National Academies Press: OpenBook

Fairness and Effectiveness in Policing: The Evidence (2004)

Chapter: 4 Explaining Police Behavior: People and Situations

« Previous: 3 The Nature of Policing in the United States
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 109
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 110
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 111
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 112
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 113
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 114
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 115
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 116
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 117
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 118
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 119
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 120
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 121
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 122
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 123
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 124
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 125
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 126
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 127
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 128
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 129
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 130
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 131
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 132
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 133
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 134
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 135
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 136
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 137
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 138
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 139
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 140
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 141
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 142
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 143
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 144
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 145
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 146
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 147
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 148
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 149
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 150
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 151
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 152
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 153
Suggested Citation:"4 Explaining Police Behavior: People and Situations." National Research Council. 2004. Fairness and Effectiveness in Policing: The Evidence. Washington, DC: The National Academies Press. doi: 10.17226/10419.
×
Page 154

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

4 Explaining Police Behavior: People and Situations C hapter 3 outlines the many things that police do. This chapter ex- amines the forces that influence how, and how much, these things are done. Knowledge about these influences is essential for imple- menting policies that contribute to the fulfillment of the two public expec- tations expressed as core themes of this volume: crime control effectiveness and fairness. Achieving greater effectiveness and fairness depends in large part on the capacity of a society to get its police to carry out legitimate policies designed to further these ends. For example, knowing which prac- tices will reduce domestic violence tells us nothing of how to ensure that officers engage in these practices at the appropriate times and places. Do the background characteristics of officers affect their enforcement practices? Can officers be trained to behave in certain ways, and what sort of training is most effective? Are certain work incentives and disciplinary practices nec- essary? Or to consider an example about police fairness, it is one thing to suggest that police who behave in a disrespectful manner toward citizens are perceived as less fair and less legitimate than those who avoid disre- spectful behavior. But it is quite another thing to determine what causes police to behave disrespectfully toward some citizens and how to devise ways of preventing disrespectful officer behavior. Can officers be trained or disciplined not to be disrespectful? Can citizens be educated to behave in ways that avoid precipitating police disrespect while maintaining their own sense of self-respect? The first step toward answering questions such as these is to appreciate the state of knowledge about the causes of police practice. 109

110 FAIRNESS AND EFFECTIVENESS IN POLICING Although most of the research in this review addresses academic ques- tions about the causes of police practice, this literature has important impli- cations for the central policy question of how to control police practice. "Control" is probably a misnomer if it is interpreted as "ruling" or "deter- mining" what police do--as in the "command and control" model of police administration that became popular as an ideal by the mid-20th century (Wilson, 1963). Most current analysts and reformers accept some degree of police autonomy in decision making as a good thing, or at least an inevi- table one (see, for example, Kelling, 1999; Moore and Stephens, 1991). Thus the term, as it used here, is intended to connote a significant degree of influence on police practice, but one that does not necessarily meet the command and control ideal of determining it. Whenever possible, we at- tempt to draw the implications of extant research for the control of what police do and how they do it. Ultimately, the findings of Chapters 4 and 5 speak to the governability of policing as communities experience it. Ac- countability of the police assumes a capacity to shape, if not determine, what they do. This review covers a wide range of police activities and policies--from how police treat citizens they encounter on the street to the kinds of policies and organizational structures implemented by police departments. The sub- stantial literature involved in this broad range has been divided into four general categories, beginning with the explanatory factors closest to every- day police work, namely, the characteristics of the situations in which offic- ers make decisions, such as whether to make an arrest, use force, or engage in community policing. Such situational characteristics include, for example, the strength of evidence available to an officer about a suspect's guilt, the personal characteristics of the suspect, and the characteristics of the victim. This chapter then examines the characteristics and outlooks of the police officers who make those decisions--such things as their age, race, sex, edu- cation, and training. We examine these two proximate elements: the degree and the ways in which characteristics of people (both officer and citizen) and situations influence police actions. In the next chapter, we consider less proximate but presumably influen- tial factors that affect police behavior. For instance, policies and other char- acteristics of the police organization that may influence police behavior, such as policies on deadly force, structures and styles of supervision, perfor- mance incentive systems, and the nature of department leadership. The com- mittee also examines forces external to police organizations, such as the social and economic makeup of the neighborhoods or jurisdiction served by the police, the political culture of the community, and political processes and decisions made in the jurisdiction, including the law. In order to distill the considerable research on each of the above types of influences, the committee considers a series of commonly expressed views

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 111 about the causes of police practice that have received publicity or consider- ation in the research literature. In many instances, there has not been enough research from which to generalize, but whenever possible the committee states a proposition about these influences based on its evaluation of the research literature. When the evidence is inadequate, the committee indi- cates the sort of research that would help to fill this knowledge gap. Before turning to these propositions, each chapter considers the nature of the evi- dence available in the research literature. NATURE OF THE EVIDENCE The studies reviewed for both this chapter and the next draw on a wide range of measurement and data collection methods and employ a similarly diverse set of designs. However, the preponderance of the literature is con- centrated in certain areas; therefore this discussion is limited to the strengths and limitations of those methods. By far the largest proportion of empirical research on police practice has concentrated on patrol officers, who consti- tute the largest portion of the nation's police force. There are a handful of relevant studies on criminal investigators, juvenile officers, other sworn spe- cialists, and telephone complaint operators and dispatchers. There are even fewer systematic studies of the behavior of police executives and middle management, constituting a major gap in knowledge about the causes of police behavior outside the realm of the rank-and-file patrol officer. Thus, most of the committee's analysis, especially when addressing the literature on the influences on individual officers, is in effect a discussion of what is known about police patrol. Research on the forces that influence police behavior has been based on: (1) police records, such as incident reports or firearms discharge re- ports; (2) direct observation of police in the field; (3) surveys of the public about their contacts with the police; and (4) surveys of police officers. Most of the research on individual officer decision making draws on field obser- vations of police, and much of that can be characterized as systematic social observation (Reiss, 1971; Mastrofski et al., 1998). Such studies employ trained observers who are assigned to accompany officers in selected beats on selected work shifts. Observers take brief field notes on officers' activi- ties and behaviors and on the citizens with whom they interact, and later code data about police actions and other variables according to a standard- ized form. Measures based on observational data are more valid than those based on police records, which serve organizational purposes and hence may be biased or incomplete. Some observational studies have linked obser- vations to surveys of officers, so that observations of individual officers can be combined with the same officers' survey responses.

112 FAIRNESS AND EFFECTIVENESS IN POLICING Several systematic observation studies of police patrol, conducted since the 1960s, have produced a significant number of publications.1 Three of these warrant more detailed description, due to their scale and the extent to which analyses of the data collected for these studies have been used by police researchers. The first large-scale observational study of police was undertaken by Albert J. Reiss, Jr., for the President's Commission on Law Enforcement and Administration of Justice (Black and Reiss, 1967). Con- ducted during summer 1966 in Boston, MA; Chicago, IL; and Washington, DC, observers accompanied patrol officers on sampled shifts in selected high-crime precincts. "In the data collection, emphasis was placed upon gaining detailed descriptions of police and citizen behavior.... The social and demographic characteristics of the participants as well as a detailed description of the settings and qualities of the encounters were also ob- tained" (Black and Reiss 1967:15; emphasis in original). The Police Services Study (PSS), which was funded by the National Science Foundation, was designed to examine the effects of institutional arrangements on the delivery of police services. The second phase of the study provided for the collection of various kinds of data about 24 police departments in 3 metropolitan areas (Rochester, NY; St. Louis, MO; and Tampa-St. Petersburg, FL), with attention focused particularly on 60 neigh- borhoods served by those departments. During summer 1977, trained ob- servers accompanied patrol officers on 900 patrol shifts, 15 in each of the 60 neighborhoods. Observers recorded information about 5,688 police-citi- zen encounters. In addition, the observed officers (and samples of other officers) were surveyed. The departments studied for this phase of the PSS ranged in size from 1 with only 13 officers to 1 with over 2,000, serving municipalities whose populations ranged from 6,000 to almost 500,000. Within jurisdictions, neighborhoods were selected with explicit reference to racial composition and wealth to ensure that different types of neighbor- hoods were represented. The departments and neighborhoods provide a rough cross-section of organizational arrangements and residential service conditions for urban policing in the United States, and thus the PSS data provide a much firmer basis for generalizing about police practices in U.S. metropolitan areas (and not only in urban, high-crime areas). Finally, the Project on Policing Neighborhoods (POPN), which was funded by the National Institute of Justice, provided for direct observation 1See the following for details on the methodology of the more widely published of these studies (Bayley and Garofalo, 1989; Black and Reiss, 1967; Frank, 1996; Frank and Travis, 1998; Klinger, 1994; Mastrofski and Parks, 1990; Mastrofski et al., 1995; Mastrofski et al., 1998; Sykes and Brent, 1980; Worden, 1989). See Riksheim and Chermak, 1993, and Sherman, 1980, for reviews of the publications that drew on some of these projects.

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 113 of police in two cities, Indianapolis and St. Petersburg, during 1996 and 1997, respectively. Observation focused on 12 selected police beats in each city and over 5,700 hours of observation (approximately 30 shifts per beat), yielding information on approximately 11,000 police-citizen contacts. Beats were selected from each of three strata of socioeconomic distress, with se- lection biased toward the more distressed beats, in order to maximize the number of police-citizen encounters subject to observation. In addition, patrol officers and field supervisors in each department were surveyed. Because of the rigor of the methodological design and the scale of systematic observation studies, they comprise the strongest data from which to draw conclusions regarding police behavior. Yet observational data are not without shortcomings. They may be tainted by officers' "reac- tivity" to observation, that is, officers might refrain from some actions (such as the use of force, running personal errands) or engage in other actions (such as stopping cars) due to the presence of observers. Efforts to assess the bias introduced by reactivity suggest that the validity of observa- tional data, in general, is quite high (Mastrofski and Parks, 1990; Spano, 2002); moreover, evidence shows that the relationships between some forms of police behavior and other variables (such as characteristics of the situation) are unaffected by reactivity (Worden, 1989). As Reiss (1971:24) observes, "it is sociologically naive to assume that for many events the presence or participation of the observer is more controlling than other factors in the situation." Observational data have other limitations. Direct observation of police is labor-intensive, making observational studies very costly; only three large- scale observational studies have been conducted. Furthermore, observa- tional studies can be conducted only with the express permission and coop- eration of the police departments, and as Fyfe et al. (1997) suggest, the findings from research in such police departments may not be generalizable to other U.S. police agencies. Observational studies are best suited to inform judgments regarding the proximate and immediate influences at work during a police-citizen en- counter. In the next chapter, data and research methodologies that examine police as organizations or look outside the police force--to the community, for instance--to explain police behavior are examined. Finally, it must be noted that the vast majority of studies in this area rely on correlational designs. A smaller number are case studies, and a very small number use quasi-experiments or experimental designs. Since all stud- ies are subject to error, the committee has rated these studies differently based on the strength of their design--that is, their ability to discount other variables that might explain the behavior under examination. Throughout we committee disclose our judgments regarding the strength and rigor of research design.

