Abstract
The imposition of appropriate sanctions for substantiated police misconduct is important, but social science offers little evidence about whether the severity of sanctions is related to the gravity of the misconduct, or disparities indicative of biased decision-making, or simply arbitrary decision-making. We examine the application of sanctions for police misconduct in one agency, analyzing sanctions imposed in 1,946 sustained personnel complaints filed against 1,026 officers between 1987 and 2001. We find that the nature of the misconduct accounts for some of the variation in sanction severity, with internal bureaucratic violations sanctioned more severely than officers’ improprieties vis-à-vis citizens. Among the latter, sanctions tend to be less severe than those prescribed by either citizens or police officers. We also find that sanction severity is affected by officers’ characteristics, but also that much of the variation in sanctions is unaccounted for, suggesting a large degree of unstructured decision-making.
Introduction
The investigation and disposition of complaints against police officers is an integral function of police administration and is also widely seen as a key element of police accountability. Much of the attention to these issues, on the parts of academicians and community advocates alike, has been fixed on the front end of the complaint process: intake, investigation, and adjudication. Far less attention has been given to the back end of the process, at which sanctions are imposed, inasmuch as the details of these personnel actions are normally shielded from public view. In the absence of systematic data, some observers opine that the punishments imposed on officers tend to be light – too light to serve as effective deterrents to misconduct. 1 Others hold that the sanctions are arbitrary, inconsistent, and excessive – so unfair as to undermine the legitimacy of the disciplinary system in officers’ eyes (Crank, 2004; Curry, 2004; Reynolds et al., 2018; Reynolds & Hicks, 2015). But the empirical evidence on the imposition of sanctions for police misconduct is thin; rarely do social scientists obtain and analyze data on sanctions to produce findings against which these suppositions could be compared. Hence the scientific literature provides no persuasive answers to questions about the severity of police discipline, whether the sanctions are proportional to the gravity of the misconduct, and the extent to which they exhibit either identifiable disparities that may be indicative of biased decision-making or inexplicable variance that may reflect the kind of arbitrary and capricious decision-making that is sometimes alleged.
The operation of police disciplinary systems is, in addition, a window on the larger police bureaucracy, albeit one that is normally shuttered to academics and other outside observers. Writing in 1981, Michael Brown observed that “the focus of bureaucratic rules and regulations in a police department and the preoccupation of supervisors is with corruption, drinking and sleeping on duty, marital infidelity, the destruction and abuse of city property, an officer’s appearance, and the like. In fact, supervisors are more concerned with the enforcement of minor rules and regulations than with a man’s conduct on the street …” (1981, p.88). Brown (and many others) characterized police departments as “punishment-centered bureaucracies” (Gouldner, 1954), and notwithstanding the rhetoric that police discipline should not be punitive, police disciplinary systems rely on punishment as a response to officers’ violations of rules and regulations (see Violanti, 2011). The effectiveness of such negative discipline in improving police officers’ performance has been questioned (e.g., Stephens, 2011). Moreover, we might reasonably speculate that how and how severely officers are punished, and for what, reflects bureaucratic priorities. Departments that expect officers to be “productive” in an enforcement sense, making arrests and issuing citations, might recognize that productivity could be discouraged if occasional excesses or mistakes of judgment are punished severely. The Los Angeles Police Department (LAPD) in which Brown conducted his field work in the 1970s, and the LAPD scrutinized by the Christopher Commission 20 years later (Independent Commission, 1991), exemplify this organizational model. And though police administrators of the 21st century are probably less concerned than their 20th century counterparts with their subordinates’ marital fidelity, a managerial priority on bureaucratic compliance would translate into harsher sanctions for administrative violations than for the mistreatment of citizens.
In this paper we analyze the sanctions imposed for substantiated misconduct by one large northeastern police agency over a nearly 15-year period. 2 We analyze sanctions imposed for all complaints, including violations of widely varying gravity filed by citizens as well as those that originated internally. We examine the extent to which the severity of the sanctions corresponds to the gravity of the misconduct, the degree to which penalties for the same categories of misconduct exhibit variation, whether that variation is associated with subject officers’ prior records of misconduct, race, sex, rank, or length of service, and whether the source of the complaint – citizen or other police – is associated with the severity of sanctions. Our objective is to advance our understanding of the sanctions imposed for misconduct and of the forces that shape the application of disciplinary sanctions.
Influences on Sanction Severity
The literature on police disciplinary sanctions is scant, so it can be reviewed without spilling much ink. We first summarize several conceptual treatments of police discipline, drawing on the concepts of just deserts (von Hirsch, 1992) and distributive justice (Cohen,1987). These expositions underscore the importance of fairness and consistency in disciplinary outcomes, and identify a number of factors that might properly be considered in fixing punishments for misconduct. Then we summarize the findings of the few empirical studies that have analyzed either sanctions as outcomes of the disciplinary process or features of the process itself. From these bodies of literature we distill some hypotheses about the forces that influence the severity of sanctions.
