Livingston

Desmarais

Greaves

Maxwell

Michalak

. . . Weaver

(2011). A study of how people with mental illness perceive and interact with the police. Retrieved from https://www.mentalhealthcommission.ca/English/system/files/private/Law_How_People_with_Mental_Illness_Perceive_Interact_Police_Study_ENG_1_0.pdf

Brown

R. A.

Novak

K. J.

Frank

(2009). Identifying variation in police officer behavior between juveniles and adults. Journal of Criminal Justice, 37, 200-208. doi:10.1016/j.jcrimjus.2009.02.004

Canadian Association of Chiefs of Police. (2008). Contemporary guidelines for working with the mental health system. Retrieved from http://www.pmhl.ca/webpages/reports/Guidelines%20for%20Police.pdf

Canadian Mental Health Association. (2014). Canadian Mental Health Association fast facts about mental health. Retrieved from http://www.cmha.ca/media/fast-facts-about-mental-illness/#.U5zBePldV8E

10.

Carrington

P. J.

Schulenberg

J. L.

(2003). Prior police contacts and police discretion with apprehended youth (Crime and Justice Research Paper Series, Report No. 85-561-MIE-003). Ottawa, Ontario: Statistics Canada.

11.

Chaimowitz

(2012). The criminalization of people with mental illness. Canadian Journal of Psychology, 57, 1-6. doi:10.1080/17522439.2013.790070

12.

Chandrasekera

(2011). Police and mental health: A critical review of joint police/mental health collaborations in Ontario (Provincial Human Services and Justice Coordinating Committee). Toronto, Ontario: Canadian Mental Health Association.

13.

Compton

M. T.

Bahora

Watson

A. C.

(2008). A comprehensive review of extant research on crisis intervention team (CIT) programs. The Journal of the American Academy of Psychiatry and the Law, 36, 47-55. doi:10.1176/appi.ajp.2009.09050670

14.

Compulsory Automobile Insurance Act, R.S.O. 1990, c.C.25.

15.

Cooper

V. G.

Mclearen

A. M.

Zapf

P. A.

(2004). Dispositional decisions with the mentally ill: Police perceptions and characteristics. Police Quarterly, 7, 295-310. doi:10.1177/1098611104267733

16.

Cordner

(2000). A community policing approach to persons with mental illness. The Journal of the American Academy of Psychiatry and the Law, 28, 326-331. doi:10.1016/S0196-1152(05)12002-X

17.

Cotton

(2004). The attitudes of Canadian police officers toward the mentally ill. International Journal of Law and Mental Health, 27, 135-146. doi:10.1016/j.ijlp.2004.01.004

18.

Cotton

Coleman

T. G.

(2010). Canadian police agencies and their interactions with persons with a mental illness: A systems approach. Police Practice & Research, 11, 301-314. doi:10.1080/15614261003701665

19.

Criminal Code of Canada, R.S.C. 1985, c. C-46.

20.

Engel

R. S.

Silver

(2001). Policing mentally disordered suspects: A re-examination of the criminalization hypothesis. Criminology, 39, 225-252. doi:10.1111/j.1745-9125.2001.tb00922.x

21.

Engel

R. S.

Tillyer

Klahm

C. F.

IV. Frank

(2012). From the officer’s perspective: A multilevel examination of citizens’ demeanor during traffic stops. Justice Quarterly, 29, 650-683. doi:10.1080/07418825.2011.574643

22.

Ericson

R. V.

(1982). Reproducing order: A study of police patrol work. Toronto, Ontario, Canada: University of Toronto Press.

23.

Godfredson

J. W.

Thomas

S. D. M.

Ogloff

J. R. P.

Luebbers

(2011). Police perceptions of their encounters with individuals experiencing mental illness: A Victorian survey. Australian & New Zealand Journal of Criminology, 44, 180-195. doi:10.1177/0004865811405138

24.

Hafemeister

T. L.

Garner

S. G.

Bath

V. E.

(2012). Forging links and renewing ties: Applying the principles of restorative and procedural justice to better respond to criminal offenders with a mental disorder. Buffalo Law Review, 60, 147-185. doi:10.1177/0306624X14554063

25.

Hatfield

R. E.

(2014). Training law enforcement in mental health: A broad-based model (Unpublished doctoral dissertation). Marshall University, Huntington, WV.

26.

Heslop

Stitt

Hoch

J. S.

(2013). Trends in police contact with persons with mental illness. Retrieved from http://police.city.london.on.ca/Headlines/ReadMore/MentalIllnessNotes2013.pdf

27.

Hiday

V. A.

Burns

P. J.

(2009). Mental illness and the criminal justice system. In Sheid

T. L.

Brown

T. N.

(Eds.), A handbook for the study of mental health: Social contexts, theories, and systems (2nd ed., pp. 478-498). New York, NY: Cambridge University Press.

28.

