Abstract
The relationship between administrative policy and deadly force has been well established in the policing literature. Surprisingly, there has been no similarly oriented research with respect to less lethal force. The current inquiry seeks to start this process. Utilizing data collected from a national multiagency use of force project, we focus on those charged with the street-level application of organizational use of force policy. In doing so, patrol officers (N = 990) from three agencies, each varying in terms of policy direction, are surveyed regarding the extent to which they believe their agency policy offers appropriate forms of guidance and restrictiveness. The findings show a number of significant policy effects. In particular, officers working in a department that uses a loosely coupled nonlinear model are significantly less likely to believe their agency policy offers adequate guidance in terms of when force can and cannot be used. However, the findings also illustrate that officers do not want to be too tightly constrained within a linear policy model in relation to restrictiveness. Such findings suggest that agency leaders may wish to consider a linear-based design that offers some degree of policy guidance, but not so much that force options are overly restricted. The implications of these findings for police practitioners and researchers are considered.
While the U.S. Supreme Court, via Graham v. Connor (1989), provides a general legal standard (i.e., objective reasonableness) for the application of less lethal force, organizationally derived use of force policies attempt to interpret this principle and provide an operational guide for officers in their coercive encounters with citizens. At the same time, the composition of less lethal force policies vary greatly across police agencies (Walker, 2007), and are sometimes linked to a continuum where various force options are detailed in severity relative to corresponding levels of citizen resistance (Terrill, 2001). Recent research has noted that even among agencies that employ a use of force continuum approach, there is widespread variation in the structural design (e.g., linear, modified-linear, wheel, matrix), the ordering of various force options (i.e., hands- and weapon-based tactics), and the incorporation of citizen resistance levels relative to permissible force options (i.e., restrictiveness; Terrill & Paoline, 2013).
What we do not know, from an empirical standpoint, is whether variation in less lethal force policies result in better or worse outcomes. That is, do different types of policies vary in terms of effectiveness? Any attempt to answer this multifaceted question would have to first consider identifying relevant outcomes (e.g., officer guidance, overall force usage, citizen complaints, and citizen and officer injuries). In addition, answers would undoubtedly come from not a single study but instead multiple research endeavors, drawing on data from a variety of sources and methodologies (e.g., surveys, official records, observation).
Utilizing data collected from a national multiagency use of force project, the current inquiry seeks to start the process of assessing the potential benefits and costs of divergent less lethal policies. 1 In doing so, we begin by going directly to those for whom the policy was written. Patrol officers (N = 990) from three agencies, each varying in terms of policy direction, are surveyed regarding the extent to which they believe their agency policy offers appropriate forms of guidance and restrictiveness.
Less Lethal Use of Force Policies
In exercising their less lethal coercive power over citizens, police officers are guided by two entities—the U.S. Supreme Court and their departmental use of force policy. The U.S. Supreme Court, via Graham v. Connor (1989), loosely directs officers to utilize a standard of “objective reasonableness” in applying the appropriate level of less lethal force, a standard that is less clear than the “defense of life” principle guiding the application of deadly force via Tennessee v. Garner (1985). That is, determining what is objectively reasonable in less lethal force situations is subject to interpretation, across a variety of audiences (e.g., the public, courts, police department, street-level officer), except in cases at the extreme (Terrill & Paoline, 2010). For example, most observers would regard the coercive actions of an officer striking a citizen in the face for passively ignoring a request to hand over their driver’s license during a traffic stop as not objectively reasonable. However, most observers would regard the coercive actions of an officer striking a citizen in the face, who is physically attacking a police officer, as objectively reasonable. The less lethal force actions outside of these illustrative extremes are subjectively interpreted as reasonable, appropriate, excessive, based in large part on a second more salient guide to the daily application of coercion—administratively based use of force policy (Walker, 2007).
Use of force policies, like other organizationally based directives (e.g., vehicle pursuits, domestic assaults, searches), are designed to assist and guide officers in their street-level decision making. As such, use of force policies not only serve as benchmarks for assessing, ex post facto, whether officers utilize proper force, but more importantly, they are designed to instruct officers as to when to use different types of force. At the same time, the construction of force policies (i.e., the design and placement of various tactics) are largely left up to the discretion of individual police agencies, as there is no commonly accepted or universal policy (Terrill & Paoline, 2013).
Existing Knowledge on Less Lethal Use of Force Policies
From an empirical standpoint, the precise nature of the variation in less lethal policies was relatively unknown until recently. Early work on use of force policies concentrated on identifying the existence of written directives, as well as the structural composition and procedures associated with such directives (e.g., permissible tactics, force definitions, reportable force, training, reporting requirements, review process; McEwen, 1997; Pate & Fridell, 1993, 1995).
In terms of force instruction, researchers also focused on whether or not departments utilized a force continuum. That is, a policy that details various levels of force in severity, relative to the type or level of resistance encountered from citizens (Terrill, 2005). McEwen (1997), utilizing a convenience sample of police agencies (n = 96), found that 25% incorporated a continuum. Alpert and Dunham (2004), using data from the 1998 Police Executive Research Forum (PERF) national survey (n = 571) of use of force reporting protocols, noted that 72% of responding agencies used a continuum approach. While such work helped to provide a foundation for understanding whether force continuums were being used by police departments throughout the United States, it did not paint a comprehensive picture of how such continuums were constructed in terms of their design, the placement of different forms of force, and if (and how) various types of citizen resistance were linked to various force options. More recent research on use of force policies has begun to address these empirical voids.