Conceptual Analyses
An ideal police disciplinary system would functionally adhere to the concepts of just deserts and distributive justice in order to achieve fairness and consistency in both its processes and outcomes. Just deserts imply that sanctions for police misconduct would be proportional to the gravity of the misconduct, such that more serious violations would be punished more severely. Distributive justice in the administration of police discipline implies that similarly-situated officers should experience similarly severe sanctions. Other extraneous factors, such as the officer’s race or gender, would bear no impact on sanctions, nor would sanctions significantly vary across the supervisory personnel who make disciplinary decisions, thus minimizing the degree to which such criteria could vary across decision-makers in fixing discipline.
Just Deserts
Judgments about the appropriate severity of sanctions for the various forms of misconduct, and their proportionality to the gravity of the misconduct, are normative and open to different perspectives. One perspective is that of the police department and, as noted above, the seriousness with which different forms of misconduct are sanctioned would reflect bureaucratic priorities. But of course other views of proportionality would spring from other perspectives on the relative seriousness of different forms of misconduct.
Another perspective is that of citizens, on which some evidence can be found in a study of citizens’ judgments about police misconduct (Seron et al, 2006). As Seron et al report, the “baseline” severity that citizens considered appropriate for the least serious form of misconduct that they examined – discourtesy – was 17.8 on a 100-point scale, equivalent to a warning placed in an officer’s personnel file. Specific elements of the officer’s or citizen’s behavior increased or decreased the appropriate punishment to as much as 35.2 – corresponding to the loss of vacation days – or as little as 6.5. The baseline punishment severity for discourtesy plus the use of unnecessary force was 30.0, equivalent to a loss of 1 week’s vacation, but again specific elements of the officer’s or citizen’s behavior increased or decreased the appropriate punishment to as much as 57.4 – corresponding to a suspension without pay for between a week and a month – or as little as 17.8.
Still another perspective on appropriate sanctions is of course that of officers, on which some evidence can be found in a study of police integrity by Klockars et al (2000). Surveying officers in 30 different agencies, focusing mainly on garden-variety corruption, Klockars et al found that officers tended to regard a verbal reprimand as the appropriate sanction for accepting gratuities of modest value, an unpaid suspension as the appropriate (and expected) sanction for excessive force following a foot pursuit, and dismissal as the appropriate sanction for on-duty theft and for accepting a bribe to not issue a traffic ticket.
Unfortunately, no evidence is available on citizens’ or officers’ views of appropriate sanctions for many other kinds of violations. We would not expect that citizens and officers would subscribe to identical views, on average; we might expect that officers would be more cognizant of the challenges of meeting the exigencies of the street within the parameters of bureaucratic regulations, leading to prescriptions for less severe sanctions, though we could also speculate that officers would be less likely than citizens to overestimate the frequency and seriousness of the threats that they confront, leading to more severe sanctions. That these two studies – Seron et al., (2006) and Klockars et al., (2000) – show agreement on the appropriate sanctions of excessive force is itself remarkable.
Distributive Justice
The distributive justice with which sanctions are applied turns to some extent on the factors that have a proper and legitimate influence on judgments about sanction severity – that is, the criteria by which one would (and would not) assess two cases as similar. The International Association of Chiefs of Police (2006), in considering the challenges of ensuring fairness in the disciplinary process, note a number of factors that arguably have a legitimate bearing on sanction severity, including the officer’s work record – “length of service, performance on the job, ability to get along with fellow workers, and dependability” – and other mitigating factors. They also observe that histories of misconduct are relevant in using progressive discipline: a concept which holds that when employee discipline is warranted, it is most effective to employ the minimum level that will alter employee behavior, and then mete out increasing levels of severity based on reoccurrences. The length of time between the occurrence of the current and prior instances might also be appropriate to consider.
Other factors that could affect discipline but are extraneous would include demographic characteristics of the officers such as their race or gender, and prior research has examined whether such factors have played a role in the sanctioning of officers (see below). Other inequities can arise as different decision-makers may use different criteria in fixing discipline and collectively produce disparate sanction outcomes for similar acts under similar circumstances (Stephens, 2011). We would expect this particularly for the less serious forms of misconduct, sanctions for which in many agencies are determined through a decentralized structure instead of via a single unit (e.g., Internal Affairs).
The violation of either of these principles – just deserts and distributive justice – would form a valid base for claims that a disciplinary system is unfair. Perceptions of inconsistent and unfair discipline have been well-documented in previous research on police. From officers’ perspective, the situationally-contingent work that they perform on the street cannot be guided by, or properly performed within, bureaucratic regulations (see, e.g., Manning, 1978; Brown, 1981). Furthermore, they see supervisors’ and administrators’ efforts to enforce rules as arbitrary, capricious, unpredictable, and uneven (Crank, 2004; Curry, 2004; Reynolds et al., 2018; Reynolds & Hicks, 2015). “Inconsistent discipline and enforcement of rules” is one reason that it is the organization, rather than the hard realities of the street, that is the greater source of police stress (Finn & Tomz, 1997; also see Crank & Caldero, 1991; Toch, 2002; Violanti & Aron, 1993) and an object of police cynicism (Niederhoffer, 1969).