114 FAIRNESS AND EFFECTIVENESS IN POLICING SITUATIONAL INFLUENCES ON POLICE BEHAVIOR Situational influences represent forces that operate at what is some- times called the tactical level of police decision making. These represent circumstances that vary from situation to situation and are expected to play a central role in shaping how officers act. Situational influences that have received considerable empirical evaluation include: the social class, race, gender, and demeanor of complainants and their dispositional preferences (e.g., whether they want offenders arrested or prefer that offenders not be arrested); the social class, race, age, gender, sobriety, and demeanor of sus- pects; the seriousness of the offense or problem and evidence available (if any); the nature of the relationships between complainants and suspects; the visibility of the encounters (whether they transpire in public or private locations, whether bystanders are present); the numbers of officers at the scene; and the character of the neighborhoods in which encounters take place. Which situational factors are studied and how they are interpreted de- pends on the researcher's theoretical perspective. For example, Donald Black and Albert Reiss (1967:8-9) posited that police action is influenced by a citizen's "sanctioning capacity," which is, in turn, a function of the citizen's status--both social (gender, age, race, class) and situational (as complain- ant, suspect, witness, etc.)--and by the citizen's "subversive capability," that is, the "capability to undermine the means the police use to attain their goals." From this perspective, situational factors (Sherman, 1980a) are the cues on which officers form judgments about how incidents should be handled (Wilson, 1968; Berk and Loseke, 1981). Perhaps the most compre- hensive statement of situational factors was that of Bittner, who posited that "the role of the police is best understood as a mechanism for the distri- bution of non-negotiably coercive force employed in accordance with the dictates of an intuitive grasp of situational exigencies" (Bittner 1970:46; emphasis added). The situational framework has been applied most fre- quently to the use of coercion by patrol officers, but also to the decisions of juvenile detectives and other investigators (e.g., Bynum, Cordner, and Green, 1982; Brandl, 1993; Terry, 1967). To the extent that Bittner's is a valid and comprehensive description of police work, it suggests that the greatest part of the variation in police officer behavior will be accounted for by establishing those situational exi- gencies that most powerfully shape police action. Other influences, such as the officer's personal characteristics and attitudes, or department policy, would manifest more subtle effects. That is, in fact, the finding of virtually all studies that compare situational influences to officer and organization characteristics (see Riksheim and Chermak, 1993, and Worden, 1989, for reviews).

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 115 One objective in much of the research on situational influences is to determine whether patterns of police behavior are affected by factors that should, in a moral or legal sense, have no bearing on police dispositions, such as race and gender (see Bernard and Engel, 2001). The analytic strat- egy of such research has been to control statistically for the effects of legal factors--particularly the strength of evidence, the seriousness of the offense, and the preferences and cooperation of complainants--which are unam- biguously legitimate criteria for police decision making and to estimate how much if any of the remaining variation in police behavior is attributable to extralegal factors, such as race. Research of this genre has found that most extralegal, situational factors have weak and inconsistent effects. LEGAL FACTORS Proposition 1: There is considerable public concern that police officer decision making ignores the constraints of the law. The evidence re- viewed by the committee indicates that the exercise of police authority to control citizens is most heavily influenced by legal factors associated with each situation, particularly the seriousness of the reported inci- dent, the evidence of wrongdoing, and the willingness of a complainant to request a controlling intervention. Public opinion surveys show that a significant minority of the Ameri- can public regards police as unfair and untrustworthy, and some fear being arrested when innocent (Gallagher et al., 2001; LaFree, 1998). However, evidence reviewed by the committee indicates that officers' use of coer- cion--their decisions to arrest or not and their use of physical force or verbal control--is most heavily influenced by legal factors (Black and Reiss, 1970; Black, 1971; Lundman, 1974; Freidrich, 1977; Lundman et al., 1978; Smith and Visher, 1981; Bayley, 1986; Mastrofski et al., 1995; Mastrofski et al., 2000; Worden and Myers, 1999; Terrill, 2001). In their encounters with suspected offenders, the likelihood that police will invoke their au- thority by making arrests, using physical force, or verbal methods of con- trol rises directly with the strength of the evidence of criminal wrongdoing. So too does the likelihood of coercive action rise with the seriousness of the offense: thus police are more likely to make arrests or use force when the offense is a felony than when it is a lesser offense. But, as noted in Chapter 1, police frequently do not invoke the law, even when they have the author- ity to do so; when they have evidence of offending; or even sometimes when the alleged or suspected offense is a serious one. Important evidence on the influence of legal factors comes from major observational studies. Black (1971), for example, found that police were less likely to arrest when they did not observe the offense themselves and had to rely on citizen testimony

116 FAIRNESS AND EFFECTIVENESS IN POLICING instead; he also reports, however, that officers arrested only slightly more than half of the felony suspects against whom they had testimonial evi- dence. Most recently, Mastrofski et al. (1995), using a more comprehensive measure of evidence, found that the likelihood of arrest rose directly, and fairly steeply, with the strength of the evidence, a finding reproduced in a later observational study that examined only those cases in which a com- plainant requested that officers do something to control another citizen (Mastrofski et al., 2000). Officers' decisions to arrest are also strongly influenced by the prefer- ences of complainants, especially (but not only) when the offense is a less serious one and especially when the preference is for leniency. Complain- ants do not always articulate a clear preference for or against legal action, but when they do, police tend to comply. Smith and Visher (1981), for example, found that police made an arrest in almost half (46.6 percent) of the encounters in which the victim requested that an arrest be made, in only one-fifth (18.8 percent) of the cases in which the victim expressed no prefer- ence, and in less than one-tenth (6.6 percent) of the cases in which the victim requested that an arrest not be made. Black (1971) observes that this tendency "gives police work a radically democratic character" and also that the standard of justice that police apply is not uniform but rather varies with the moral standards of complainants. This is a pattern that has been observed in domestic incidents (Berk and Loseke, 1981), and it is one that recent pro-arrest statutory and policy changes have sought to alter, under the assumption that victims of abuse are not always in a position to request legal action against their abuser (see Ferraro, 1989; Jones and Belknap, 1999). Researchers have observed that the preference of the complainant is most influential when they request levels of police control lower than mak- ing an arrest: advice and persuasion, warnings and threats, and banishment from the scene (Mastrofski et al., 2000). Furthermore, the success of a complainant's arrest request was highly sensitive to the strength of evidence available; the likelihood that police officers would fulfill a request for an arrest was found to be much higher in situations in which evidence was strong compared to those in which it was weak. Complainants requesting lesser forms of control experienced high levels of police compliance, regard- less of evidence strength, although even here, stronger evidence produced a significantly higher chance of having the request fulfilled. One dispositional factor, juvenile status, does not appear to affect po- lice practice, in that patterns of decision making are based on the same criteria and weighed in the same ways (Worden and Myers, 2000). Riksheim and Chermak (1993) note that in the 1970s age was inversely related to the likelihood of arrest, but in the 1980s, controlling for other factors, suspects' age did not affect the likelihood of arrest (e.g., Smith and Visher, 1981; but

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 117 compare Mastrofski et al., 1995). This shift may reflect the last decade's well-documented trend in public attitudes and justice system practices to- ward treating juveniles more like adults, especially those suspected of seri- ous offenses (Triplett, 1996; National Research Council, 2001:Ch. 5). One 1970 observational study of drunk-driving arrests in a Midwest- ern city did produce results somewhat at odds with others reviewed above (Lundman, 1998). It found that, although some legal variables showed sig- nificant effects (whether the officers had to chase the suspect and the degree of intoxication), they were less powerful than several extralegal influences, such as the suspect's social class and demeanor. Before turning to the extralegal considerations in police practice, two caveats are in order. First, being influenced by the law is not the same as being governed by it. Studies of arrest show that as evidence of wrongdoing increases, so does the probability of arrest, but these studies do not judge how often the police ignore the specific standard of legal evidence that ap- plies to the case, such as probable cause. Indeed, because we know that police often overlook minor violations, even when the evidence is strong, we must be careful not to interpret these findings as suggesting that police serve as legal automatons. Second, the available research suggests that of- ficers tend to be constrained by law, but there are occasions when they clearly act outside it. However infrequent such incidents might be, they have a large impact on the perceived legitimacy of the police, in part be- cause when they become known to the press, they are highly publicized, an issue considered in Chapter 8. Extralegal Factors Although most research shows that many police actions constrained by law (e.g., arrest) are most strongly influenced by legal considerations, it is still possible for extralegal influences to exert a significant effect. Indeed the available research shows that police behavior is also influenced by extrale- gal factors, but, for the most part, findings have not been consistent as to the nature and strength of those effects. Citizens' Demeanor Toward the Police Proposition 2: It is widely believed that officers punish citizens based on the citizen's untoward demeanor toward the police, even when that demeanor is itself not a legal violation. The committee finds conflicting evidence regarding the impact of suspects' demeanor on police actions toward suspects and victim-complainants.

118 FAIRNESS AND EFFECTIVENESS IN POLICING The proposition that police officers respond punitively to suspects who fail to accord them deference emerged from some of the earliest systematic inquiry into police behavior. Westley (1953, 1970) found that the mainte- nance of respect is an important norm among police. Disrespect for the police, he reports, is symbolized by "the `wise guy,' the fellow who thinks he knows more than they do, the fellow who talks back, the fellow who insults the policeman" (1970:123; see also Van Maanen, 1978) and, fur- thermore, such disrespect legitimates the use of force to compel deference. Analyses of data collected in the 1960s and 1970s consistently found that the demeanor of suspects toward police affects the likelihood that they will be arrested and the likelihood that officers will use physical force against them (Black and Reiss, 1970; Black, 1971; Lundman, 1974, 1994, 1996, 1998; Sykes et al., 1974; Smith and Visher, 1981; Worden, 1989, 1995a; Worden and Shepard, 1996; Worden and Myers, 2000; compare Mastrofski et al., 1995; also see Van Maanen, 1978). Given the tendency of the police to underenforce the law (Wilson, 1968; also see Black, 1971; LaFave, 1965), this means that suspects who fail to show deference to police authority are less likely to get a break--to avoid justifiable arrest or to receive the benefit of an evidentiary doubt. Moreover, the magnitude of the estimated effect was substantial: one analysis of data collected in 1977 indicated that a disrespectful demeanor raised the estimated likelihood of arrest from .11 to .28 (Worden and Shepard, 1996), and the results of another analysis of the same data (Engel et al., 2000) indicated that police were 5.8 times more likely to use force against disrespectful suspects than against more deferential suspects. Analyses of more recent data, however, are mixed. Two studies using data on police intervention into disputes, collected in 1986, yielded mixed findings on the effect of demeanor on arrest (Klinger, 1994, 1996). An analysis of data collected in 1992 showed that the likelihood of arrest was greater when the suspect resisted police authority--if, for example, they refused to comply with an explicit police command, acted threateningly, or offered physical resistance (Mastrofski et al., 1995). The effect of resistance on arrest was limited to citizens' actions that were illegal; resistance that did not take the form of illegal action did not affect the likelihood of arrest. Data collected for the Project on Policing Neighborhoods in 1996-1997 shows that disrespect by suspects raises the probability of arrest (Worden and Myers, 2000), and it is by far the most powerful situational influence on whether the officer will act disrespectfully toward the suspect (Mastrofski et al., 2002a). Importantly, however, it has no detectable effect on officers' use of coercion more generally (Terrill and Mastrofski, 2002). However, another study, based on officer self-report data on custody arrests in six jurisdictions, found that an antagonistic demeanor, as well as physical resis- tance, substantially increased the likelihood of police use of physical force

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 119 (Garner et al., 2002). Thus it appears that the effect of demeanor on police behavior may be quite complex, perhaps contingent on other factors: the era (contemporary police may be less prone to apply the "attitude test"),2 the police department, and even the nature of the encounters in which po- lice and suspects interact. The demeanor of complainants and victims toward police has also dem- onstrated mixed effects on how police treat them. In one study, disrespect- ful demeanor from complainants was found to influence the likelihood that police will exert the degree of control on another citizen requested by the complainant, but this effect was evident in only one of the two cities studied (Mastrofski et al., 2000). Another study of a single department found that a complainant's display of disrespect toward the police did significantly re- duce the likelihood that police would try to control the targeted offender (Snipes, 2001). A disrespectful demeanor by the offender had no bearing on the police response in the two-city study (Mastrofski et al., 2000), but an uncooperative demeanor toward the police did produce a significantly re- duced likelihood that the police would try to control the offender in the single-city study (Snipes, 2001). When citizens asked the police for assis- tance that did not require controlling another person (e.g., help with a flat tire), the citizen's demeanor was found to exert no influence on the out- come of the request (Snipes, 2001). However, this study found that disre- spectful citizens requesting any form of assistance were generally less likely to be treated by police in a friendly or comforting manner. This contrasts with the finding of the two-city study, in which the likelihood that police comforted citizens experiencing some form of distress was shown to be unrelated to the citizen's demeanor toward the officer (Mastrofski et al., 1998). What can be taken from the studies on citizen demeanor that produce such a mixed pattern of findings? First, some of the diversity of the findings may be attributable to variations in how researchers have defined and mea- sured citizens' demeanor (Worden et al., 1996). Some consider it to be any- thing that police might interpret negatively; others emphasize failure to show deference (involving both verbal and physical acts); and others distinguish verbal acts of disrespect from acts of resistance (some defining it as physical only, and others including both physical and verbal resistance). In general, physical acts of resistance fairly consistently increase the risk of a punitive police response. Second, effects of citizen demeanor may vary according to the particular feature of police behavior under consideration: arrest, use of force, granting citizens requests, and affective displays toward the citizen. 2However, a recent survey of police (Weisburd et al., 2000) revealed that nearly half agreed with the statement "A police officer is more likely to arrest a person who displays what he or she considers to be a bad attitude."