Some of the latest research on less lethal policies has been driven by a contemporary controversy surrounding conducted energy devices (CEDs; for example, where on the continuum they should be located). The U.S. General Accounting Office (2005) extensively reviewed use of force policies across seven police departments, finding that the placement of CEDs was far from uniform. That is, some agencies placed it high on the use of force continuum, while others placed it low. Using a larger national survey of 210 police agencies, Thomas, Collins, and Lovrich (2010) reported that on a scale of 0 (i.e., officer presence) to 10 (i.e., lethal force), more than 60% of the responding agencies placed CEDs between a 5 and 7. In another study, Thomas, Collins, and Lovrich (2012) assessed how CEDs should be used when interacting with various types of resistant citizens. Using a subset of their original 210 police agencies (n = 124), the researchers found that 60% required that citizens be “actively” resisting before officers deployed their CEDs. While the above research began to scratch the surface regarding the placement of force options and consideration of citizen resistance on the continuum, the scope was somewhat narrow in that it concentrated on one type of force (i.e., CED weapons).
To date, Terrill and Paoline (2013) have provided the most extensive examination of less lethal force policies. Based on a multiwave national mail survey (n = 662) of police agencies throughout the United States, this work illuminated both the collectiveness (i.e., continuum use and design option) and variation (i.e., where force tactics and citizen resistance are situated) in contemporary use of force policies. More specifically, in line with extant research, the authors found that 80.8% of responding agencies were utilizing a use of force continuum, of which the “linear” design (i.e., one characterized by a series of steps that hierarchically move upward in severity for officers to consider as they apply coercive tactics) was the most prevalent (73.4%). With respect to the placement of different forms of hands and weapon-based force tactics along the continuum, Terrill and Paoline (2013) highlighted great variation, as they found “123 different permutations” across “3 to 9 different levels” where force options could be placed (p. 47). In terms of CED placement specifically, they found that nearly two thirds of departments require some type of hands-on force before being permitted to use a CED. Conversely, less variation, but far from a consensus, was found for the ordering of citizen resistance on force continuums, as “23 permutations were uncovered ranging from 3 to 7 different levels” (p. 51). When agencies linked permissible force relative to citizen resistance, as an indicator of policy restrictiveness, the researchers again noted great variation in approaches (i.e., 103 permutations when CEDs and chemical sprays were located in a general “weapons” category, and 169 permutations when CEDs and chemical sprays were individually located among other force choices).
In sum, the burgeoning empirical research on less lethal policies is beginning to produce some direction as to the various ways agencies attempt to guide officers’ discretionary decision making. More specifically, despite limited claims regarding the phasing out of the use of force continuum (Peters & Brave, 2006; Petrowski, 2002), a large majority of police agencies utilize a continuum approach, of which the linear design is most preferred. However, agencies are much more autonomous and less uniform when it comes to the ordering of force and the linking of force options to citizen resistance within a continuum framework. Hence, there are a variety of ways in which officers are being guided in their coercive power, especially when it comes to the restrictiveness of their organizational policy (i.e., allowable force relative to citizen resistance).
Less Lethal Use of Force Policies and Outcomes: The Missing Link
Given the structured variation in less lethal direction, a logical question arises whether one approach is more or less beneficial than another. For example, is a policy design that ranks various hands-on force tactics (e.g., soft hands, pain compliance controls, hard hands) and weapons (e.g., batons, chemical sprays, CEDs) in one manner going to produce better or worse outcomes compared with another policy that ranks the progression of force in a different way? Likewise, in considering a second key component of use of force continuums (i.e., the incorporation of citizen resistance in directing which force option to use), are more restrictive policies better or worse than others?
Researchers interested in attempting to answer such questions, in relation to policy effectiveness, must first determine the relevant outcomes. For instance, is an effective policy one that, compared with others, provides officers better guidance and direction, and results in lower amounts of overall force usage, fewer complaints in use of force encounters, less excessive force, and/or fewer injuries to citizens and officers? Tackling all of these outcomes would be an arduous task that would require a variety of research methodologies. Nonetheless, working toward such a goal is certainly worthwhile, especially in light of recent requests for a national use of force reporting system, similar to that of the Uniform Crime Reports for crime (Fyfe, 2002; McEwen, 1996; Pate & Fridell, 1995). These efforts would undoubtedly be enhanced if use of force directives, via agency policy, were also standardized. So, not only would departments report all of their less lethal force (from lower level hands tactics to upper level weapons), but the criteria for when to apply each tactic and weapon (and against which type of resistant citizen) would be similar, allowing for a more clear assessment and comparison of force (i.e., leveling the playing field) across various locales. Such long-term goals would benefit from a collection of empirical knowledge on the benefits and costs of varying policies on force-related outcomes.
The Current Inquiry
In the present study, patrol officers (N = 990) from three agencies, each varying in terms of policy direction, are surveyed regarding the extent to which they believe their agency policy offers appropriate forms of guidance and restrictiveness. To account for the multidimensional nature of force, policy guidance and restrictiveness are considered in terms of the policy in general (i.e., overall view of their policy), as well as across various encounters with different types of resistant citizens (i.e., view of their policy in terms of citizens who are nonresistant, passively resistant, verbally resistant, physically resistant in a nonassaultive nature, and physically resistant in an assaultive manner). Multivariate analyses are utilized to tease out potential policy effects, while controlling for other theoretically relevant causal factors of patrol officer attitudes and/or the use of force (i.e., officer demographics and orientations toward the primary work environments).