Existing empirical research provides some fairly persuasive evidence that these perceptions – accurate or not – are real in their consequences for officers’ performance. For example, in their study of the Philadelphia Police Department, Wolfe and Piquero (2011) found that “organizational justice” – that is, officers’ perceptions of the fairness with which organizational rewards and sanctions are distributed, the fairness of rules and disciplinary procedures, and the fairness with which officers are treated – bore a substantively and statistically significant relationship to officers’ misconduct in the form of citizen complaints, Internal Affairs (IA) investigations, and disciplinary charges. Similarly, Tyler et al (2007) found among officers in a city police department and a federal law enforcement agency that the procedural justice with which the organization was perceived to operate shaped officers’ views of its legitimacy, and legitimacy in turn shaped officers’ conformity to organizational regulations.
In an analysis of the effects of disciplinary sanctions in one large agency, Harris and Worden (2014) found that officers who are sanctioned are both more likely to engage in further misconduct and to do so more quickly when compared to unsanctioned officers. The most plausible interpretation of this pattern, they believed, is that of a “defiance” effect, as officers react to substantive and/or procedural injustice that they perceive in the administration of discipline (cf. Sherman, 1993; also Bouffard & Piquero, 2010; Paternoster & Piquero, 1995), though their interpretation was speculative, as they had no data on officers’ perceptions.
Empirical Findings on Police Discipline
Although limited, prior research has tested data on sanctions for evidence of bias, particularly along racial and gender lines, as well as the degree to which sanctions are influenced by career features such as rank and length of service, or features of the disciplinary process itself. Rojek and Decker (2009) analyzed data from the pseudonymous Centerville police for the 5-year period 1999 to 2003. Distinguishing major disciplinary actions, defined as suspensions and demotions in rank, from minor sanctions such as reprimands, they found that officers’ race was weakly related to sanction severity, with an estimated effect that could not be reliably differentiated from zero. Neither did they uncover compelling evidence of gender bias. Factors that had substantial bearings on disciplinary outcomes included officers’ length of service and prior records of sustained complaints, and characteristics of the incident including the number of allegations, the number of involved officers, and the source of the complaint. Officers with longer service were somewhat less likely to be punished severely. Officers with prior sustained complaints were more likely to suffer major discipline, though the effect did not achieve a customary level of statistical significance. Complaints that involved multiple allegations, those that originated internally, and those that involved a single officer, were all more likely to eventuate in major discipline. But no other control was imposed for the nature of the misconduct.
Fyfe et al (1998) analyzed the filing of charges and specifications in the New York City Police Department’s higher tier of punitive discipline, covering more serious and chronic violations. They found disparities by race and ethnicity that stemmed only from mandatory, non-discretionary charges relating to off-duty behavior, and no disparity by gender. They also found, however, that rates of formal discipline were inversely related to officers’ rank.
Curry (2004) found no racial bias in the imposition of sanctions in the Lansing (MI) police in the 1998–2003 period. His tables (7 and 8 on pp. 13–14) also show that the severity of the sanctions was directly related to the seriousness of the misconduct: 14% of major violations resulted in termination and 28% resulted in suspension, while 5.3% of minor violations resulted in termination and 15.3% in suspension. Other factors that might bear on the severity of sanctions were not examined. Curry (2004) further reports that in his focus group interviews with Lansing police, “inconsistency and unfairness in disciplinary penalties was also a frequently expressed opinion. One attendee stated, ‘they drop the hammer on one guy and give a verbal to another doing the same thing’” (p. 4).
Hickman et al (2000) tested for gender bias in disciplinary sanctions, focusing on internally-generated charges against patrol officers and disposed by the Philadelphia Police Department’s Police Board of Inquiry (PBI) during the years 1991 to 1998. Limiting their analysis to cases of first offenses, among which about two-thirds resulted in a sanction more severe than a reprimand, they found women disproportionately represented among those punished more severely, though they attributed the disproportionality not to post-charging discrimination by the PBI but rather to differential charging by police supervisors. 3
More recently, Mrozla (2021) analyzed the dispositions of complaints filed against patrol officers in a mid-size police agency in the Midwestern U.S. Sanctions were analyzed as binary outcomes: “minor” or “severe” discipline. Prior allegations against the officer increased the likelihood of severe discipline by 24%. Severe discipline was also associated with the duration of the investigations, the number of officers involved in the incident, and the level of subject officers’ arrests.