120 FAIRNESS AND EFFECTIVENESS IN POLICING Despite the variability in how citizen demeanor is defined, there does seem to be a high level of consensus among multivariate studies of police use of force that the suspect's demeanor is a powerful influence on whether the police will resort to force (7 of 8 studies; see Garner et al., 2002). Most of the debate about the impact of a suspect's demeanor on arrest centers around how to control for and interpret the aspects of that demeanor that may be considered legally relevant to an arrest. This debate remains unre- solved. The resolution of this issue is not merely of academic importance, since there is no legal justification for punishing a citizen whose demeanor is unpleasant but not illegal. Citizens' Social Class Proposition 3: Some members of the public are concerned that the po- lice distribute coercion and assistance based on the citizen's social class. The evidence on the effects of social class on police behavior is scarce and the findings are mixed, precluding a judgment about its effects. Research on policing in the 1960s and early 1970s noted that officers were more likely to invoke their authority against lower-class suspects (see, e.g., Lundman, 1994, 1998; Reiss, 1968, 1971; compare Friedrich, 1980) and more generally to treat lower-class parties in a coercive fashion (Black, 1980:Ch.5). More recent studies have yielded mixed findings on this issue. Some show that the likelihood of arrest is unaffected by suspects' social class (Mastrofski et al., 1995), while others demonstrate that class has a substantial effect on the use of force (Terrill and Mastrofski, 2002). One possible explanation for this discrepancy is that force as defined in the Terrill and Mastrofski study included coercion as minimal as verbal commands and threats, whereas the first study calculated the effect of social class only on arrest. Arrest must be documented and therefore, as a behavior, it is subject to more scrutiny, possibly leading to a more scrupulous behavior on the part of the officer. One recent study that did find a significant, though small, social class effect examined the likelihood that officers would com- fort distressed citizens (Mastrofski et al., 1998); middle-class citizens were more likely to be comforted than lower-class citizens. One might expect that if officers are influenced by citizens' social class, they would be affected by the social class not only of a suspect but also of a complainant: while police might be more likely to invoke the law against a lower-class suspect, they might be less likely to take action on an allegation by a lower-class complainant (Black, 1976). Such complex causal dynamics require social scientists to estimate correspondingly complex causal models, and null findings could be artifacts of oversimplified modeling. But when analyses have taken such potential interactions into account, they have not

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 121 revealed class effects ostensibly hidden by simpler models (Mastrofski et al., 1995, 2000; Snipes 2001). Finally, the pattern of mixed results with social class effects may be due to inconsistencies in or problems with measuring social class. Studies vary in how social class is conceptualized and measured. More fundamental are problems with reliable measurement. Survey researchers may solicit from respondents information about their wealth, education, and occupation-- all of which can be challenging to determine from direct observation during a brief field encounter. Field observers must nearly always rely on infer- ences made from the circumstances, usually without detailed background information about the citizens observed. These inferences rely on the citizen's physical state, dress, speech patterns, home, vehicle, or other prop- erty at hand, as well as the neighborhood of the police-citizen encounter. Balancing this complex web of indicators, which sometimes produces con- flicting signals, increases the risks that different observers would judge the social class of the same citizen differently. Furthermore, if the observed officer's knowledge and assumptions about these appearances differ radi- cally from those of the observer, the likelihood of an erroneous inference increases greatly, assuming that the observer's task is to code how the citi- zen appeared to the officer. Research that tested the extent to which police officers' perceptions of social class match those of field researchers would help to establish the nature and scope of reliability problems with the mea- surement of social class. Citizens' Sex Proposition 4: Some are concerned that police officers are biased in favor of or against female suspects and victims. There is mixed evidence concerning whether and how citizen gender influences police behavior; the committee is unable to draw firm conclusions about the existence of widespread bias in police practices linked to gender bias. The "chivalry hypothesis" holds that female offenders receive preferen- tial treatment from the police (Visher, 1983). While the results of some early research were consistent with this expectation (i.e., women were less likely to be arrested than men), later research produced mixed results (e.g., Friedrich, 1980; Smith and Visher, 1981; see Riksheim and Chermak, 1993, for a review). Visher (1983) points out that preferential treatment may be extended by police only to women who fulfill sex role expectations. She found a complex pattern of interactions: for example, women were more likely than men to be arrested for property offenses, but less likely to be arrested for violent offenses. The more recent research finds support for the chivalry hypothesis (lower likelihood of arrest) even when only the main

122 FAIRNESS AND EFFECTIVENESS IN POLICING effect is estimated (Mastrofski et al., 1995; Worden, 1995b; Terrill and Mastrofski, 2002). The "leniency hypothesis" holds that police are less likely to invoke their authority on behalf of female victims (Fyfe et al., 1997). Many advo- cates for victims of domestic violence have supposed that the police are reluctant to take legal action against their (often male) assailants. They have advanced--and many states have adopted--statutory changes that promote arrest. They have also advanced--and many departments have adopted--pro-arrest policies. The underlying assumption is that police treat incidents of domestic assault differently from other assaults, and thus that the protection of the law is less likely to be extended to (predominantly female) victims. The evidence on this hypothesis is also equivocal. Some research has found, in analyzing police responses to incidents of domestic conflict, that the pattern of police decision making follows the same pat- terns that have been observed in police encounters with suspected offenders generally (Berk and Loseke, 1981; Worden and Pollitz, 1984), with the implication that police handle domestic conflicts in much the same way that they handle other incidents. Other research that has examined more di- rectly the proposition that police are less likely to invoke the law in inci- dents of domestic assault than in other cases of assault (Klinger, 1995; Fyfe et al., 1997) has generated mixed results. Looking at a broader range of disputes, one study found that when a citizen asked police to control another person, the gender of the requester had no significant effect on the probability that the police would exert the requested control (Mastrofski et al., 2000). However, the gender of the targeted citizen did make a difference in one of the two cities studied. There police were less likely to control males than females. However, a similar study conducted in another city found that females there typically received a less compliant response from officers than did males (Snipes, 2001). Another pair of analyses focused on the treatment of distressed citizens. One found that females were more likely to receive comfort or reassurance from police than males (Mastrofski et al., 1998), but the other found no significant difference between the sexes (Snipes, 2001). One possible source of the mixed results reported on the influence of the citizen's sex is that these studies varied in jurisdiction and time. Taking into account the legal and community context of these effects should be an important element of future research that attempts to test hypotheses about the impact of citizens' sex on police practice. Citizens' Race Proposition 5: There is a widespread perception of systematic police bias against racial and ethnic minority groups. The evidence is mixed,

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 123 ranging from findings that indicate bias against racial minorities, find- ings of bias in favor of racial minorities, and findings of no race effect. The results appear to be highly contingent on the measure of police practice, other influences that are taken into account, and the time and location context of the study. Studies of police behavior have routinely examined the degree to which officers' treatment of citizens varies with their race. Some of the earliest inquiries reported disparities in the treatment of white and black suspects, to the expected disadvantage of the latter, but these disparities were attrib- uted to causal factors other than race itself: to the more frequently disre- spectful demeanor of black or other minority suspects (Black, 1971) or to the more frequently pro-arrest preferences of black complainants (Black and Reiss, 1970). One scholar reconsidered his earlier findings in later work (Black, 1980:107-108). Some later, more technically sophisticated analyses (Smith and Visher, 1981; Worden, 1995a) showed that race has an effect on the arrest decision independent of other factors. Lundman (1998) found in a Midwestern city in the early 1970s that blacks stopped on suspicion of drunk driving had a significantly higher likelihood of being arrested than did whites. Some studies of the use of deadly force also provide support for this hypothesis (Meyer, 1980; Geller and Karales, 1981:123-125; Fyfe, 1982), as have some studies of less-than-lethal force (Smith, 1986; Worden, 1995a; Terrill and Mastrofski, 2002). Even so, research has not consis- tently supported the proposition that minorities are treated more harshly than whites. Null findings have been reported in many studies of arrest (e.g., Mastrofski et al., 1995), the use of nonlethal force (Friedrich, 1980; Kavanagh, 1994; Garner et al., 1995, 2002; Engel et al., 2000), and the use of lethal force (Milton et al., 1977; Fyfe, 1980, 1981b; Blumberg, 1982; Geller and Karales, 1981; Alpert, 1989). Some studies find that blacks are significantly more likely to be arrested at one time period and set of locales, but they fail to find that effect at other time periods and locales (as did Engel and Silver, 2001, in their analysis of nontraffic misdemeanor sus- pects). And a few studies even find that whites are in some communities targeted for more harsh police demeanor than blacks (Reiss, 1971; Mastrof- ski et al., 2002a). Smith, Visher, and Davidson (1984) conducted one of the more thor- ough analyses of race effects on police arrest decisions, using the PSS obser- vation data from the late 1970s. They report that the effect of citizens' race depends on other factors. When only suspects are present at the encounter scene, the suspect's race shows no effect for males, but black women are much more likely to be arrested than white women. When both suspects and victims are present, an arrest is more likely if the victim is white and the crime is a property offense. The police are also more likely to comply with the preference of a white victim for arrest. These "interaction effects" be-

124 FAIRNESS AND EFFECTIVENESS IN POLICING tween race and other features of the encounter must be interpreted cau- tiously, because when researchers test a large number of them, some may be significant only by chance. Indeed, these results were not replicated in sub- sequent research at later times and other places (Mastrofski et al., 1995, 1998). More recently, hypothesized racial biases have concerned officers' deci- sions to stop suspected violators or to conduct searches incident to stops. Charges of racial profiling by the police have occasioned scores of data collection efforts, although the knowledge that has been gained has not been commensurate with the effort because the results are ambiguous and difficult to interpret; many have not been guided by the logic of scientific inquiry (see Chapter 5 for more discussion of racial profiling). Social science research has focused on untangling the causes of police actions that are highly constrained by law (arrests, citations, and use of force), but by doing so, researchers have failed to produce results that are as useful as they might be to courts charged with reviewing police practices for potential racial bias. For example, much of the field research suggests that when other extralegal factors, such as the citizen's demeanor, are taken into account along with the citizen's race, no race effects are found. It may be that racial minorities have a more negative demeanor toward police (Ander- son, 1990, 1999), and it may be that demeanor, not minority racial status, increases the likelihood of arrest. Nonetheless, it must be noted that a citi- zen cannot be arrested on demeanor evidence alone. Under the constitution, the Supreme Court has ruled that an officer cannot arrest without probable cause that an individual has committed a crime or is about to commit a crime (Illinois v. Gates, 462 U.S. 213 (1983)).3 Officers who make a deci- sion based only on demeanor evidence, regardless of whether the officer ties this behavior to race or not, remain outside the bounds of the law. Even if demeanor increases the likelihood of arrest, the reasons for this consistently more disrespectful demeanor may in fact be anchored in a larger race-based story. For purposes of making a legal judgment about whether police arrest practices are racially biased, researchers should estimate the effects of citi- zens' race, controlling only for those factors that the law says should be taken into account. An analysis conducted by a member of the committee, built on an earlier analysis of POPN data (Mastrofski et al., 1998), found no race effects when a number of legal and other extralegal factors were taken into account. Similarly, additional analysis was also conducted on data in Richmond, Virginia, in the early 1990s (Mastrofski et al., 1995). 3The probable standard long predates Gates, which changed the standard by which applica- tions for warrants based on anonymous tips should be assessed.