Method
Data
Data are drawn from the Assessing Police Use of Force Policy and Outcomes project, a National Institute of Justice (NIJ)–funded study involving a host of use of force issues, with the overarching goal to examine policy variation and related outcomes. The project included two primary phases. The first phase consisted of surveying a stratified random sample of 1,083 U.S. police agencies to illuminate the various use of force policies in operation nationally. Based on the variation in force policy identified in the first phase, the second phase involved a deeper exploration of eight police departments, each of which instructed the application of force in a different manner. 2 As part of the second phase, a variety of data were collected regarding the use of force (e.g., officer perceptions, use of force reports, citizen complaints, citizen and officer injuries, lawsuits) and the organization as a whole (e.g., policies and procedures, organizational charts, use of force history, official rosters of sworn personnel, training, calls for service, arrests, crime). Given the focus of the current study, we rely on two data sources collected during the second phase—surveys of patrol officers and organizational use of force policies.
In the fall of 2007, trained project staff administered a 116-item survey to patrol officers during roll call sessions. The survey’s primary aim was to query officers regarding their use of force policy. In developing a structured plan for the administration of the survey, project staff obtained an official roster of officer assignments, which suggested that, across the three agencies, 1,458 officers should (i.e., on paper) be at designated roll calls. After accounting for officers’ scheduled days off, court, training, military obligations, injuries, suspensions, or other absences for unspecified reasons, there were 1,006 patrol officers physically present to be surveyed, of whom 990 participated. This resulted in a response rate of 98.4% of those physically present at the survey’s administration, and 67.9% of the official count of patrol personnel.
In addition to the patrol officer survey, a second source of data focused on the specific interpretation and detail of each agency’s use of force policy. These data were also collected during the site visits as part of the second phase of the project. After procuring and reviewing the official written departmental policy on less lethal force (including both the narrative and graphical depiction components), project staff conducted a series of open-ended interviews (using a snowball method) with a variety of police personnel of varying ranks and assignments (e.g., patrol officers, middle managers, top-level executives, trainers, special tactics personnel, public information officers, internal affairs) in an effort to detail how policies were structured and implemented within the organization. This allowed for richly detailed data on use of force directives beyond simply obtaining a copy of the department’s policy. These data also illuminated the precise ways in which the policies varied, which we describe in the following section.
Independent Variables
As noted previously, police departments use a number of different policy approaches seeking to guide officer discretion in the use of force. Thus, the primary explanatory variables for the current study are the various organizational approaches to directing the use of coercion, via each agency’s less lethal force policy.
Less Lethal Force Policy 1: The Charlotte–Mecklenburg Police Department (CMPD) Approach
Drawing on three interrelated use of force policy directives including a graphic depiction (i.e., use of force continuum, less lethal, and deadly force), the CMPD, North Carolina, uses a linear continuum design, with explicit force levels, and a link to varying types of resistance. 3 Although the policy specifies that it is not designed to be a step-by-step progression model, it explicitly outlines six levels of citizen resistance (cooperative, verbal and nonverbal, passive, defensive, active aggression, aggravated active) along with seven levels of force (professional presence and verbal dialogue and commands, soft hands, chemical spray, hard hands, CEDs, impact weapon, and deadly force). While there is not a complete overlap between resistance and force options, there is fairly explicit direction as to which types of force are most appropriate given different types of resistant behaviors (e.g., the policy indicates that oleoresin capsicum spray will normally be used when the officer is confronted with defensive resistance). The progression of force is somewhat measured (e.g., the use of soft empty hand control prior to oleoresin capsicum spray before hard empty hand control) and restricts the use of empty hand tactics, CEDs (e.g., TASER®), and impact weapons (e.g., ASP baton) to suspects presenting a range of resistance between defensive and active. On the whole, CMPD’s policy would appear to offer fairly explicit guidance to officers, while also being somewhat restrictive in nature.
Less Lethal Force Policy 2: The Albuquerque Police Department (APD) Approach
Drawing on one policy directive that incorporates three use of force components along with a graphic depiction (i.e., less lethal force, nondeadly, and deadly force), the APD, New Mexico, also uses a linear force continuum design (i.e., the Reactive Control Model or RCM), which lays out varying force options that are linked to varying forms of citizen resistance. This model specifies four levels of citizen resistance (cooperative, noncooperative, unarmed assailant, armed assailant) along with four broad based “officer mental conditions” (alert, control, active, survival), which serve as an overarching context to varying force tactics. Within each of the four broad levels, there are specific force types outlined (e.g., under active, it instructs officers to consider anything from verbal commands to the use of an intermediate weapon).