Shane (2012) reported on a qualitative analysis of disciplinary proceedings in the Newark (NJ) police. Shane observed department trial board hearings, from which major discipline is dispensed (i.e., violations subject to more than 5-day suspensions), and command-level disciplinary conferences, which can result in minor discipline. Both kinds of proceedings, he reports, are informal and subject to wide discretion by hearing officers and board members. No formalized system provides direction about appropriate punishment or information about previous punishments. Decisions were based on discussions about “the quality of the investigation …; whether the officer was known to any trial board members and what they personally thought of the officer; the officer’s longevity; the length and severity of the officer’s disciplinary history; and the nature of the infraction” (Shane, 2012, p. 72). Department policy also stipulates the relative severity of punishments, and it establishes five types of infractions that are subject to major discipline and are thus defined as more serious: aggravated insubordination; criminal acts; unauthorized discharge of firearms; refusal to submit to drug screening; and violations of radio discipline. Otherwise, however, policy provides for no proportionality in sanctioning.
Hypothesized Influences on Sanction Severity
The central research question we address is: what factors influence the severity of disciplinary actions for police misconduct? Specifically, to what degree is sanction severity proportional to the seriousness of the misconduct? Is discipline appropriately severe? As well, to what extent do the disciplinary outcomes adhere to principles of distributive justice?
From the current literature we can derive several hypotheses about the factors that influence the severity of disciplinary sanctions for police misconduct, which we organize in terms of just deserts and distributive justice.
Just Deserts
(1) The severity of sanctions is proportional to the seriousness of the misconduct. We might expect this pattern to hold both as a matter of fairness and as a matter of affirming the values of the organization. That the alternative null hypothesis rests largely on the concerns that prompt discussions about discipline matrices, and the generality of Shane’s observations in Newark: that the effect of the type of misconduct is weak, without sharp distinctions in how different types of violations are punished and with wide variation around a central tendency for any one type. (2) The severity of sanction is greater when the complaint originates internally, rather than being filed by a citizen. We would expect such a pattern insofar as police managers see violations that occur in the course of police-citizen interactions – e.g., discourtesy, or the use of more force than necessary – as to some extent inevitable errors of judgment or spontaneous acts or statements in the heat of the moment, compared with presumably more willful violations of administrative rules that signal a breakdown of internal regulation.
Distributive Justice
(3) The severity of sanctions varies more widely, other things being equal, among cases disposed at lower command levels. Insofar as different commanders hold different views about the seriousness of particular violations, the fairness and utility of sanctions, and the weight (if any) to be given to various aggravating and mitigating factors, then we would expect to find greater heterogeneity in the sanctions in cases handled at the precinct level rather than at the departmental level (by internal affairs), holding seriousness and prior history constant, even if the heterogeneity is not patterned by race, sex, or rank. As a corollary, however, we would expect greater biases by race, ethnicity, sex, and perhaps rank at the lower level. (4) The severity of sanctions is greater when the officer in question has a history of substantiated misconduct, and the worse the history – the more numerous the previous violations – the greater the severity of sanctions, as progressive discipline would dictate.
4
(5) The severity of sanctions is impacted by an officer’s demographic and service characteristics. Specifically: (a) The severity of sanctions is mitigated, ceteris paribus, in proportion with the officer’s length of service. We would expect such a pattern if police managers give consideration – and the benefit of any doubt about the degree of culpability – to officers who have accumulated years of service. As a corollary, we would expect that sanctions will be particularly severe for academy trainees, in whom an agency has a fairly modest investment, and who have not established themselves as competent officers with some credibility. (b) The severity of sanctions is greater for racial and ethnic minorities and for women. We would expect such a pattern insofar as police managers subscribe to stereotypes about the unsuitability of minorities and/or women for police work, or make presumptions that minorities and women were held to a lower standard for hiring and/or academy training. Previous studies have not detected such disparities, but we cannot reject such hypotheses on that limited basis. (c) The severity of sanctions is greater for officers of lower rank. Rank may have its privileges. But rank also may signify a level of service to and/or achievement in the agency that is a mitigating factor in fixing punishments.
Data and Methods
In the course of examining the early intervention system of a large, full-service police agency in the northeastern United States, we gathered data on personnel complaints. 5 The agency has a reputation as a very professional, bureaucratized organization with high personnel standards and tight internal controls. The agency takes personnel complaints in person, by mail, by phone, by fax, and via e-mail from the agency’s website. Any officer can take a complaint, whereupon it must be filed with the IA division. There is no “statute of limitations” on the filing of complaints, meaning a complaint can be filed any time after the incident, regardless of duration. This agency also takes anonymous complaints, and so long as a complainant provides enough information that an officer(s) can be identified, an investigation is conducted.
Between 1987 and 2001, the period covered by our inquiry, the agency recorded approximately 740 personnel complaints per year, 15 to 25% of which included allegations against more than one officer, and about one-quarter of which were generated internally (e.g., by supervisors). Overall, 36% of the complaints were sustained, though internal complaints were far more likely to be sustained than civilian complaints (82.6% and 17.8%, respectively). We focus on complaints filed against officers whose service began in or after 1987, since it is only those officers for whom we have a complete history of complaints. Of the personnel hired in this period, 88.9% were men, 78.9% were white, 11.2% black, and 9.3% Hispanic. One or more complaints (6,095 in all) were filed against 1,820 of these officers. We examine 1,946 sustained complaints filed against 1,026 officers. 6 Of these officers, 90.3% are men; 71.6% are White, 17.7% Black, and 9.9% Hispanic.