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 125 When only race and two legally relevant variables (seriousness of offense and strength of evidence) were used to account for patterns of arrest, race still showed no significant effects in all three cities covered by these two data sets.4 Only a few studies have explored the effects of race on the order main- tenance and assistance aspects of police practice. One study examined how police responded to citizens' requests to control other citizens present at the scene (Mastrofski et al., 2000). In St. Petersburg, when a white citizen re- quested control of a minority citizen, the probability that police would com- ply was substantially greater than that with other racial pairings, but this did not achieve statistical significance at the conventional .05 level used by social scientists. A replication of this analysis in Richmond, Virginia, showed no race effect (Snipes, 2001). Three studies have examined the effects of race on the officer's demeanor toward citizens (whether the officer is friendly or offers comfort or reassurance), and none found a race effect (DeJong, 2000; Mastrofski et al., 1998; Snipes, 2001). An analysis of the police re- sponse to citizens' requests for assistance not involving the control of an- other citizen also found no race effect (Snipes, 2001). In general, research on police behavior suggests that race effects are highly contingent. Some studies find effects--and those effects differ in di- rection--and others find no effects. Differences in how variables are mea- sured, what other influences are taken into account, sampling, and the set- ting (both time and place) make it impossible to discern meaningful patterns at present. In the committee's view, the most fruitful research strategy for the future will be to systematically take into account the organizational and environmental context in which police-citizen interactions transpire. This, of course, requires some theory or logic about the aspects of the larger context in which they occur. One possibility is to consider the political power wielded by minority groups in the community and their representa- tion in the police force itself, as well as the history of race relations (Mastrofski et al., 1998). One of the few studies showing black citizens with more positive evaluations of the police than whites was conducted in a city where local government was dominated by black elected officials and the majority of the police force was staffed by blacks (Frank et al., 1996). Given the regular recurrence of allegations of racial injustice by the police and the inconclusive nature of the available findings, the committee judges it a high research priority to establish the nature and extent to which 4In one of the cities, St. Petersburg, the effects of race were significant at a probability of .09, which is above the .05 standard characteristically used by social scientists. In St. Petersburg, the odds of a minority suspect's being arrested, controlling for the seriousness of the offense and the strength of evidence, was about half again as great as the odds for white suspects.

126 FAIRNESS AND EFFECTIVENESS IN POLICING race and ethnicity affect police practice, independent of other legal and extralegal considerations. Because of the contingent nature of race effects demonstrated in extant research, public policy would most benefit by a well-organized program of research that provides for the systematic testing of race effects across the most relevant contingencies. Such a program should consider a variety of contexts under which race effects might vary. Varia- tion over time and place in legal sanctions for racial bias (e.g., changes in state and federal civil rights laws) should be evaluated. Researchers should carefully scrutinize the effects of variation in the community context in which policing occurs, such as the political power of racial and ethnic mi- norities in local government and their representation in police agencies. A wide array of police practices should be considered, including both those that are punitive and controlling and those that render some service. It will be especially important to consider the cumulative impact of police decision making. Most of the extant research considers each type of decision as if it were unconnected to prior choices made by police and others who mobilize them. For example, the racial distribution of arrests is dependent on prior decisions about whether and when to mobilize police to engage suspects. An analysis of police disrespect toward suspects showed that, although mi- nority citizens were not more likely than whites to be treated disrespect- fully, they were far more likely (than their representation in the neighbor- hood residential population) to have contact with the police, making minority citizens about twice as likely as whites to experience police disre- spect (Mastrofski et al., 2002a:543). Research that incorporates the contin- gent nature of police decision making will better untangle the sources of racial disparity that may be uncovered. Mental Illness Proposition 6: Police have been accused of arresting mentally disor- dered citizens without legal justification. Although some initial studies supported this claim, later studies found that, other things being equal, mentally disordered suspects are no more likely or even less likely than other suspects to be arrested. Given the small number of rigorous stud- ies on this topic, the committee recommends further research. Urban police frequently encounter mentally disordered persons (Bittner, 1967; Teplin, 1984, 1986, 1994; Teplin et al., 1996), and the frequency of these contacts has increased considerably since the 1960s because of the deinstitutionalization of the mentally ill, more demanding standards for civil commitment, and the limited availability of community treatment pro- grams (Engel and Silver, 2001:225-227). Recently urban police have been encouraged to initiate more interventions in dealing with the sorts of public

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 127 disorders in which mentally disordered people may engage (Reiss, 1985; Wilson and Kelling, 1982; Kelling and Coles, 1996). Police may arrest and jail mentally disordered persons because (a) they have violated the law; (b) they wish to punish or control them, even though they may not have vio- lated the law; or (c) no suitable alternative is readily available to care for the person, a complaint frequently voiced by officers. There is no dispute that mentally disordered persons are overrepresented among the incarcerated population of the nation, but the key question is whether that reflects a misapplication of the police authority to arrest, one that "criminalizes" mental illness (Engel and Silver, 2001:228). Perhaps the most influential study offering support for the criminaliza- tion hypothesis--based on systematic field observations of police interac- tions with 506 Chicago suspects--found that mentally disordered suspects were arrested at 1.67 times the rate of those not judged to be mentally disordered, and this relationship held at different levels of incident serious- ness (Teplin, 1984). Subsequently, some researchers have questioned the validity of the claim that the police had criminalized mental illness, noting that the Chicago study had not controlled for a variety of legal factors shown by prior research to exert considerable influence on police arrest practice (see earlier section on legal factors) (Engel and Silver, 2001). They conducted a separate analysis of systematic field observations drawn from the Police Services Study and the Project on Policing Neighborhoods. Their analyses drew on much larger samples of suspects drawn from 26 urban police agencies of varying sizes. Using multivariate regression, the research- ers controlled for a large number of legal and extralegal variables (including evidence of the suspect's prior illegal behavior and the suspects' behavior toward police and others during the encounter that constitutes legal viola- tions). From the POPN data they found that mentally disordered suspects were almost three times less likely to be arrested as those not judged to be mentally disordered. From the PSS data they found that mentally disor- dered suspects were not significantly different from other suspects in their chances of being arrested. The researchers suggested two possible reasons that their findings differed from the Chicago studies: the inability of the Chicago study to control for legally justifiable reasons to arrest suspects and different methods used to measure mental disorder. The Chicago study used clinical criteria, while the POPN and PSS studies used observers' per- ceptions, thought to more accurately reflect police officers' perceptions. If observers (or police) consistently failed to classify as mentally ill suspects who deserved that classification and who were arrested, this would under- state the effects of the criminalization process. Whether officers have the knowledge and skill to make these determinations is not well established. Officers typically receive little training on handling mental health problems (Police Executive Research Forum, 1989), but one study found that officers

128 FAIRNESS AND EFFECTIVENESS IN POLICING are able to make reasonable judgments about the risk of violence and make judgments about mental health fairly consistent with the clinical diagnoses of mental health professionals (Menzies, 1987; Steadman, 1986). Another reason for the divergent findings is simply that police practices may change over time and differ from jurisdiction to jurisdiction--or even from neigh- borhood to neighborhood--factors not taken into account. Finally, the small number of mentally disordered suspects available for study in all of the field observations sums to only 133, hardly constituting a basis for a stable estimation of effects. Given the limitations of the available research, the Committee is unable to draw a conclusion about the independent influence of a suspect's mental illness on the police arrest decision. For policy purposes, a more meaningful analysis would go beyond the mere determination of whether there was sufficient legal justification for an arrest. Future research should take into account certain contextual factors that may influence an officer's choices, especially the availability of adequate treatment alternatives (e.g., a medical or mental health facility willing to accept such cases). Without such alterna- tives, officers faced with severely disordered persons must usually choose between the least-worst alternatives of doing nothing, encouraging friends and family to exert control, or making an arrest. OFFICERS' OUTLOOKS AND CHARACTERISTICS Over the last 150 years, reformers in America have focused much effort on shaping the practices of police by shaping the characteristics and out- looks of police officers (Fogelson, 1977; Walker, 1977). A long line of re- formers has made this the focal point of their efforts, including Theodore Roosevelt (Berman, 1987), August Vollmer (Douthit, 1975), O.W. Wilson (1963), and Herman Goldstein (1990). These and other reformers have concerned themselves with hiring, developing, and retaining officers who have high moral character (e.g., absence of a serious criminal record or abuse of drugs), who have an acceptable personality (e.g., that avoid certain psychological deficiencies), who embrace certain ideals (e.g., a commitment to public service, to observing legal constraints on police powers, to crime fighting, to working in partnership with the community), who have certain predispositions and not others (e.g., regarding the treatment of minority groups), whose personal characteristics represent certain underrepresented groups in policing (e.g., according to race, ethnicity, and gender), who have higher levels of education and training, and who have (or have the capacity to develop) a set of skills valuable to police work (communication, persua- sion, mediation, physical adeptness). Police policies designed to shape the characteristics of who does police work typically take form in the following areas: recruitment and hiring policies; policies regarding performance re-

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 129 view, promotion, and career advancement; disciplinary policies; job descrip- tions and personnel assignment practices; training policies; and approaches to and priorities of supervision and management. These policies are expected to affect officers' behavior through two types of intervening mechanisms: officer attitudes and beliefs ("outlooks") and officer knowledge, skills, or ability. Policies may influence police prac- tice through other mechanisms, such as organizationally provided incen- tives and disincentives to engage in certain behaviors. The following chap- ter is reserved for discussing the organization-level influences on police practice. In this part of the chapter we first consider a variety of proposi- tions about officers' outlooks (both general and specific) before turning to the effects of knowledge, skill, and ability and closely associated character- istics, such as training, education, race, and gender. General Police Outlooks Researchers have been conducting surveys of police officers for over five decades, and there is no consensus about whether outlooks are impor- tant to examine and, if so, how they can be best measured. Studies have examined a number of general aspects of police personality, including authoritarianism, cynicism, and job satisfaction. The influence of these broad personality measures on police practice has not been extensively or rigorously tested. Relevant research on these general constructs is reviewed below. Authoritarian Personalities Proposition 7: A popular stereotype about police holds that officers embrace attitudes that are more authoritarian than the average person and that these attitudes influence police behavior in undesirable ways. The available evidence on the existence of an "authoritarian personal- ity" among the police that distinguishes them from the general public is inconclusive, and there is very little research on the extent to which variation in authoritarianism among police accounts for differences in police behavior. The committee judges research on this issue to be in- sufficient to offer an assessment. Some research in the 1960s and 1970s was based on the premise that police use and misuse of their authority stems from their authoritarian per- sonalities, which predispose them to use force (Balch, 1972, and more gen- erally, Frankel-Brunswik, Levinson, and Adorno, 1993). Authoritarianism is a cluster of beliefs that includes a strong adherence to conventional, middle-class values, a tendency to think in terms of rigid categories, identi-