Despite the similarity between CMPD and APD in terms of both using a linear continuum design, there are distinctive differences between the two policy approaches. First, APD’s resistance and force categories are quite broad. Passive, verbal, and defensive resistance are all grouped together into one category (i.e., noncooperative), as is the recommended officer force response (i.e., control), which suggests anything from the use of verbal persuasion, to compliance holds, to chemical agents, to leveraged or impact takedowns. Second, such a setup demonstrates that the recommended progression of force is less measured, as officers are permitted to use relatively higher levels of force on lower levels of resistance. For instance, unlike CMPDs policy, APDs policy does not specify the expectation that oleoresin capsicum spray should primarily be used solely when officers are confronted with defensive resistance. Third, a CED is not explicitly identified on the RCM, but the policy identifies the device as being equivalent to chemical sprays, with but one exception—that the weapon should not be used on passively resistant suspects. Thus, a TASER® may be used on a verbally resistant suspect as well as a suspect demonstrating defensive resistance (e.g., pulling/moving or running away). Conversely, CMPD restricts the use of a TASER® to suspect resistance closer in form to active aggression resistance (e.g., suspects attempting or actually attacking an officer or other citizen). Such differences suggest that APD’s policy would appear to offer less explicit guidance and restrictiveness, compared with CMPD’s policy approach.
Less Lethal Force Policy 3: The Colorado Springs Police Department (CSPD) Approach
Drawing on three interrelated use of force policy directives along with a graphical depiction (i.e., use of force continuum, less lethal, and deadly force), the CSPD, Colorado, relies on a nonlinear design policy (i.e., Situational Force Model). This policy envisions an officer standing in the middle of a circle (or wheel-like object) with various force options placed in random surrounding him or her to indicate there is no natural progression of force (e.g., lethal force placed next to soft control techniques, impact weapons placed next to voice commands). Unlike CMPD and APD’s linear-based models, there is no depiction of citizen resistance as to which types of force are most appropriate given different types of resistance (e.g., officers who encounter verbally resistant citizens are not explicitly required to use a specific type of force in response). Thus, officers are provided a substantial degree of flexibility with respect to what type of force to use when dealing with any number of citizen resistant behaviors. In fact, departmental officials reported that officers operate under the philosophy that there are “no absolutes” and that the use of force policy is not intended to indicate that officers may never do something, but rather that their actions should be justified based on the situation. Within this context, CSPD’s policy appears the least restrictive of the three agencies.
Despite the great leeway offered to officers via the use of a “Situational Force Model,” the policy does identify various types of force options (e.g., presence, voice, soft control, control and compliance tools, hard control, impact munitions) and resistance types (i.e., psychological intimidation, verbal, passive, defensive, active, and aggravated active). Part of this description includes language offering officers guidance in varying situations (e.g., soft control techniques are appropriate when a subject fails to respond to verbal direction or resists in a defensive manner), although other parts of the policy offers conflicting direction (e.g., CEDs may not only be used to control violent or potentially violent persons, but also states that the device may be appropriate when dealing with nonaggressive noncompliant persons). Moreover, the policy calls for officers to escalate and de-escalate force as the subject escalates and de-escalates resistance, as well as consider the least amount of force possible given the situation. 4
Based on the less lethal directives utilized by these three agencies, officers working within the more explicitly designed linear parameters set forth by CMPD’s model should hold the most favorable assessments of policy guidance, compared with APD’s linear design and especially when compared with CSPD’s circular/wheel design. In contrast, the latitude and flexibility in exercising less lethal force afforded to CSPD officers should result in more favorable attitudes toward the restrictive nature of their policy, compared with those working within APD’s policy and especially when compared with the more tightly structured CMPD model.
Table 1 lists CMPD, APD, and CSPD across a multitude of agency and city characteristics. With respect to sworn officers, while the total number varies between 1,638 in CMPD to 669 in CSPD, when the number of sworn officers per 1,000 population is considered, the range is just 2.23 to 1.79. We also see that varying socioeconomic indicators 5 are relatively closely situated. For example, percentage unemployed is tightly grouped between 3.1 in Colorado Springs and 3.8 in Albuquerque, as is percentage of female-headed households (ranging from 7.1% to 8.0%). We tend to see a little more variation on some other measures. For instance, percentage of non-White ranges from 19.3% in Colorado Springs to 36.0% in Charlotte–Mecklenburg.
Study Sites.
For the present study, a series of dummy variables are utilized to measure the different organizational approaches to guiding the use of coercion (0 = all other, 1 = CMPD; 0 = all other, 1 = APD; 0 = all other, 1 = CSPD). 6
Control Variables
We include two sets of variables as statistical controls in our analyses. The first set focuses on a variety of officer assessments relating to their work environments. In a recent symposium on emerging issues, academic and police practitioner experts agreed that a vital missing component in the understanding of police use of force is the attitudinal “mindset” that patrol officers hold toward citizens (on the street) and supervisors (within the organization; Community Oriented Policing Services, 2012, pp. 27-28). The participants further recommended that surveys of patrol officers be conducted and include questions related to, among other things, “their sense of the relative dangerousness of their jobs,” “the perception of members of the community,” and “their beliefs about the supportiveness of police leadership” (Community Oriented Policing Services, 2012, p. 28). To control for officer’s attitudinal mind-set toward force, we include five indices, all of which are measured on a 4-point Likert-type scale (1 = agree strongly, 2 = agree somewhat, 3 = disagree somewhat, and 4 = disagree strongly), where higher values reflect stronger endorsement of each dimension.