Highway Traffic Act, R.S.O. 1990, c.H.8.

29.

Hoch

J. S.

Hartford

Heslop

Stitt

(2009). Mental illness and police interactions in a mid-sized Canadian city: What the data do and do not say. Canadian Journal of Community Mental Health, 28(1), 49-66. doi:10.7870/cjcmh-2009-0005

30.

Holley

H. L.

Arboleda-Florez

(1988). Criminalization of the mentally ill: Part I. Police perceptions. Canadian Journal of Psychiatry, 33, 81-86. doi:10.1016/0376-8716(91)90038-Z

31.

Iacobucci

(2014). Police encounters with people in crisis. Retrieved from https://www.torontopolice.on.ca/publications/files/reports/police_encounters_with_people_in_crisis_2014.pdf

32.

Interuniversity Consortium for Political and Social Research. (2007). Project on policing neighborhoods in Indianapolis, Indiana, and St. Petersburg, Florida, 1996-1997. Retrieved from http://www.icpsr.umich.edu/icpsrweb/ICPSR/studies/03160#method

33.

Kappeler

V. E.

Gaines

L. K.

(2011). Community policing: A contemporary perspective (6th ed.). Boston, MA: Anderson Publishing.

34.

Kesic

Thomas

S. D. M.

Ogloff

J. R. P.

(2013). Use of nonfatal force on and by persons with apparent mental disorder in encounters with police. Criminal Justice and Behavior, 40, 321-337. doi:10.1177/0093854812474425

35.

Klinger

D. A.

(1996). More on demeanor and arrest in Dade County. Criminology, 34, 61-82. doi:10.1111/j.1745-9125.1996.tb01195.x

36.

Kochel

T. R.

Wilson

D. B.

Mastrofski

S. D.

(2011). Effects of suspect race on officers’ arrest decisions. Criminology, 49, 473-512. doi:10.1111/j.1745-9125.2011.00230.x

37.

Lamb

H. R.

Weinberger

L. E.

DeCuir

W. J.

(2002). The police and mental health. Psychiatric Services, 53, 1266-1271. doi:10.1176/appi.ps.53.10.1266

38.

Lamb

H. R.

Weinberger

L. E.

Gross

B. H.

(2004). Mentally ill persons in the criminal justice system: Some perspectives. Psychiatric Quarterly, 75, 107-126. doi:10.1176/appi.ps.53.10.1266

39.

Levesque

(2010). Closing the door: Mental illness, the criminal justice system, and the need for a uniform mental health policy. Nova Law Review, 34, 711-720.

40.

Liquor Licence Act, R.S.O. 1990, c.L.19.

41.

Lurigio

A. J.

(2000). Persons with serious mental illness in the criminal justice system: Background, prevalence, and principles of care. Criminal Justice Policy Review, 11, 312-328. doi:10.1177/0887403400011004003

42.

Mahoney

(2013). Addressing the criminalization of the mentally ill: The importance of jail diversion and stigma reduction. Journal of Medicine and Law, 17, 327-342. doi:10.1007/s10464-010-9345-6

43.

Mastrofski

S. D.

Worden

R. E.

Snipes

J. B.

(1995). Law enforcement in a time of community policing. Criminology, 33, 539-563. doi:10.1111/j.1745-9125

44.

Mercer

(2015, April 20). Police insist lethal force the last record. The Record. Retrieved from http://therecord.com/news-story/5564461-lethal-force-is-last-resort-waterloo-regional-police-say/

45.

Novak

K. J.

Engel

R. S.

(2005). Disentangling the influence of suspects’ demeanor and mental disorder on arrest. International Journal of Police Strategies & Management, 28, 493-512. doi:10.1108/13639510510614573

46.

Ontario Mental Health Act, R.S.O. 1990, c.M.7.

47.

Patch

P. C.

Arrigo

B. A.

(1999). Police officer attitudes and use of discretion in situations involving the mentally ill: The need to narrow the focus. International Journal of Law and Psychiatry, 22, 23-35. doi:10.1016/j.ijlp.2014.02.014

48.

Perreault

(2013). Police-reported crime statistics in Canada, 2012 (Report No. 85-002-X). Ottawa, Ontario: Statistics Canada.

49.

Plecas

McCormick

A. V.

Cohen

I. M.

(2012). RCMP Surrey ride along study: General findings. Abbotsford, British Columbia, Canada: Centre for Public Safety & Criminal Justice Research, University of the Fraser Valley.

50.

Price

(2005). Commentary: The challenge of training police officers. The Journal of the American Academy of Psychiatry and the Law, 33, 50-54. doi:10.1016/j.ijlp.2014.02.003

51.

Reuland

Schwarzfeld

Draper

(2009). Law enforcement responses to people with mental illnesses: A guide to research-informed policy and practice. New York, NY: Council of State Government Justice Center.

52.