130 FAIRNESS AND EFFECTIVENESS IN POLICING fication with powerful figures, and a concern with displays of strength and toughness. Research on the police has little to say about the extent to which these personality traits vary among officers. Rather, it has been concerned mostly with whether the degree of authoritarianism among police as an occupational group differs from that of citizens. This research has turned on the question of a modal (and pathological) "police personality" that has all the earmarks of an authoritarian personality. Results have been incon- clusive (Balch, 1972; compare Lefkowitz, 1975), in terms of whether police are more authoritarian than the population as a whole. But this research did not examine whether these personality attributes might account for variations in behavioral propensities among police. Elements of authoritarianism might contribute to the issue of some "problem officers." Hans Toch has described the "violence-prone" officer, whose reactions to catalytic events are different from those of other officers and whose reactions might be attributed to personality characteristics. In a survey of police psychologists, Scrivner (1994) found some evidence of such an effect: respondents reported that among the officers referred to them because of their use of excessive force, one group had personality disorders that placed them at chronic risk. "These officers have pervasive and endur- ing personality traits (in contrast to characteristics acquired on the job) that are manifested in antisocial, narcissistic, paranoid, or abusive tendencies. These conditions interfere with judgment and interactions with others, par- ticularly when officers perceive challenges or threats to their authority. Such officers generally lack empathy for others.... Individuals who exhibit these personality patterns generally do not learn from experience or accept re- sponsibility for their behavior, so they are at greater risk for repeated citi- zen complaints"(1994:3). It remains to be seen whether such personality traits are disproportion- ately characteristic of officers who use force improperly or excessively, and as Scrivner pointed out, "the number who fit this profile is the smallest of all the high-risk groups." Police Culture Proposition 8: Many patterns in officer behavior are attributed to the existence of a police culture that pressures police to share a similar view about their work, involving such things as being suspicious, being iso- lated from outsiders, and extreme loyalty to fellow officers. There is insufficient evidence to establish whether these or any other "cultural" perspectives distinguish police officers from the citizenry generally or from other occupations. There is no rigorous evidence that measures the influence of the police culture on actual police practice.

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 131 While the authoritarian police personality faded into intellectual obscu- rity, the police culture remains a widely accepted analytic construct, despite having some similar shortcomings. The police culture is a social psychologi- cal construct, rather than a psychological one, although it rests on similar presumptions about the degree of homogeneity among police officers. The police culture is a set of widely shared outlooks that are formed as adapta- tions to a working environment characterized by uncertainty, danger, and coercive authority and that serves to manage the strains that originate in this work environment (see Bittner, 1974; Brown, 1981; McNamara, 1967; Reuss-Ianni, 1983; Skolnick, 1966; Sparrow et al., 1990; Van Maanen, 1974; Westley, 1970). Officers characterize how they cope with the danger and uncertainty of their interactions with the public as being "suspicious" and "maintaining the edge." Officers read people and situations in terms of the potential danger that they pose, and they maintain the edge in an authoritative, take- charge approach to citizen interactions (Sykes and Brent, 1980; Manning, 1994). Officers characterize how they cope with bureaucratic risks as adopt- ing a "lie-low" attitude and a "crime-fighter" or "law enforcement" orien- tation. They minimize their risk of exposure and sanction. They also de- scribe law enforcement as "real" police work compared with the more ambiguous order maintenance or service roles, and the inner-directed, ag- gressive street cop is seen as the cultural ideal. The problems that officers confront in their working environment, as well as the coping mechanisms prescribed by the police culture, produce other defining characteristics of the police culture: social isolation and group loyalty. Officers' coercive authority distances them from society, and they further distance themselves in coping with the danger of their jobs. Some analysts hold that the professionalization of the police (i.e., taking the politics out of policing, focusing on scientific crime-fighting, and using motorized patrol, along with anticorruption measures) has fur- ther isolated police and strengthened the police culture (Brown, 1981; Spar- row et al., 1990). In this context, officers develop an "us versus them" attitude toward citizens (Sparrow et al., 1990) and strong norms of group loyalty. New recruits are tested before being accepted as one of the group, and officers are expected to provide mutual support vis-à-vis a hostile citi- zenry and a punitive bureaucracy. To the extent that it exists and holds sway, police culture is important because, first, it is thought to represent a major obstacle to holding officers accountable, and second, it is an impediment to organizational change. The culture buffers officers from organizational sanctions and also serves to promulgate views about how police work should be performed. Yet several studies, which were based on fieldwork conducted in the 1970s, cast some doubt on the extent to which officers subscribed to cul-

132 FAIRNESS AND EFFECTIVENESS IN POLICING tural views. These studies (White, 1972; Muir, 1977; Broderick, 1977; Brown, 1981) formed typologies of police officers based on two or three attitudinal dimensions. While research on the police culture has stressed what appeared to be the central tendencies in the occupational group, the typologies highlighted the variance. Although these studies were based on field research conducted in different places and at different times, and al- though they use different attitudinal dimensions to define the types, their descriptions of the types suggest that five composite types can be identified (Worden, 1995b). Their findings contrast with conventional views of the police culture, particularly with respect to the predominance of a role orien- tation defined by law enforcement, the idealization of an aggressive ap- proach to policing, and an us versus them attitude toward citizens. More recent developments in policing presumably have further attenu- ated the cohesiveness of the police culture. The increasing social diversity of the police, with the recruitment of women and racial minorities (Walker, 1985), and the rising education level of police might be expected to weaken the social bonds among officers (see Haar, 1997). The adoption of commu- nity policing as an operational philosophy might further challenge tradi- tional cultural tenets. One study of a department that was trying to imbue a "community policing culture (proactive community building) in its patrol officers in the early 1990s found that the department had achieved modest success in getting its officers to share its vision of community policing, but there was tremendous variation in the styles of patrol that officers exhib- ited, suggesting that the police culture of that department was more frag- mented than the "monoculture" literature would have predicted (Mastrofski et al., 2002b). Wood and colleagues (2003) also found that increasing gen- der, racial, and educational diversity had fragmented what was reputedly a monolithic police culture, and that various elements of these new police cultures either supported or opposed what was ultimately not a very suc- cessful attempt to institute community policing. While these studies cast some doubt on the claim that the police culture is an overwhelming influ- ence on police practice, they do not preclude the possibility that it nonethe- less exerts a powerful and significant influence. Despite its popularity as a reputed cause of officer behavior, there have been no rigorous tests of the impact of a modal police culture containing all or most of the elements described above, whether it is a culture that pro- motes traditional or community policing perspectives. This flows in large part from the problem that most of the literature on the police culture views it as more or less uniform over time and place. Without variation in police culture, it is difficult to demonstrate its impact on police practice. At a minimum, such a study would require comparison of different police cul- tures (varying across place or over time). To the extent that police cultures co-vary with patterns of police practice, and to the extent that alternative

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 133 causes of variation in those practices can be discounted, police culture will have been shown to have an impact. However, the measurement of police culture remains an obstacle to such a test. The theoretical framework for establishing the locus or font of the police culture has not been articulated. Police culture is a characteristic of an occupation, or at least a police orga- nization, not an individual officer. To measure a culture's effects, it must exist independent of the outlooks of the individual officers it presumably affects, because we must be able to entertain the possibility that some, per- haps many, officers will not embrace the dominant cultural outlook, even though they may feel its pressure. But it is not clear where researchers should look to measure the police culture. Researchers might look to the formal or informal leadership of the entire occupation or a given department (e.g., high-visibility leaders, professional associations, unions). They might also consider the presumed structural sources of culture (e.g., size and degree of bureaucratization) (Herbert, 1998). Cynicism Proposition 9: Some researchers have hypothesized that the onset of cynicism among police officers leads them to use their authority abu- sively. Very little empirical research is available to test this claim, mak- ing it impossible to offer a judgment as to its validity. Cynicism among police is the subject of numerous empirical inquiries, much of it based on the application of Arthur Niederhoffer's (1967) cyni- cism questionnaire or some modification of it. Indeed, few research instru- ments have enjoyed more widespread use in criminal justice research than has the Niederhoffer questionnaire. But in a review of this research, Robert Langworthy concludes that "we are left with both an elegant theory of the development of police officer attitudes and ample evidence that the measure Niederhoffer developed to test the theory is inadequate" (1987:33). Much of the research is implicitly predicated on a misconception of police cyni- cism, and, as a consequence, this research has produced inconsistent results. Cynicism entails a loss of faith in people, an unwillingness to attribute to them any but self-interested motives, a derisive skepticism of the efficacy of social institutions, and a posture of indifference. Niederhoffer's 20-item questionnaire, which he administered to samples of New York City police- men, has been administered (in its original or modified forms) to other samples in other police departments, and it has been subjected to more intensive scrutiny. These analyses show that the number and nature of the attitudinal dimensions tapped by Niederhoffer's questionnaire vary from department to department (Lefkowitz, 1975; Weiner, 1974; Regoli, 1975, 1976; Poole and Regoli, 1979).

134 FAIRNESS AND EFFECTIVENESS IN POLICING One treatment of police cynicism, that of William Muir (1977), sug- gests that officers vary in their degree of cynicism, and also that it shapes how they use (and misuse) their authority. Cynicism is one dimension that serves as the basis of Muir's four-fold typology of police officers (the other dimension being the officer's comfort in using force). Muir's study was based on 28 officers in one department. Other scholars who have devised typologies of police have also described cynicism, even though it is not an explicit component of their typologies. More importantly, perhaps, this re- search forms the basis for--but does not systematically test--the hypothesis that officers' cynicism drives their behavior in important ways. An effort to replicate Muir's study to determine the strength of the relationship between his typology and patterns of behavior showed virtually no relationship be- tween an officer's place in the typology (based on interview comments) and his observed behavior on the job (Snipes and Mastrofski, 1990). Job Satisfaction Proposition 10: Much of the police management literature argues that maintaining adequate levels of job satisfaction is essential for sustain- ing good or satisfactory police performance. No research is available to validate this claim. Police research has borrowed concepts and methods from industrial psychology to examine the job satisfaction of police officers. Job satisfac- tion is a multidimensional construct: an officer may be satisfied with some elements of her job while she is dissatisfied with other elements. Police of- ficers tend to be satisfied with the intrinsic elements of their jobs: the variety of tasks and the challenge and significance of the work (see, e.g., Zhao et al., 1999). They tend to be less satisfied with supervision and management. One might hypothesize that officers who are more satisfied with their jobs also tend to perform better--that is, that happy workers are productive workers--yet there is no research that could speak to this issue. It is espe- cially challenging to determine how much job satisfaction influences work performance and how much work performance influences job satisfaction. This requires measuring both job satisfaction and work performance over time to link job satisfaction at an earlier time with work performance at a later one. Research is also needed on the factors that affect job satisfaction, especially those that can be promoted by police administrators. Further- more, researchers should monitor the sources of satisfaction and dissatis- faction associated with new and experimental policing programs. Knowing what irks and pleases officers will prove valuable in designing more effec- tive methods of implementing innovations in the future.