Three of the five indices examine components of a patrol officer’s occupational (i.e., street) environment. Perceived citizen distrust is measured with two items, adopted from previous research utilizing the Project on Policing Neighborhoods (POPN) patrol survey (Paoline, 2001, 2004): (a) Police officers have reason to be distrustful of most citizens (reverse coded) and (b) Police officers have reason to be suspicious of most citizens (reverse coded). Cronbach’s alpha scale reliability analysis for the two items in this index revealed a coefficient of .80. Perceptions of dangerousness of the job is measured using three items from Cullen, Link, Cullen, and Wolfe (1989): (a) I work in a dangerous job (reverse code); (b) In my job, a person stands a good chance of getting hurt (reverse coded); and (c) My job is a lot more dangerous than other kinds of jobs (reverse coded). Cronbach’s alpha scale reliability analysis for the three dangerousness items revealed a coefficient of .59. 7 Finally, in an effort to capture patrol officers’ overall (i.e., global, facet-free) job satisfaction (Dantzker & Kubin, 1998), we utilize three items from Brayfield and Rothe (1951; see also Forsyth & Copes, 1994): (a) I would not consider taking another job (reverse coded); (b) I like my job better than the average police officer does (reverse coded); and (c) I find real enjoyment in my job (reverse coded). Cronbach’s alpha scale reliability analysis for the three job satisfaction items revealed a coefficient of .66.
The remaining two indices focus on patrol officers’ attitudes toward their organizational environment (i.e., supervisors). Both indices are adopted from prior research utilizing the POPN patrol survey (Paoline, 2004; Terrill, Paoline, & Manning, 2003). Patrol officers’ attitudes toward top management is based on three survey items: (a) When an officer does a particularly good job, top management will publicly recognize his or her performance (reverse coded); (b) When an officer gets written up for minor violations of the rules, he or she will be treated fairly by top management (reverse coded); and (c) When an officer contributes to a team effort rather than look good individually, top management here will recognize it (reverse coded). Cronbach’s alpha scale reliability analysis for this index revealed a coefficient of .77. A second measure focuses on more immediate direct supervision, and is based on two survey items: (a) My supervisor looks out for the personal welfare of his or her subordinates (reverse coded) and (b) My supervisor’s approach tends to discourage me from giving extra effort. Cronbach’s alpha scale reliability analysis on the two direct supervision items revealed a coefficient of .66.
The second set of variables serving as statistical controls focuses on officer-level characteristics, all of which have been used in studies of police officer attitudes and/or the use of force in general (see, for example, Paoline & Terrill, 2004, 2007; Worden, 1993; Zhao, Thurman, & He, 1999). Specifically, we control for gender (0 = female, 1 = male), race (0 = non-White, 1 = White), experience (number of years as sworn police officer), and education (1 = high school, 2 = some college, but no Bachelor’s degree, 3 = Bachelor’s degree or higher).
Dependent Variables
We draw on a number of questions from the survey to tap into two primary dimensions of officers’ views toward their use of force policy: guidance and restrictiveness. For guidance, officers were queried regarding the extent to which their agency policy “provides adequate guidance in terms of when force can and cannot be used,” with response categories measured along a 4-point Likert-type scale (1 = agree strongly, 2 = agree somewhat, 3 = disagree somewhat, 4 = disagree strongly), which was reverse coded so that higher values reflected more positive attitudes. Using this measure, officers were asked their views of the policy overall, as well as across five different types of resistant citizens (i.e., nonresistant, passively resistant, verbally resistant, physically resistant in a nonassaultive nature, and physically resistant in an assaultive manner).
For restrictiveness, respondents were queried in two different manners. First, they were asked the extent to which they believe their agency policy is “too restrictive.” Second, they were asked the extent to which they believed their agency’s policy is “not restrictive enough.” Once again, patrol officers were selected from four responses categories (and coded so that higher values reflected more favorable attitudes), and were asked about restrictiveness of the policy overall, as well as across five different types of resistant citizens ranging from nonresistant to physically resistant in an assaultive manner (see Table 2 for descriptive statistics).
Descriptive Statistics.
Analyses and Findings
We begin by examining the distribution of officers’ views toward their force policy across each of the three dependent variables, starting with the extent to which they believe their agency policy provides adequate guidance when force can and cannot be used. As shown in Table 3, officers in each of the agencies generally have favorable views of their policy in this regard. A closer look, however, shows that Colorado Springs (CSPD) officers have significantly less favorable views compared with officers in the other departments. For instance, while 90.3% of Albuquerque (APD) officers and 88.2% of Charlotte–Mecklenburg (CMPD) officers agree (strongly or somewhat) that their policy offers adequate guidance, the percentage of officers in CSPD drops to 81.8%. A similar type of pattern plays out when looking at officer views across nearly all forms of citizen resistant categories, with the greatest distinctions found more toward the lower end of the continuum. For example, if we look at the percentage of officers indicating concern toward their policy in terms of properly dealing with nonresistant citizens (i.e., noting they somewhat or strongly disagree that the policy offers adequate guidance), just 9.0% of CMPD officers and 9.9% of APD officers indicate such concern, compared with nearly 20% of CSPD officers. Moreover, the percentages increase even more when looking at officers’ views toward passively resistant citizens (i.e., APD 10.9%, CMPD 14.3%, and CSPD 29.8%). The chi-square statistics show a significant difference across the three agencies for the policy overall, as well as each individual category of resistant citizen type with the exception of physical assaultive.