Riksheim

E. C.

Chermak

S. M.

(1993). Causes of police behavior revisited. Journal of Criminal Justice, 21, 353-382. doi:10.1016/0047-2352(93)90019-J

53.

Roberg

Crank

Kuykendall

(2000). Police and society. Los Angeles, CA: Roxbury Publishing.

54.

Royal Canadian Mounted Police. (2010). Policing persons with mental illness: Issues and trajectories. Retrieved from http://www.crcc-ccetp.gc.ca/en/archived-policing-persons-mental-illness-issues-and-trajectories

55.

Ruiz

Miller

(2004). An exploratory study of Pennsylvania police officers’ perceptions of dangerousness and their ability to manage persons with mental illness. Police Quarterly, 7, 359-371. doi:10.1177/1098611103258957

56.

Schulenberg

J. L.

(2014). Systematic social observation of police decision-making: The process, logistics, and challenges in a Canadian context. Quality & Quantity: International Journal of Methodology, 48, 297-315. doi:10.1007/s11135-012-9769-1

57.

Schulenberg

J. L.

(2015). Moving beyond arrest and reconceptualizing police discretion: An investigation into the factors affecting conversation, assistance, and criminal charges. Police Quarterly, 18, 244-271. doi:10.1177/1098611115577144

58.

Shenton

A. K.

(2004). Strategies for ensuring trustworthiness in qualitative research projects. Education for Information, 22, 63-75.

59.

Sherman

L. W.

(1980). Causes of police behavior: The current state of quantitative research. Journal of Research in Crime & Delinquency, 17, 69-100. doi:10.1177/002242788001700106

60.

Short

T. B. R.

MacDonald

Luebbers

Ogloff

J. R. P.

Thomas

S. D. M.

(2014). The nature of police involvement in mental health transfers. Police Practice & Research, 15, 336-348. doi:10.1080/15614263.2012.736717

61.

Slate

R. N.

Johnson

W. W.

(2008). The criminalization of mental illness: Crisis and opportunity for the justice system. Durham, NC: Carolina Academic Press.

62.

Smith

Novak

Frank

Lowenkamp

(2005). Explaining police officer discretionary activity. Criminal Justice Review, 30, 325-346. doi:10.1177/0734016805285260

63.

Smoke-Free Ontario Act, R.S.O. 1994, C.10.

64.

Teplin

L. A.

(2000). Keeping the peace: Police discretion and mentally ill persons. National Institute of Justice Journal. Washington, DC: National Institute of Justice.

65.

Teplin

L. A.

Pruett

N. S.

(1992). Police as streetcorner psychiatrist: Managing the mentally ill. International Journal of Law and Psychiatry, 15, 139-156. doi:10.1016/0160-2527(92)90010-X

66.

Terrill

Paoline

E. A.

III . (2007). Nonarrest decision making in police-citizen encounters. Police Quarterly, 10, 308-331. doi:10.1177/1098611107299998

67.

Trespass to Property Act, R.S.O. 1990, c.T.21.

68.

Wallace

Turner

Matarazzo

Babyak

(2009). Measuring crime in Canada: Introducing the Crime Severity Index and improvements to the Uniform Crime Reporting Survey (Report No. 85-004-X). Ottawa, Ontario: Statistics Canada.

69.

Watson

A. C.

Angell

(2012). 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. Psychology, Public Policy, and Law, 19, 30-39. doi:10.1037/a0027931

70.

Watson

A. C.

Corrigan

P. W.

Ottati

(2004). Police responses to persons with mental illness: Does the label matter? The Journal of the American Academy of Psychiatry and the Law, 32, 75-85. doi:10.1007/s10597-011-9384-y

71.

Watson

A. C.

Morabito

M. S.

Draine

Ottati

(2008). Improving police response to persons with mental illness: A multi-level conceptualization of CIT. International Journal of Law and Psychiatry, 31, 359-368. doi:10.1016/j.ijlp.2008.06.004

72.

Watson

A. C.

Ottati

V. C.

Morabito

Draine

Kerr

A. N.

Angell

(2010). Outcomes of police contacts with persons with mental illness: The impact of CIT. Administration and Policy in Mental Health, 37, 302-317. doi:10.1007/s10488-009-0236-9

73.

White

Perrone

(1997). Police discretion and differential policing. In White

Perrone

(Eds.), Crime and social control: An introduction (pp. 34-63). New York, NY: Oxford University Press.

74.

Worden

R. E.

Shepard

R. L.

(1996). Demeanor, crime, and police behavior: A reexamination of the Police Services Study data. Criminology, 34, 83-105. doi:10.1111/j.1745-9125.1996.tb01196.x

75.

Worden

R. E.

Shepard

R. L.

Mastrofski

S. D.

(1996). On the meaning and measurement of suspects’ demeanor toward the police: A comment on “Demeanor and arrest.” Journal of Research in Crime & Delinquency, 33, 324-332. doi:10.1177/0022427896033003003