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 135 Effects of Specific Attitudes on Behavior Proposition 11: Attitudes toward specific aspects of police officers' work and work environment are expected to affect how officers per- form that work. However, a small body of quantitative research at- tempting to validate this linkage has with few exceptions found only weak relationships between specific officer attitudes and behavior. The proposition that police officers' patterns of behavior correspond to their outlooks is certainly plausible, given the latitude that officers have in performing police work on the street. One might expect that officers' occu- pational attitudes would manifest themselves especially in the more discre- tionary forms of behavior, those that are not regulated by law or standard operating procedure, or that are difficult to subject to supervisory oversight or other review. Decisions to stop or other actions that officers initiate on their own authority (e.g., public relations contacts), might particularly be affected by their attitudes. However, the few efforts to test such propositions systematically have produced, in the main, little or no support. One of the earliest and most striking findings from systematic observational research on police was that, while many officers professed to be prejudiced against blacks, they did not act on that prejudice in their encounters with citizens (Reiss, 1971). Freidrich confirmed that the use of force is unrelated to either officers' job satisfaction or (among white officers) their attitudes toward blacks (1980), and he also found that these outlooks bore weak relationships to other forms of behavior (1977). Worden (1989, 1995a) found that officers' occu- pational attitudes--for example, their conceptions of the police role, atti- tudes toward citizens, attitudes toward legal restrictions--are only weakly related to their patterns of behavior on the street in general and to their use of force in particular. Mastrofski et al. (1994) found that officers' attitudes, including their individual enforcement priorities, bore weak relationships to their patterns of enforcement of drunk-driving laws (also see Heeren et al., 1989). Terrill and Mastrofski (2002) found that officers' use of coer- cion was unrelated to their attitudes. Only two studies have reported rela- tionships of a noteworthy magnitude. Mastrofski et al. (1995) found that officers who had more positive attitudes toward community policing were less likely to make arrests than their more negative colleagues. Brehm and Gates (1993) reported that officers' rates of "shirking" or time spent "goof- ing off" or "loafing" were influenced by their professionalism, their atti- tudes toward their jobs, and their satisfaction with their supervisors.5 The 5"Professionalism" was measured in terms of summary characterizations by observers of how officers conducted themselves across a work shift; it is probably better interpreted as behavioral than attitudinal in nature.

136 FAIRNESS AND EFFECTIVENESS IN POLICING results of this research are hardly definitive, and propositions that officers' outlooks or personality traits affect their behavior remain plausible but are not established scientific facts. This small body of quantitative research is consistent with a much larger body of social psychological research on attitude-behavior consistency, which has shown that the estimated relationships between attitudes and behavior are counterintuitively small (Weinstein, 1972; Ajzen and Fishbein, 1977). Innumerable studies have analyzed various forms of behavior and their relationships to attitudes. In their review of this research, Schuman and Johnson (1976:167-168) reported that "few plausible studies fail to find significant relationships," however, "in most cases investigated, atti- tudes and behaviors are related to an extent that ranges from small to mod- erate in degree." This body of research also offers some clues about why the associations between attitudes and behavior are not stronger: it is attenu- ated by "situational cross-pressures," including the norms of reference groups. In policing, such cross-pressures might include the expectations of coworkers and the limitations imposed by bureaucratic controls. Another source of the weak attitude-behavior linkage is invalid and an unreliable measurement of both attitudes and behaviors. Often, the attitudes mea- sured are not the ones most theoretically relevant to the behavior in ques- tion. Also, attitudes are frequently measured using responses to a single question, rather than the more reliable scales derived from combining re- sponses to many questions that tap into the same attitude dimension. It is fair to say that some of the research on police attitudes is subject to these methodological limitations. Finally, the whole point of creating police organizations is to attenuate the link between the views of individual officers and how they practice policing. Chiefs might prefer that their officers not hold racially prejudicial views, but what counts is whether officers act on those views. A department that structures police work and the consequences of police work to inhibit acting on prejudicial attitudes has accomplished its goal precisely by cutting the link between attitude and behavior. Thus, the failure to find substantial links between personal attitudes and behaviors is not necessarily bad news for control of police discretion, if that means that officers are behaving consistent with the organization's goals instead of their own beliefs. One study that explored this possibility in making drunk-driving arrests found that officers' personal beliefs about the importance of drunk-driving en- forcement had far less influence on their arrest productivity than their capa- bility (experience and training) and opportunity (work shift) to engage in this kind of enforcement activity (Mastrofski et al., 1994).

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 137 KNOWLEDGE, SKILLS, AND ABILITY Proposition 12: It is widely believed that officers with greater knowl- edge, skills, and abilities will perform better than those with less of those traits. Research indicates that knowledge, skills, and ability do tend to predict better scores on performance tests, but these tests are of unknown value in predicting actual behavior and performance on the job. Consequently, the committee is unable to offer propositions about what forms of knowledge, skill, and ability are most useful for various aspects of actual job performance. Police officer knowledge, skills, and abilities appear to have a moder- ately strong influence on the measures of their performance that researchers have used, but those measures bear a weak or unknown relationship to actual police practice. We are therefore unable to draw conclusions about the influence of knowledge skills and abilities on officer behavior. It is widely assumed in American society that those who have greater knowl- edge and more developed skills will generally perform better than those who do not. Reformers have made this a fundamental precept of professionalizing police, although police officers may be more inclined to adhere to the notion that knowledge derived from classroom instruction is of limited value, while learning by doing (or watching one's fellow officers) is how the most valuable knowledge and skills are obtained (Rubinstein, 1973; Van Maanen, 1978; Bayley and Bittner, 1984). Furthermore, police and outside experts may differ in what knowledge, skills, and abilities they believe are most valuable for police or police in a specific job assignment. Police knowledge, skills, and abilities are typically measured by test scores--for example, reflecting the officer's grasp of search and seizure law or the ability to pass a physical fitness test. Most of what is known about the relationship of knowledge, skills, and abilities and job performance comes from research on the validity of various psychological tests. A recent extensive review provides some insight into this literature (Hogg and Wil- son, 1995:42-52).6 Ability tests fall into four groups: cognitive aptitude (the capacity to learn the job with sufficient training), basic achievement (the extent to which an individual has knowledge or skills after training), specific ability (current performance ability for specific tasks), and intelligence tests (general capac- ity for problem solving, abstract thought, and understanding complexity). These abilities tend to be highly correlated, although different types of tests work better at predicting different sorts of things. An important caveat is 6The literature review in this section draws heavily on this comprehensive review.

138 FAIRNESS AND EFFECTIVENESS IN POLICING that most of the evaluations of these tests rely on measures of officer perfor- mance that are very general in nature and far removed from the sorts of specific police actions toward which most of the commentary in this chap- ter has been directed. For example, "performance" might be defined as not being fired or not resigning, getting positive supervisor evaluations, or suc- cess in police training (higher test scores). These measures reflect outcomes, subjective rater judgments, or proxy measures of behavior (test scores) and must therefore be regarded with considerable caution for the purpose of predicting specific types of behavior, such as making an arrest, using force, or engaging in some form of community policing. While many people believe that, other things being equal, more intelli- gent police officers will perform better, some have postulated that people of both low and high intelligence will not perform policing as competently as those in the middle range (Kenney and Watson, 1990). The explanation is that more intelligent officers may be understimulated by the routine aspects of police work. There is some research to support this curvilinear relation- ship as well as other research that produces mixed results (Hogg and Wil- son, 1995:44). Results are undoubtedly dependent on the type of job under consideration, and one might arguably expect a direct relationship between intelligence and job performance for officers assigned to a problem-solving specialty. However, level of educational attainment showed no such effect in predicting the amount of time officers spent on problem solving in one field study (DeJong et al., 2001:54). In general, research tends to show that ability measures are predictors of police training performance, but it is less clear that training performance is itself a good predictor of performance in the field (Hogg and Wilson, 1995:45-46). It is assumed that testing during recruit academy training may successfully weed out those with the least ability, but this has not been demonstrated because of the lack of follow-up on those who were weeded out. One meta-analysis of 40 law enforcement studies showed that cogni- tive ability tests were fairly strong predictors of academy success, but only about one-third as effective in predicting actual job performance (Hirsh et al., 1986). Police agencies use a variety of specific task assessment methods to evaluate ability: in-basket exercises, teamwork tests, and job simulations in assessment centers (Hogg and Wilson, 1995:51-52). Assessment center ex- ercises have been shown to do a better job of predicting job performance than ability tests, but they are much more costly to run. Structured inter- views are less valid in predicting job performance than ability tests, but they have achieved respectable levels of validity, especially when based on a job analysis, use multiple interviewers, are situation- and job-focused, and are standardized. All of this suggests that knowledge, skills, and abilities do influence performance, and that ability tests in policing (as well as most

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 139 occupations) are better at predicting job performance than are personality tests (which concentrate on attitudes, perceptions, and predispositions). However, because most of the measures of performance used in these stud- ies have a weak or unknown relationship to actual officer practice, we are unable to draw firm conclusions about the degree of influence that knowl- edge, skills, and abilities exert on officer practice. EXPERIENCES OF OFFICERS Thus far this chapter has considered what is known about the impact of outlooks and capabilities on police officer practice. Police organizations and government generally attempt to manipulate outlooks and develop ca- pabilities by promoting such things as education and training and recruiting officers with certain characteristics. In recent years, equal opportunity and affirmative action policies have promoted the recruitment of a more racially and sexually diverse police force nationwide. By focusing on these consider- ations, police agencies are in effect attempting to manipulate the experi- ences that officers bring to bear on their work. They do this by selecting officers who have certain characteristics (and presumably certain types of experiences before becoming officers) and by socializing and training those selected to serve as police. In this section we consider what is known about the influence of these factors on police practice. Education Proposition 13: It is widely believed that higher education improves officer behavior, and, as a consequence, much effort has been invested over the years in raising education levels. The evidence reviewed by the committee does not permit conclusions regarding the impact of educa- tion on officer decision making. Prior research has not employed strong measures of police behavior and performance; it has not taken into account the content of the education; and it has not controlled for the effects of many other influences on behavior and performance. There is widespread enthusiasm for college education as a valuable de- veloper of knowledge, skill, and ethics for policing generally and commu- nity policing in particular (Shernock, 1992; Baro and Burlingame, 1999; Smith and Flanagan, 2000). The proportion of departments serving large cities that require new officers to have at least some college climbed from 19 to 37 percent between 1990 and 2000; the proportion requiring at least a 2-year degree increased from 6 to 14 percent during that period (Bureau of Justice Statistics, 2002). Much of the impetus for this trend can be cred- ited to the desire to increase the regard in which others hold the police, but

140 FAIRNESS AND EFFECTIVENESS IN POLICING many advocates also expect that a college education improves the capacity of police to analyze and communicate, as well as to develop greater sensitiv- ity to democratic values they are sworn to uphold in their work. Commu- nity policing, especially those approaches that emphasize problem solving and working closely with citizen organizations, is thought to call for the greater knowledge and skill that college is supposed to develop (Goldstein, 1990). The available evidence is limited in both quantity and quality, and the results are, at best, mixed--failing to show the sorts of dramatic differ- ences reformers had hoped to see. Much of the research on the effects of education focuses on attitudes or aspects of performance that are only indirectly related to community polic- ing (Smith and Ostrom, 1974; Cascio, 1977; Weirman, 1978; Sherman and Blumberg, 1981; Worden, 1989). Several reviews of the literature have con- cluded that evidence on the effects of higher education on police perfor- mance is weak and mixed (Hayeslip, 1989; Buckley, 1991). A number of early studies indicated that college-educated officers are less dogmatic, au- thoritarian, cynical, prejudiced, intolerant, and punitive, but these studies measured attitudes, not performance (Shernock, 1992:73). Some studies found college education to be associated with fewer civilian complaints, quicker response times, and higher levels of enforcement activity (arrests, citations, stops, and checking businesses) (Finckenauer, 1975; Cascio, 1977; Hudzik, 1978; Sherman, 1980; Carter and Sapp, 1990). One study found that officers with a four-year college degree were substantially less likely to say they would arrest a mentally disordered person than were officers with- out the degree (LaGrange, 2000). An attempt to measure actual officer per- formance rather than attitude was based on department records. Control- ling for age, length of service, and father's occupation, the researchers found that education had no significant relationship to the use of lethal force (Sherman and Blumberg, 1981). This and virtually all other studies have serious methodological flaws: weak measures of performance, inadequate measurement of the quality of college education, limited capacity to control for the effects of other variables (many of which are presumably correlated with a college degree), and lack of consensus on what constitutes good performance (Mastrofski, 1990:17; Shernock, 1992:74). A few systematic observation studies have focused on actual police prac- tices. Worden (1995a) identified three types of police behavior in handling suspects: no force, reasonable force, and improper force. He found that officers with a bachelor's degree were significantly more likely to use rea- sonable levels of force, but that they were indistinguishable from officers without a college degree in the use of improper force. Data from the POPN study found that college education had no significant effect on the amount of time officers spent on problem solving (DeJong et al., 2001), and no effect on officers' responsiveness to citizens' request to control other citi-