Bivariate Distribution: Policy Provides Adequate Guidance When Force Can and Cannot Be Used.
Note. CMPD = Charlotte–Mecklenburg Police Department; APD = Albuquerque Police Department; CSPD = Colorado Springs Police Department. Totals across the dependent measures, within agencies, vary due to officers’ failure to respond equally to all items.
Turning to our second dependent variable (Table 4), which centers on querying officers as to the extent to which they believe their agency policy is too restrictive, we find officers vary quite significantly in terms of overall policy views, particularly in CMPD. While less than half of the officers in APD (48.9%) and CSPD (42.1%) believe their policy overall is too restrictive (agreeing strongly or somewhat), two thirds of CMPD officers (66.4%) have similar views. Looking at the varying levels of citizen resistance, it becomes clear that the primary concern lies at the top of the continuum range, as there are no statistically significant differences at the nonresistant, passively resistant, or verbally resistant categories. However, CMPD officers believe the manner in which they are instructed to use force when dealing with physically resisting suspects (both nonassaultive and assaultive) is overly restrictive. For instance, 45.6% of CMPD officers feel their policy is too restrictive at the physical assaultive level, compared with just 34.6% of APD officers and 31.8% of CSPD officers.
Bivariate Distribution: Policy Is Too Restrictive.
Note. CMPD = Charlotte–Mecklenburg Police Department; APD = Albuquerque Police Department; CSPD = Colorado Springs Police Department. Totals across the dependent measures, within agencies, vary due to officers’ failure to respond equally to all items.
Our final measure turns the previous query on restrictiveness upside down, this time inquiring as to whether officers believe their agency policy is not restrictive enough. As depicted in Table 5, few officers in either of the departments agree with the statement that their overall force policy is not restrictive enough. APD officers had the highest percentage of officers agreeing with such a statement, but still only accounts for 7.5% of all respondents, with CMPD not too far behind at 7.2%. The percentage drops in half, however, in CSPD (3.1%), which contributes to a statistically significant difference overall, albeit just under the traditional p > .05 threshold level. Examining differences across varying types of citizen resistance shows that the only significant difference is at the physical nonassaultive level, with 18.9% of APD officers agreeing that their policy is not restrictive enough, compared with 13.5% of CMPD officers and 10.1% of CSPD officers.
Bivariate Distribution: Policy Is Not Restrictive Enough.
Note. CMPD = Charlotte–Mecklenburg Police Department; APD = Albuquerque Police Department; CSPD = Colorado Springs Police Department. Totals across the dependent measures, within agencies, vary due to officers’ failure to respond equally to all items.
While the bivariate analyses illustrate a connection between policy and officer views across several of the measures, it does not allow for an assessment of independent effects. Thus, we estimate a series of ordinal regression multivariate models to examine the relationship between agency policy and officer views, controlling for a host of other factors previously shown to influence officer use of force and/or police attitudes. Once again, we examine potential influences in relation to the overall policy, as well as across varying levels of citizen resistance.
The first set of models assesses the extent to which officers believe their agency policy provides adequate guidance when force can and cannot be used. Given the nominal dichotomous nature of our primary variable of interest (i.e., agency), CMPD initially serves as the reference category, as it represents the most explicit and definitive policy approach of the three departments (as specified in the methods description). Hence, we posit that APD and CSPD officers will offer a lower degree of favorable agreement on this measure, compared with CMPD officers whom we would expect a relatively higher degree of favorable agreement (i.e., the policy provides adequate guidance when force can and cannot be used).
As Table 6 illustrates, in four of the six models, CSPD officers are significantly less likely (at the p < .01 level) to believe their agency policy offers adequate guidance compared with CMPD officers, while there are no differences between APD and CMPD officers. This pattern is found for overall policy views, as well as for instances when officers’ are faced with nonresistant, passive, and verbally resistant citizens. To determine whether there are differences when altering the reference category, we estimated a series of additional models across each of the dependent variables (not shown). When CSPD served as the reference category, both CMPD and APD officers were significantly more likely to believe their agency policy provides adequate guidance. When APD served as the reference category, there were no significant difference between CMPD and APD, while CSPD officers were significantly less likely to believe their agency policy offers adequate guidance. The findings indicate that both CMPD and APD officers view their agency policy similarly in terms of believing it offers adequate guidance. Note (Table 6) that the policy effects dissipate, however, once getting to the two physical resistance categories (i.e., nonassaultive and assaultive). Finally, in terms of the control variables, only two measures are significantly related to officer views toward policy guidance. Officers with favorable views of both their direct supervisor and top management are significantly more likely to believe their policy offers adequate guidance, and such findings hold across all levels of citizen resistance.
Ordinal Regression: Policy Provides Adequate Guidance When Force Can and Cannot Be Used.
Note. APD = Albuquerque Police Department; CSPD = Colorado Springs Police Department.
p < .05. **p < .01.
Next, we take a similar multivariate approach to exploring the extent to which officers believe their agency policy is too restrictive. Once again, CMPD initially serves as the reference category, as it represents the most restrictive policy approach of the three cities as specified in the policy description above. Hence, we posit that APD and CSPD officers will offer a lower degree of favorable agreement on this measure, compared with CMPD officers whom we would expect a relatively higher degree of favorable agreement (i.e., the policy is too restrictive).