For each complaint we have information on the nature of the violation(s), the source of the complaint (citizen or agency personnel), the date of the incident, the date on which the investigation was finalized, the level(s) – precinct or agency – at which the case was investigated and closed, the sanction(s) imposed, and the officer’s rank and precinct assignment at the time of the complaint. For each officer we also have his/her date of hire, from which we derive for each complaint a precise measure of the officer’s length of service at the time of the incident, and we also construct for each complaint counts of previous sustained complaints (citizen and internal, respectively). We also form an indicator for officers who at the time of the complaint had not yet completed the 6-month recruit academy. 7
Complaints were closed at either the department level, by IA, or the precinct level, whether or not they were investigated at the same level. IA conducted only 129 of the 1,946 investigations (6.6%), and most of those (71%) were also closed at the department level. Of the other 1,817 cases investigated at the precinct level, 23% were closed at the department level. Complaints investigated and/or closed at the department level were generally the more serious allegations, and/or those against officers with prior histories of complaints, as we show below. 8 Insofar as the more serious allegations were selected for department investigation and/or closure, we can think of these features of a complaint partly as additional indicators of seriousness (at least from the agency’s perspective), though as we show below, they are also features of the discipline process that structure the outcomes.
Decisions about disciplinary sanctions were determined through a “consultative” process involving the interested (agency) parties. At the precinct level, at which the maximum punishment was a letter of censure, a loss of five vacation days, and a transfer, the precinct commander would consult with the subject officer and a representative of the officer’s bargaining unit, whereupon the officer would sign a waiver indicating that s/he would accept the precinct commander’s punishment; the outcome was also subject to approval by the IA commander, though higher-ranking executives might also become involved depending on the particulars of the case. A similarly consultative process prevailed in cases handled at the departmental level. Thus the process bore a resemblance to plea negotiations. We have no way of knowing how common or unusual such a “consultative” process is among police departments’ disciplinary systems, or how closely the outcomes resemble those ultimately produced in agencies whose decisions are frequently altered in binding arbitration. But we would note that procedural fairness cannot be assumed even when, in the end, the punished does not contest the punishment; there is nothing intrinsic to the plea bargaining process that ensures the hallmarks of procedural fairness (O’Hear, 2008), and many felony defendants who cop pleas nevertheless feel a sense of procedural injustice (Casper et al, 1988).
Descriptive Statistics.
Our principal measure of the nature of the misconduct is derived from the agency’s 71 individual complaint classification codes, up to 4 of which were used to describe the allegations in any one complaint. We distilled this set of codes into a more parsimonious set of 13 categories that are substantively coherent and that also preserved, as much as possible, the distribution of sanction severity in the original categories (Appendix A provides detail regarding where each of the 71 complaint classification codes were placed within each of our 13 categories); thus we maximize the degree to which the nature of the misconduct accounts for the severity of the sanction imposed. See Table 1. But of course these categories mainly reflect differences in kind, while differences of degree remain within categories. Complaints that included allegations in multiple categories were classified in multiple corresponding categories, so the set is not mutually exclusive, and the sum of the percentages in Table 1 thus exceeds 100.
Individual officers’ identities are stable, of course, and so are their sex, race, and ethnicity, but other characteristics can vary across their careers, so we display descriptive statistics as if officers’ characteristics – their length of service, rank, sex and race – were features of the complaints filed against them. Officers’ race and ethnicity is measured in terms of the categories specified in the agency’s records: White, Black, Hispanic, Asian-American, and Native American. Complaints against White and Black officers are by far the most numerous (63.8% and 25.2%, respectively), and given their small numbers, we collapse other race/ethnic groups into a non-white category, though most of those complaints (all but 8, or 96%) are against Hispanic officers. 10 percent of complaints were filed against female officers.
Most sustained complaints (93.8%) are filed against personnel of patrol officer rank, but 3.4% were filed against detectives, 2.4% against sergeants, and only .5% against higher-ranking officers (mostly lieutenants and captains). Given the small numbers of complaints filed against each of the ranks higher than sergeant, we combine those rank categories.
Officers’ length of service is measured in years. Among complaints filed and sustained against officers whose service began in 1987 or later, the typical sustained complaint was filed against an officer with 5.7 years of service. Restricting the analysis to officers who were hired after 1986 also restricts the range of officers’ length of service, though complaints decline in number after year three of service (Harris, 2009).
Analytical Strategy
We examine simple crosstabulations of sanction severity by the nature of the misconduct, but our principal analytic strategy is an ordered logit regression, first of all sustained complaints, and then of the two separate subsets: those closed at the department level, and those closed at the precinct level. Ordered logistic regression is a statistical technique that can be used with an ordered (from low to high) dependent variable, here sanction severity.