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 141 zens in encounters (Mastrofski et al., 2000). However, college education was found to have a negative effect on providing comfort to citizens (DeJong, 2000). In the latter case, the negative effect of a college degree was concentrated among male officers. Similar results demonstrating no effect of a college education were found in a study in Cincinnati that used the same observation protocols as the POPN study (Novak et al., 1999:171). Given the methodological limitations of past research on the effects of higher education on police practice, it is not surprising that the results are mixed and the effects generally weak. Even when relationships are found, researchers have not been able to distinguish how much of the effects of higher education are due to changes imparted by the educational experience and how much are due to the selection and screening involved in getting accepted into a college program (such as intelligence, initiative, wealth, fam- ily background, and knowledge, skills, and abilities acquired before col- lege). Other limitations of existing research on the effects of college educa- tion are numerous. Almost all studies measure the quantity of higher education that officers receive but ignore the content of that education. Does the student's curriculum make a difference (e.g., majoring in criminal justice versus other majors, a theory-focused curriculum versus an applied, experiential one)? Controlling for age, is there a difference in the behavior of a person with a college education acquired before becoming a police officer com- pared with the behavior of a person who obtained a college education while on the job? To what extent does the trend toward more college education produce a police force with less understanding and empathy for society's disadvantaged persons, and what are the consequences for street-level po- lice practice? The committee finds the available evidence inadequate to make recommendations regarding the desirability of higher education for improv- ing police practice and strongly recommends rigorous research on the ef- fects of higher education on job performance. Training Proposition 14: For many decades it has been assumed that more and better police training leads to improved officer performance. The com- mittee finds that past research on this topic has not addressed the com- plexities of the subject. Few studies evaluate the impact of training programs on actual performance on the job. Prior research has not taken into account the substantive content of training programs, modes of instruction, the abilities of the instructors, the timing of training, or the organizational support for reinforcing the objectives of the train- ing program.

142 FAIRNESS AND EFFECTIVENESS IN POLICING There is limited evidence available, scarcely more than a handful of studies, on the effects of training. Some research suggests that the influence of training depends on the extent to which training lessons are reinforced in the larger organizational environment, through top leadership, supervision, performance monitoring, and incentive systems. Most of the research on training suffers from one or more serious methodological limitations: poor measures of the actual performance that training is designed to improve, an inability to distinguish training effects from those of other influences, and failure to take the characteristics of the organizational environment into account. Research on the effects of training content, timing, instructor quali- fications, pedagogical methods, dosage, and long-term effects is virtually nonexistent. Police training has been a growth industry since 1967 (Mastrofski, 1990:14). Between 1990 and 2000, median classroom training hours re- quired of recruits grew from 760 to 880, and field training requirements similarly grew from 520 to 600 hours during the past decade (Bureau of Justice Statistics, 2002). In addition, states require that minimum levels of in-service training be completed annually. Finally, when something goes wrong in a police department, additional training is usually at the top of the list of remedies proposed and undertaken. Knowledge of the effects of police training is limited primarily to whether more training produces the desired change in police practice. This is typically acquired by conducting a controlled experiment (comparing police who have received training with those who have not) or by a study that measures the correlation between the amount of training officers have received and some police practice, while statistically controlling for the ef- fects of other influences, such as years of experience. There are too few of either type of study available to shed light on the effects of training. Often evaluations of police training are embedded in assessments of larger organizational change interventions. An assessment of San Diego's Community Profile Development project in the early 1970s is an example (Boydstun and Sherry, 1975). This program attempted to implement many of the features that would later be incorporated into Goldstein's (1990) notion of problem-oriented policing. The program was intended to increase officers' knowledge and empathy for the people they served, and to get officers to focus on initiating innovative patrol strategies to deal with tar- geted problems in their assigned areas after carefully observing and analyz- ing those problems. It was evaluated using an experimental design that ran- domly assigned officers to treatment and control groups, taking pre- and post-treatment measures of a wide range of police behaviors, practices, atti- tudes, and perceptions. This program included not only training (60 hours), but also supervision and management direction given in support of these objectives. The evaluators found that the program had several encouraging

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 143 outcomes: officers in the treatment group had a more developed sense of beat responsibility, more knowledge about their beats, and a greater in- volvement with community leaders. However, the evaluators were unable to distinguish the specific contribution that training made to produce these effects. A randomized experimental evaluation that focused exclusively on the effects of training was conducted in Detroit in the 1980s (Rosenbaum, 1987). The 3-day training program, delivered to police officers undergoing a 16-week basic training curriculum, was designed to sensitize officers to the needs of victims. Surveys administered immediately before and after the training showed that the officers receiving the training showed much more favorable attitudes, perceptions, and intentions regarding victims than did officers in the control group. However, victims served by the treatment and control groups reported no differences in the way that officers treated them, suggesting that training may have affected officers' views and inclinations, but that it had no effect once they were actually at work. These two studies illustrate a dilemma for understanding what effects training has on police practice and how to improve its effectiveness. The San Diego study evaluated an entire organizational change program, of which training was a significant part, but it was not the only organizational change strategy employed. It found generally positive results, but the re- searchers were unable to identify the specific contribution that training made, or whether it made any contribution at all. The Detroit study was able to isolate the effects of training, but it was unable to determine why its effects were limited. Future experimental evaluations should attempt to do both by increasing the number of training programs that are evaluated and varying the nature of the training and the context in which it occurs. More numerous than experimental evaluations are studies that attempt to establish the relative contribution of police training and compare it with the effects of other possible influences, such as supervision and various in- centives for performance. Typically, these studies examine the amount of training officers have received at a given point in time and then correlate it with some measure of officer performance (e.g., number of arrests), while controlling for the effects of other possible influences. The results of this research are mixed. The amount of training in community policing had no influence on the amount of time officers spent on problem solving (DeJong et al., 2001) and no effect on whether officers comforted citizens (Mastrof- ski, 1998). However, officers who had high levels of community policing training and who had a broad conception of their role (treating minor dis- orders as a priority) were significantly more likely to respond positively to citizens' requests to control other citizens who were troubling them (Mas- trofski et al., 2000).

144 FAIRNESS AND EFFECTIVENESS IN POLICING The contingent nature of training's effects appears to depend not only on what kind of officer receives it, but also on the kind of organizational environment in which the officer is operating. For example, training to deal with drunk-driving violations (driving under the influence or DUI) was found to be related to the number of arrests patrol officers made and that the effects of this training depended on the organizational environment in which the officers worked (Mastrofksi et al., 1996; Mastrofski and Ritti, 1996). When training was supported by ongoing supervisory and manage- ment practices that favored DUI arrest (leading by example, having officers work closely with victims' groups, closely monitoring arrest statistics, and supporting and protecting from internal criticism officers whose DUI arrest rates were exceptionally high). In departments that failed to provide a nur- turing environment to sustain the DUI training, the amount of training re- ceived had no effect on the officers' DUI arrest rate. A study of community policing training produced similar conclusions to the DUI study, in that academy training that was not reinforced in the field usually failed to produce lasting changes in officers' attitudes and be- liefs (Haarr, 2001). This study of Arizona police recruits found that at the end of a 16-week program of training, recruits tended to show a more positive outlook toward community policing, problem solving, and tradi- tional policing than when they began the training. However, these effects dissipated during the 12-week field training experience (except for views on the need for good police-public relations), and by the end of their 1-year probationary period, they tended to hold more negative attitudes toward community policing than they did at the end of academy training (except for police-public relations). Overall, with the exceptions of views on the need for good police-public relations and self-assessed capabilities in problemsolving, recruits entered the academy with more positive views than they held after one year on the job. During field training, the most powerful predictors of attitude were department policies supporting community po- licing. The change in attitude from pre-academy to the end of the one-year probationary period was strongly influenced by the officers' perceived view of coworkers' attitude toward community policing and the assigned work shift (presumably representing opportunities to spend time doing commu- nity policing on less busy shifts). Although this study focused on attitude changes and not performance, it underscores the notion that to sustain training's effects, it must be reinforced. A survey of training for community policing in over 500 police agencies indicated that few police agencies were even going so far as to require that their field training officers have knowl- edge of community policing (McEwen, 1997). There are numerous limitations to the correlational research on train- ing, including measurement of training itself. Most of this research is lim- ited to measuring the amount of training but does not consider that the

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 145 character, quality, and timing of the training can vary, even within the same department.7 For example, some observers have criticized much police train- ing for attempting to change officers' belief systems, while paying little or no attention to giving them the skills and incentives to change their prac- tices, a common approach in the areas of domestic violence, cultural sensi- tivity, and community policing (Buerger, 1998; Haar, 2001). According to critics, such training is doomed to fail because it attempts to overcome strongly embedded habits and norms using a relatively light dose (at most, a few days of training), offers no really useful tools that officers can use, and fails to reinforce the training on the job. Nonetheless, much contempo- rary police training does attempt to imbue officers with a different set of outlooks about their work, so it would behoove researchers to take the content of the training into account. Another important unresolved issue about training content is determining what sort of curriculum is most effec- tive in promoting the practice of various aspects of community policing. Minimum training standards are established by each state's police officer standards and training council, but there is virtually no rigorous research to guide them on how to structure recruit training curricula most effectively (for example, whether to integrate community policing training seamlessly throughout the curriculum or whether to highlight it in special segments).8 Similarly, evidence is lacking on what sort of curriculum will best promote effective problem-solving projects. Should it be academic in nature (teach- ing recruits the rudiments of social science evaluation research), or should it be more inductive and experiential? Correlational research has also failed to look for different patterns of training effects according to the type of officer who receives it. For ex- ample, it is conceivable that training on a given topic (such as handling domestic disputes) will be most effective when introduced in discrete seg- ments over time, rather than all at once. Officers may need an opportunity to learn basic skills and try them out before moving on to more advanced techniques. The notion that training should be designed to build on past skill acquisition is different from the more common approach of simply retraining officers with the same material periodically. Until studies get more detailed information on when training of a given type was received in an officer's career, little will be known about the most effective way to develop a long-term training program for police. What is known about the effects of training on police performance and 7Officers are typically asked to indicate how many hours of training on a given topic they have received in the last x time period. 8One analysis suggests that even after more than a decade in which community policing became very popular among police, training for officers has not changed much from its tradi- tional focus on reactive activities (Bradford and Pynes, 1999).