As shown in Table 7, compared with CMPD, both APD and CSPD officers are significantly less likely (at the p < .01 level) to believe that their overall agency force policy is too restrictive. Moreover, although APD officers show a significant effect at the verbal resistance level, when examining potential effects across varying levels of citizen resistance, we find that the policy effects are reserved for the upper end of the continuum involving physical resistance. APD and CSPD officers are significantly less likely (at the p < .01 level) to believe their agency policy is too restrictive when dealing with suspects posing a physical threat. By implication, this suggests that CMPD officers are significantly more likely to believe their policy is too restrictive. To confirm, we estimated a series of additional models (not shown) for nonassaultive physical and assaultive physical, as well as the policy overall, altering the reference category to APD and CSPD. In each case, CMPD officers are significantly more likely to believe their policy is too restrictive, while no statistically distinguishable effects were found between APD and CSPD, lending support that CMPD officers view their policy as more restrictive compared with officers in the other two cities.
Ordinal Regression: Policy Is Too Restrictive.
Note. APD = Albuquerque Police Department; CSPD = Colorado Springs Police Department.
p < .05. **p < .01.
In addition to noted policy effects, several of the statistical controls are consistently related to officer views of their policy being too restrictive. Officers with favorable views of both their direct supervisor and top management are significantly less likely to believe that their agency policy is too restrictive across all levels of resistance. In addition, we find that officers with greater distrust of citizens are more likely to believe their policy is too restrictive. This finding holds for the policy overall, as well as for lower forms of citizen resistance (up to verbal). The only other significant effects are experience level for overall policy views and education level for verbal resistance. In both case, these effects just reach statistical significance (p = .043 and p = .042, respectively).
The final set of models examines the extent to which officers believe their agency policy is not restrictive enough. In this instance, CSPD (not CMPD) serves as the reference category, as it represents the least restrictive policy approach of the three cities as specified in the policy description above. Hence, we expect that CMPD and APD officers will offer a lower degree of favorable agreement on this measure, compared with CSPD officers whom we would expect a relatively higher degree of favorable agreement (i.e., the policy is not restrictive enough). As illustrated in Table 8, neither of the agency variables is significantly related to officer views on this dependent measure, either in terms of overall policy or across any of the citizen resistance types. That is, officers in all three of the departments have similar views. This is not overly surprising given the bivariate results, which indicated that few officers, irrespective of their agency’s less lethal direction, agree with the statement that their overall force policy is not restrictive enough. 8
Ordinal Regression: Policy Is Not Restrictive Enough.
Note. APD = Albuquerque Police Department; CMPD = Charlotte–Mecklenburg Police Department.
p < .05. **p < .01.
While no policy effects were found regarding this dependent measure, several control measures are statistically significant. White officers, those with more education, and those with favorable views of top management are often less likely to believe their policy is not restrictive enough. Conversely, those officers who are more distrustful of citizens and have favorable views of top management are more likely to believe their policy is not restrictive enough.
Discussion
The purpose of the present inquiry was to empirically examine the effect of varying types of less lethal force policies on street-level officer perceptions. Patrol officers (N = 990) from three agencies, each varying in terms of policy direction, were surveyed regarding the extent to which they believe their agency policy offers appropriate forms of guidance and restrictiveness. The findings show a number of significant policy effects.
First, CSPD officers are significantly less likely to believe their agency policy offers adequate guidance in terms of when force can and cannot be used, especially when dealing with suspects posing nonphysical forms of resistance (e.g., passive and verbal resistance). Recall that CSPD uses a loosely coupled nonlinear model policy approach whereby officers are provided a substantial degree of flexibility with respect to what type of force to use when dealing with any number of citizen resistant behaviors. Conversely, CMPD uses a linear continuum design, with explicit direction as to when to use varying types of force in relation to varying forms of citizen resistance, while APD also uses a linear-based model but one somewhat more broadly construed. Such results indicate that specificity, at least in terms of the linear-based models practiced by CMPD and APD, matter to the extent that officers feel these policies provide an advantage in relation to more appropriately guiding officer discretionary choices.
Second, officers in both the APD and CSPD’s are significantly less likely to feel that their agency policy is too restrictive, compared with CMPD officers, particularly when dealing with suspects posing a physical threat. Conversely, CMPD officers are significantly more likely to believe their policy is too restrictive. Such a finding indicates that APD and CSPD officers believe their policy offers more appropriate leeway in terms of force options when dealing with physical resistant suspects, while CMPD officers feel constrained in their discretionary force options.
Third, we found no policy effects in terms of any agencies having policies that were not restrictive enough. In fact, few officers in either of the cities believe their force policy is not restrictive enough. This finding holds for officers’ views of their agency policy overall as well as across all forms of citizen resistance.
So what should police practitioners take away from such findings? For one, providing a substantial degree of discretionary flexibility in the form of a loosely coupled nonlinear policy model does not necessary translate to officers believing it helps them decide when and when not to use force. While one may reasonably posit that street-level officers would want as much discretion and freedom as possible, our findings show that officers believe such an approach offers significantly less guidance, at least when dealing with passive and verbally resistant suspects. While such forms of resistance may not pose the same threat probability to officers in terms of violence, citizens failing to respond to officer direction either passively or verbally may present a greater dilemma in terms of what to do. That is, officers are generally well trained on how to handle suspects who physically resist control. For instance, nearly all policies permit officers to use some form of hands-on tactics or less lethal weapons on suspects displaying assaultive physical resistance. Conversely, how to deal with a suspect simply refusing to show identification is more problematic in terms of what forms of coercive act may be permissible. In this sense, perhaps officers need greater direction.