We estimate the parameters of a regression with robust standard errors for clustered data, inasmuch as complaints are to a degree nested within officers (though as we observed above, officers’ rank and length of service may vary for individual officers). Because some officers are included multiple times in the dataset, the observations are not always independent. Using the estimator, our robust standard errors relax the usual requirement that all observations be independent by allowing for intragroup correlation within groups of repeated officers. We also report the p-values of the logit coefficients and note those that achieve conventional levels of statistical significance, for readers who are interested. But we caution that for the 1987–2001 period, we analyze the entire population of substantiated complaints against officers whose careers began no earlier than 1987, and for this population, the estimated coefficients are not sample statistics subject to sampling error; as a sample, the complaints we analyze are a non-random sample.
Findings
Sanction Severity by Seriousness of Misconduct.
Be that as it may, Table 2 also shows that sanction severity varies within many of the categories. Sanctions imposed in use of force complaints exhibit the greatest variation, with 30% resulting in no sanction, nearly as many resulting in a less severe sanction, and 40% resulting in a more severe sanction. But sanctions also vary considerably within other categories, such as insubordination, procedure, and unlawful/improper police action. This variation could reflect the heterogeneity of misconduct within the categories – differences of kind or degree – and/or (1) variation across decision-makers (e.g., precinct commanders), and (2) the consideration of criteria other than the misconduct itself, such as an officer’s history of misconduct. Nearly three-quarters of these cases were closed at the precinct level, and so decision-making was susceptible to a wider range of criteria.
Insofar as complaints may involve multiple forms of misconduct, however, these bivariate results cannot isolate the independent contributions of particular forms of misconduct to sanction severity. Nearly half – 47% – of the complaints involved two or more allegations in different categories; 11% involved at least three, and 2.7% involved four. Some categories overlapped with others more frequently: e.g., 81% of the integrity complaints also contained another allegation, as did 86% of the complaints for insubordination.
Ordered Logistic Regression of Sanction Severity.
aAdministrative violations omitted as reference category.
*p < .10.
**p < .05.
Other hypotheses are also confirmed, or partially confirmed, by the regression results. Complaints that are investigated departmentally, by IA, tend not to result in more severe sanctions, but citizen complaints are 44% less likely to result in a most severe sanction than internally-generated complaints, even controlling statistically for the nature of the misconduct, thus providing some support hypothesis 2 which supposed that sanction severity is greater when complaints originate internally as these are more likely to be deemed willful violations of rules.
Several characteristics of the officer affect sanction severity at the margin. Misconduct by academy trainees is over 6 times more likely to be severely sanctioned than that of other officers, providing support for hypothesis 5a that academy trainees would be subject to harsher discipline because they have yet to establish themselves as competent officers. Officers’ records of prior sustained internal complaints – but not of sustained citizen complaints – affect sanction severity, thus partially supporting hypothesis 4 that progressive discipline would yield greater penalties as officers accumulate complaints. Each additional sustained internal complaint in an officer’s record increases the likelihood of another severe sanction by 39%.
Furthermore, longer tenure or service mitigates sanction severity somewhat, also providing support for hypothesis 5a. Each year of service reduces the likelihood of a most severe sanction by about 9%. Officers who are neither Black nor White – mostly Latino – are sanctioned somewhat more severely, with non-White officers 39% more likely to receive a most severe sanction relative to White officers. But officers’ sex has no substantively or statistically significant effect on the severity of sanctions, and officers’ rank has a weak, if any, effect. These results provide only partial support for hypothesis 5b and 5c, which predicted that sanctions would be greater for racial minorities and for women.
Sanction Severity by Level of Closure.
Ordered Logit Analysis of Sanction Severity, Cases Closed at Precinct or Department Levels.
aAdministrative violations omitted as reference category.
*p < .10.
**p < .05.
Discussion
A well-administered regime of disciplinary sanctions for police misconduct could be expected, a priori, to serve several purposes. The imposition of appropriate sanctions for violations of policy or law serves to hold officers accountable for their actions. Sanctions might also have deterrent effects: a specific deterrent (or corrective) effect on the officers who are punished, discouraging repeat misconduct, and a general deterrent effect on all officers, discouraging misconduct through the prospect of punishment. Furthermore, sanctions might be expected to contribute to the regulation of police behavior through an educative process, insofar as they affirm the values of the organization and define behavior that is unacceptable. Finally, to the extent that sanctions are meted out fairly and proportionately in accordance with procedural justice, officers’ adherence to departmental rules and regulations might increase, thereby enhancing the deterrent effects of the disciplinary system.
In any analysis of a well-administered regime of disciplinary sanctions for police misconduct, one would expect to find that similar categories of substantiated misbehavior to receive sanctions of similar severity, and that severity, while not objectively defined, would bear a consistent relationship to the seriousness of the violation. Residual variation in the severity of sanctions would mostly turn on an officer’s prior disciplinary history, with more repeated violations being punished more severely, as progressive discipline would dictate. The remainder of the variation in sanction severity would be attributable to features of the disciplinary process, such as the level at which complaints are investigated and disposed. Little to no variation would be attributable to the personal characteristics of the officer such as their race, or sex, or their occupational characteristics, such as their rank. Most importantly, any model which incorporates these factors should be able to account for the bulk of the variation in sanction severity.