Second, policy attempts to hone in or tightly control discretionary force choices lead to officers believing they are too tightly constrained, especially when considering physically resistant citizens. While officers may wish to have greater specification on how to deal with passive and verbally resistant suspects, it appears they also do not want to be overly regulated as to the types of permissible hands-on or weapon-based tactics once resistance rises to the level of physicality. For instance, the CMPD policy is quite specific and directs officers to use a linear and specific progression of less lethal force (i.e., chemical spray, hard empty hand control, conducted energy weapon, and then impact weapon). In reviewing responses to an open-ended question posed at the end of the structured survey asking officers if there was anything about their policy that should be changed, the most lobbied criticism, by far, was the overly restrictive nature of CMPD’s policy. Much of the concern was directed at where the TASER® was located on the continuum. In fact, 44.9% of the CMPD officers (n = 101 of 225) responding to this question specifically identified the placement of the TASER® as being problematic (being located too high on the continuum), compared with just 11.9% of APD officers (n = 14 of 118) and 11.8% of CSPD officers (n = 8 of 68).
In all, within the context of the three policy approaches examined here, it appears APD’s policy may offer a viable middle ground in terms of providing appropriate guidance and restrictiveness. First, APD officers report a high degree of satisfaction with their policy in relation to properly guiding them as to when and when not to use force across the full range of citizens (from those posing no resistance to those engaging in assaultive resistance). Second, APD officers do not believe their policy is overly resistive nor do they believe it is not restrictive enough. Thus, there appears to be a proper balance of restrictiveness in place.
The current study aimed to start the process of empirically assessing the impact of varying less lethal force policies on police outcomes. However, it is limited to three agencies and patrol officer views across two dimensions of administrative policy. Future research should extend beyond that which was done here to other agencies (and locales) with diverse policies to capture an even more comprehensive picture of officers’ perceptions of their use of force policy. In doing so, different research methodologies could also be incorporated that tap into a variety of outcomes. For example, observational methods could be utilized to examine use of force encounters with citizens to measure the potential costs and benefits of different policy approaches as they play out behaviorally on the street. As part of such inquiries, observers could also incorporate debriefing techniques (i.e., open-ended discussions with officers following citizen encounters) to tap into the operational effectiveness of less lethal policy. For example, debriefing could examine (and detail) the extent to which policies specifically guide officer behavior during different types of force encounters or instances where policy direction comes up short (e.g., in dealing with non–physically resistant citizens). Other work, on outcomes from a patrol officer perspective, could focus on the impact of varying policy approaches on injuries to officers, as perhaps policies that are not properly guiding (or those that are too restrictive) are putting officers in harm’s way.
Given that the use of coercion by the police is viewed and interpreted through a variety of lenses (e.g., officers, administrators, citizens; Terrill & Paoline, 2010), the effectiveness of less lethal policy could extend beyond the patrol officer unit of analysis examined here. Police supervisors and upper level managers also have a stake in use of force policy effectiveness, especially as it relates to the organization as a whole. Outcomes from this perspective could focus on the impact of force policy on overall levels of force, citizen complaints, and lawsuits. In teasing out such relationships, researchers, utilizing organizational official data, could examine the extent to which variation in policy guidance and restrictiveness leads to higher (or lower) levels of force, complaints from citizens based on their encounters with the police, and lawsuits filed for improper/excessive use of force. Because the findings of this study suggest that overall perceptions of immediate supervisors and upper level managers influence attitudes toward use of force policy guidance and restrictiveness, review processes of use of force encounters could also be a relevant outcome. For instance, is the review of use of force reports a “rubber stamp” process or is there a level of behavioral scrutiny whereby officers are being held accountable to the standards set forth in their organizational force directives?
Finally, there are a number of outcomes of less lethal policy effectiveness that are of interest to the public. Researchers interested in the citizen perspective could focus on variation in policy guidance and restrictiveness as it relates to police legitimacy (Tyler, 1990). In doing so, outcomes of relevance would be abuses of coercive power (e.g., excessive force) and injuries to citizens, both of which could work to erode public trust. It could be that those policies that do not effectively guide officers, and those which are not restrictive enough, result in a higher probability of officers using more force than is necessary and also a greater likelihood of unnecessary harm to citizens. Official records, in the form of citizen complaints and injury documentation from use of force reports, would be useful for capturing such outcomes, especially as they relate to reported abuses and injuries. In addition, researchers could interview citizens involved in force encounters with the police to supplement departmental records, in the event that allegations of abuses of coercion and/or injuries were not officially recorded by police (or not reported by citizens).
Footnotes
Acknowledgements
We would like to thank the Editor, Curt R. Bartol for his extraordinary service to this journal, and the anonymous reviewers for their helpful comments and guidance in the preparation of this article.
Author’s Note:
This article is based on data from the Assessing Police Use of Force Policy and Outcomes Project, supported by Grant 2005-IJ-CX-0055 by the National Institute of Justice, Office of Justice Programs, U.S. Department of Justice. Points of view are those of the authors and do not necessarily represent the official position or policies of the U.S. Department of Justice.