Our empirical inquiry into one department’s disciplinary outcomes reveals that the severity of sanctions varies with the nature of the misconduct, and it also shows that sanction severity is affected – properly or improperly – to some degree by factors other than the nature of the misconduct, but with a substantial fraction of the variation in severity unaccounted for by the variables in our analysis. Above we posed two central research questions: First, to what degree is sanction severity proportional to the seriousness of the misconduct? Is discipline appropriately severe? Because seriousness is not an objective property of acts of misconduct, these questions of just deserts can be addressed only by invoking some benchmarks against which to assess the proportionality of the sanctions. Second, to what extent do the disciplinary outcomes adhere to principles of distributive justice? Concerns about the fairness with which discipline is administered arise when we see substantial variation in sanction severity, even within categories of misconduct, which either is associated with factors that should not have a bearing on disciplinary outcomes or that cannot be explained in terms of other legitimate factors.
Just Deserts
We might consider whether the sanctions imposed for the various forms of misconduct are of appropriate severity, and proportional to the gravity of the misconduct from both citizens’ and officers’ perspectives based on prior research discussed above (Klockars et al, 2000; Seron et al, 2006). Recall that, from a citizen’s perspective, Seron et al (2006) report the “baseline” severity that citizens considered appropriate for the least serious form of misconduct that they examined – discourtesy – was equivalent to a warning placed in an officer’s personnel file. The baseline punishment severity for discourtesy plus the use of unnecessary force was that equivalent to a loss of 1 week’s vacation.
Against these standards for appropriately severe sanctions, the sanctions examined here are arguably light, on balance. If by citizens’ lights the baseline discipline for the use of unnecessary force is a loss of 1 week’s vacation, then certainly the force cases in which no sanction was applied are not commensurate with citizens’ prescriptions. A distribution of sanctions that is more in line with citizens’ expectations would probably be centered in the more severe category examined here, including losses of vacation and unpaid suspensions, but fewer than half of the sanctions in use of force cases are that severe, and the values predicted by the regression model are concentrated instead in the less severe category of severity.
Another perspective on appropriate sanctions is that of officers. Recall that Klockars et al found that officers tended to regard a verbal reprimand as the appropriate sanction for accepting gratuities of modest value, an unpaid suspension as the appropriate sanction for excessive force, and dismissal as the appropriate sanction for on-duty criminal conduct (both theft and for accepting a bribe). Against these standards, some of the sanctions examined here are more severe and others are somewhat less severe. Gratuities were the subject of only 4 substantiated complaints but resulted, in 3 cases, in more severe sanctions – more severe than a reprimand. Sanctions in use of force cases, as noted above, were mainly less severe than unpaid suspensions. But on-duty criminal conduct resulted in the most severe sanction in 3 of 4 cases.
Insofar as the severity of sanctions imposed for various forms of misconduct is a behavioral indicator of the agency’s own views of seriousness, we can surmise that it is deviations from bureaucratic imperatives that are treated as the more serious violations, compared with the propriety of police action vis-à-vis the citizenry. Citizen complaints eventuate in less severe sanctions, on average and other things being equal, than internal complaints. Integrity lapses, insubordination, and even traffic violations are, on average, sanctioned more severely than the improper use of force, other unlawful or improper police actions, and “service” complaints. To the extent that police discipline systems serve an educative function, affirming the values of the agency, then these kinds of patterns arguably send signals that the public would find questionable.
Distributive Justice
The nature of the misconduct accounts for about half of the explained variation in severity. Other factors also could legitimately shape disciplinary outcomes, as we discussed above: officers with histories of substantiated misconduct might appropriately be subjects of progressive discipline; officers with longer service, other things being equal, might appropriately have their service taken into account in mitigating the severity of discipline. But some of the patterns revealed by the results might reasonably raise questions about the fairness with which discipline is administered. The effect of officers’ ethnicity is modest but detectable. Rank appears to affect the severity of sanctions, to the benefit of higher-ranking personnel, at the precinct level.
Moreover, much of the variance in sanction severity is not accounted for by the model, and the model accounts for still less, proportionally, among the more routine and numerous cases handled at the precinct level. Some of this unaccounted for variation is surely legitimate, based on mitigating and aggravating particulars of the conduct and the context in which it occurred – particulars that social science is hard-pressed to accommodate in parsimonious theories and practical measures, and that are beyond the level of detail in our data. But the remainder of the unexplained variation can be plausibly attributed to the formally unstructured discretion of command staff responsible for making decisions about discipline. And this seemingly unpatterned – and hence arbitrary and arguably unfair – variation in sanction severity is the outcome that we would expect if we took as a given the extant research reports of officers’ perceptions of police disciplinary systems. Officers engaging in very similar forms of misconduct may receive disciplinary sanctions of varying severity.
Not all rules serve the purpose of legitimating managerial prerogatives equally well in police organizations; it may be the rules governing internal compliance more than those governing the delivery of police services that tend to preoccupy police managers and supervisors, as Brown (1981) contends. The findings of our inquiry are compatible with this portrayal, inasmuch as sanctions for violating rules concerning internal compliance tended to be more severe than those for violations of rules governing police-citizen interactions.
While this study provides a much-needed examination of an understudied issue, it is not without caveats. First, these data originate in a single agency and cover a time span mainly in the late 20th century. Policing has changed in a number of respects in the past 20 years, as demands for police accountability have intensified, and accountability mechanisms – including early intervention systems, discipline matrices, and civilian oversight bodies – have further proliferated (see, e.g., Harris & Perry, 2022). Whether and how much these changes have altered patterns of police discipline are open questions, however. Existing evidence cannot tell us how generalizable these findings are to other agencies or to other times. The purpose of our research is to contribute to a body of empirical evidence, on the basis of which tentative inferences can be drawn about how police organizational systems work (and fail to work), and on which future research can further build. If these findings are time-bound, then it will behoove the field to theorize about what has changed over time and how those changes affected the administration of sanctions for misconduct.
Second, these data lack the granularity that would be ideal to represent the seriousness of the many forms that police misconduct takes. The agency under study utilized a large number of categories for complaints filed against officers and we strove to preserve the relevant variation and at the same time establish a set of analytically tractable categories, but unmeasured variation in seriousness remained. Finally, we do not have data on the perceptions of the officers in the study agency with respect to rules and rule enforcement, or “organizational justice” more generally; neither do we have data about the practice of positive discipline in the agency, such as advising, coaching, or mentoring. 11 And unfortunately, police research does not, to our knowledge, offer a basis on which to speculate about whether and, if so, how officers differentiate among the rules to which they are subject – that is, whether some are regarded as legitimate rules and others as illegitimate.
Conclusions
No one would claim that sanctions for police misconduct are unimportant. They are significant both as a matter of justice and accountability, potentially providing appropriate consequences for actions or omissions by officers, and as means to the end of regulating officer conduct. And surely few would dispute that discipline should be administered with procedural and distributive fairness. Police executives, police officers, and the public all have a stake in disciplinary systems that operate, and are perceived to operate, with fairness. Fairness is of intrinsic value, and it also may have instrumental value. Previous research gives us reason to believe that discipline that is administered in ways that officers experience as unfair will lead to outcomes that are detrimental for the officers, their agencies, and the communities that they serve.
We know very little about the actual operation of police disciplinary systems, as only a handful of studies have empirically investigated the disciplinary sanctions imposed by police agencies for substantiated misconduct, from which we can draw some inferences about the process, and only one study (Shane, 2012) has described the process by which those outcomes are realized. The generalizability of those findings is unknown. We suspect that our findings may be limited to departments that are “professional” in the sense that Brown used that term – fairly well insulated from external political pressures, highly bureaucratized, and with a strong emphasis on command-and-control – and perhaps to departments that practice the consultative process of the study agency. Judgments about external validity obviously must await further research in other agencies, and further judgments about the appropriateness of the sanctions imposed will require additional evidence on what various stakeholders consider appropriate.
So too must conclusions about the efficacy of police discipline await further inquiry in other agencies. The effectiveness of systems that rely on negative discipline has been challenged, but seldom systematically evaluated. The effectiveness of alternative approaches, or even adjustments (such as discipline matrices) to traditional systems, is also unknown. These are important questions, whose answers have implications for the quality of police service that citizens receive and for the quality of work life that officers experience.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Notes
Appendix
Categorization of classification codes
Improper attire
Lack of military courtesy
Moonlighting
Sexual harassment (2 codes)
Tardy
Untimely response
Criminal association—place
Criminal association—person
Criminal/illegal conduct—on duty
Criminal/illegal conduct—off duty
Penal law violation
Alcohol use
Drug use
Damaged issued equipment
Damaged other property
Improperly maintained equipment
Improperly maintained police vehicle
Loss of issued property
Unsafe operation of police vehicle
Firearms acquisition
Firearms discharge
Firearms—excessive force
Firearms—other
Firearms transfer
Improper use of firearms
Excessive force not involving a firearm
Improper use of force not involving a firearm
Physical abuse—other than prisoner
Physical abuse of a prisoner
Improper police action
Unlawful arrest
Unlawful search
Absent without leave
Disobeyed directive
Disobeyed instructions
Disobeyed order
Insubordination
Abuse of authority
Abuse of sick leave
Accepting a gratuity or reward
Bribery
False entry in records
False report
Unethical conduct
Communications system misuse
Confidential disclosure
Escaped prisoner
Failure to safeguard a prisoner
Failure to safeguard property
Incompetence
Improper handling of evidence
Neglect of duty
Failure to assist motorist
Harassment
Insensitive
Poor judgment
Racial discrimination
Service complaints
Towing complaint
Tow truck operator complaint
Verbal abuse
Traffic offense
Traffic violation/improper use of a motor vehicle
Discredited agency
Unprofessional conduct