146 FAIRNESS AND EFFECTIVENESS IN POLICING practice is very limited, making the need for a systematic and rigorous re- search program in this area quite compelling. For example, recommenda- tions about what constitutes desirable training to control police abuse of force do exist, but there is virtually no empirical validation of these claims (Geller and Toch, 1995:318). A number of issues that in the committee's view require attention: · What should the content of training be? Should it focus on changing attitudes, developing skills, or both? How generalized can training be, and to what extent must it be tailored to the needs of the locale (Fyfe, 1995:171)? · What training methods work best? Is realistic training that attempts to simulate conditions on the job the most effective (Fyfe, 1995:167)? For what kinds of training topics is "roll-call" training effective, and for what types of topics is more training immersion required? · Who make the most effective instructors, those who are experts in training or those who are experts in the content area of interest?9 Does the sworn versus civilian status of the instructor affect the willingness of offic- ers to accept and implement the training? · At what point in their career should officers receive training of a given type, what kind of follow-up training is effective, and when and how should it be given? · What is the appropriate duration, intensity, or dosage of a given type of training? · For a given type of training, what kind of on-the-job reinforcement is required to produce the desired change in officer behavior? How impor- tant are departmental rewards and sanctions for performing according to training compared with an individual officer's sense that the skills learned in training are useful? Is training an effective way to initiate change in an organization, or to be effective must it follow other organizational changes in such areas as supervision, incentive and disciplinary systems, and perfor- mance accountability? · How long do the effects of training last? That is, how quickly do any effects decay over time? The received wisdom is that police academy train- ing is quickly undercut by what officers learn from their more experienced colleagues on the job (Bayley and Bittner, 1984; Haar, 2001). Is this neces- sarily so for all training? What kinds of recruit training, if any, do officers find useful and follow as they gain experience? · What kinds of control groups are appropriate? 9Experts may vary considerably in their ability to teach, but they will generally be far more knowledgeable about their topic, motivated to "sell" the training, and have higher credibility with the trainees (Buerger, 1998).

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 147 Given that departments invest extraordinary resources in training every year, the committee strongly recommends more research on police training. Perhaps the ultimate question for future research is to determine the limits of the effects of training. That is, just how much can training influence officers' attitudes, beliefs, knowledge, and skills? Much can undoubtedly be learned by studies of the effects of training in the military, the clergy, medi- cine, and the law. Ultimately, however, training competes with many other forces that come into play, so it is especially important to gain a realistic sense of what sorts of transformations can be expected. The committee anticipates that research on this issue will benefit from a careful consider- ation of two things: what skills and values police trainees bring to their experiences and the organizational context to which they return when they have been trained. Specifically, it is important for training evaluations to go beyond a narrow focus on the training program itself; they need to incorpo- rate the entire package of management decisions made for monitoring, su- pervising, and rewarding desired behaviors. EQUAL EMPLOYMENT OPPORTUNITY, AFFIRMATIVE ACTION, AND THE EFFECTS OF OFFICER RACE AND SEX As discussed in Chapter 1, the trend in federal and state laws over the last 35 years has been to force police agencies to open their doors wider to racial and ethnic minorities and women. Although the principal legal justi- fication for these laws has been to end and rectify employment discrimina- tion, reformers have also argued that minority and female officers (a) will perform better (at least in certain situations or with certain groups of people) than white and male officers and (b) that their presence on the force will help to change the predispositions of the police subculture (Walker, 1998:232). Equal employment opportunity and affirmative action have served as principles that appear to have increased the representation of ra- cial minorities and females on America's police forces (Walker, 1985; Mar- tin, 1990), but does the race and gender of the officer have a significant effect on the way that officers exercise discretion? The short answer to this question is that the limited research available provides little support for the notion that race and gender have a significant influence on officer behavior. Some recent research shows that female po- lice officers are more inclined to engage in community policing and caregiving behavior, but the pattern is mixed and the number of studies limited. Indeed, the received wisdom from the research community is that whatever influence race and gender may exert on behavior is overwhelmed by the unifying effects of occupational socialization (see Donohue and Levitt, 2001, for a review). This may be disappointing to some reformers, who expected improved performance, but it may also be interpreted as good

148 FAIRNESS AND EFFECTIVENESS IN POLICING news, inasmuch as it discredits past discriminatory practices that equal em- ployment policies attempted to rectify, and it shows that no appreciable differences in policing practice by officer race or sex should be anticipated. Officer's Race Proposition 15: Many reformers have argued that increasing the num- ber of racial and ethnic minority group officers will lead to improved policing and better police-community relations. This proposition is based on the assumption that, for example, black officers will be less likely to shoot, arrest, or stop black citizens than white officers. The committee finds that in the small body of relevant studies there is no credible evidence that officers of different racial or ethnic backgrounds perform differently during interactions with citizens simply because of race or ethnicity. Reformers of American police have for some time couched their criti- cisms and claims as if the race of the police officer has a significant influ- ence on how the officer behaves (Kerner Commission, 1968:315). Many have argued that police forces will be more caring and service-oriented when the racial makeup of the police force approximates that of the department's jurisdiction, and for most center-city urban areas, that has meant increasing the number of minority officers on the force. Underlying this notion is that people with the same racial background will be more solicitous of each other. A contrary hypothesis is offered by Black (1976, 1980:9), who ar- gues that citizens of high or dominant status are more likely to receive favorable police response when the lawgiver (i.e., police officer) is of a lower or nondominant status. Conversely, when the citizen is of a lower status than the officer, the probability of a favorable police action is lowest. Like- status individuals fall between these two poles, according to Black. Virtu- ally all of the available studies compare whites and blacks. Some research has found that officers of different races do tend to have different occupational outlooks (Alex, 1969; Rossi, 1974; Jacobs and Cohen, 1978; Decker and Smith, 1981; Leinen, 1984; Paoline et al., 2000; Weisburd et al., 2000) and knowledge about their neighborhoods (Mastrofski, 1983), but these differences do not seem to translate into sig- nificantly different patterns of behavior. Virtually all multivariate analyses that have tested for the effects of an officer's race on the use of coercion (arrest or force) show no appreciable difference between races (Reiss, 1968; Fyfe, 1981a; Smith and Klein, 1983; Worden, 1989, 1995a; Mastrofski et al., 1998; Engel, 2000; Terrill, 2001). A study of police disrespect toward suspects found that, in one of two cities studied, white officers were more

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 149 inclined to be disrespectful to white suspects than minority suspects, an effect that the researchers suggested might have been due to the minority chief's strenuous efforts to reduce both police incivility and racial discrimi- nation (Mastrofski et al., 2002a). As with the punitive aspects of police work, little evidence in support of officer race effects has been found when researchers have attempted to pre- dict various forms of order maintenance, police assistance, and engagement in community policing. Novak et al. (1999:176) found no effects of officer's race on the probability that officers would initiate order maintenance activ- ity with suspects and disputants. DeJong (2000) found that the officer's race had no significant effect on the likelihood that a citizen would be com- forted. Mastrofski (1998) found the same result when looking at officer- citizen race dyads. Two analyses of the effects of officer-citizen racial pair- ings found no effect on the likelihood that officers would grant citizens' requests to control another citizen (Mastrofski et al., 2000; Snipes, 2001). Only one of the reviewed studies found an officer race effect (Engel et al., 2000). White officers spent more time on problem-solving activities than did black officers. In contrast to the above studies, all of which are based on field observa- tions of individual officers' encounters with the public, is an analysis of arrest rates in 122 U.S. cities with populations greater than 100,000 for a time period spanning 1977-1993 (Donohue and Levitt, 2001). This study found that increases in the number of minority officers were associated with increases in arrests of whites (but not minority citizens), while more white police produced increases in the arrest rate of nonwhites (but not white citizens). These effects were particularly strong for minor offenses, such as public order offenses, prostitution, drunk driving, and other minor crimes. Extrapolating their results, the researchers estimated that "moving from random assignment of officers by race to a scenario in which same- race policing is maximized would lead arrests to decrease by over 15 per- cent" (Donohue and Levitt, 2001:390). It is important to note that this research does not allow us to make valid predictions about the effects of race on individual officers' behaviors, because the unit of analysis in this study is at an aggregated level--the entire municipal department. Arrest rates by officers' race may be affected by a number of factors not taken into account in this study (beat and shift assignment patterns), as well as other policies and practices that are associated with both the racial distribution of the police force and the distribution of arrests across citizens of different races. Moreover, arrest rates are calculated on the basis of people in a given population rather than on the number of incidents that could have resulted in an arrest. Inasmuch as quantitative analyses of individual officers' arrest

150 FAIRNESS AND EFFECTIVENESS IN POLICING practices have not produced similar cross-race effects as found in this aggre- gate study, it should be interpreted with great caution.10 One possible race effect is that the distribution of officers by race in a department affects the way that individual officers (by race) behave (Walker, 1985; Mastrofski et al., 1998). The vast majority of officers in most Ameri- can police agencies are white, but possibly in those few departments in which the majority of officers are black, both black and white officers may behave differently, because there are a sufficient number of minority offic- ers to sustain an alternative culture to that which has been found repeatedly in white-dominated departments. Underlying this notion is that a police force that is well integrated provides a context for developing greater mu- tual respect and understanding, as officers come to know each other as individuals. That is, the process may also change police officers' stereotypi- cal views about people of a race other than their own. It also seems reason- able to hypothesize that when there is substantial racial heterogeneity in a police force, and that force also experiences substantial race-based tensions, officers' race may have an effect on how they practice policing, although the nature of the effects could well differ from situations in which there is racial heterogeneity and no such tension. Although nearly all of the available multivariate research suggests that an officer's race is not a significant influence on police behavior, this issue should be explored more fully by considering different contexts in which the officer's race might matter. Researchers could test more fully the possi- bility that the effects of the officer's race depend on that of the citizens with whom they interact. More importantly, researchers should compare officer race effects in departments in which officers of a minority race constitute the majority of the sworn force, thus considering the possibility that these effects differ from departments in which minority race officers also com- prise a minority of the sworn force. 10Indeed, a field observation study of Richmond, Virginia, in the early 1990s found that white officers dealing with minority citizens were the most likely to receive a compliant re- sponse when ordering a citizen not to engage in undesired behavior, and minority officers dealing with white citizens were the least likely (Mastrofski et al., 1996). Like-race pairings of officer and citizen were indistinguishable from each other in their success at securing citizen compliance and fell between the two racially heterogeneous pairings. Assuming that citizen compliance mitigates the need for arrest, these findings would predict that minority officers would have fewer enforcement alternatives to arrest when dealing with white citizens, while white officers dealing with minority citizens would have less need for arrest, given their higher rate of compliance success with that racial group.

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 151 Officer's Sex Proposition 16: Many argue that the employment of more female offic- ers will lead to changes in policing. This assumption is based on the belief that women are less aggressive and more nurturing than men and therefore more likely to use less coercion, relying instead on persuasion and assistance in dealing with citizens who cause problems or need help. The committee finds that the body of available research is too small and the findings too variable to draw firm conclusions about the effects of officer sex on police practice. Relatively little work has been done on the differences in how men and women practice policing. A number of studies fail to find differences be- tween male and female officers' beliefs and perceptions (Worden, 1993; Lasley, 1994; Finn and Stalans, 1997; Stewart and Maddren, 1997), and early evaluations of female officers indicated no sex-based differences in officer performance (see Feinman, 1994, for a review), focusing on the "masculine" or enforcement-oriented aspects of police work. Some of the early empirical work on police behavior suggested that female officers are less aggressive, less inclined to make arrests and citations, and less inclined to misbehave (see Sherman, 1980; Mastrofski, 1990; Riksheim and Cher- mak, 1993, for reviews). But this research suffers a number of methodologi- cal limitations, such as inadequate control for such confounding factors as age, experience, and duty assignment. Recent studies that control for many of these potentially confounding effects also fail to show significant differ- ences between male and female officers in making arrests, issuing citations, and using force (Worden, 1989, 1995a; Engel, 2000; Terrill, 2001). Some work has focused on specific community policing and order main- tenance practices. Two of six systematic observation studies of patrol indi- cated that women were more inclined to engage in assistance or community policing; one study indicated that females were less inclined to do so; and three showed no difference. Engel et al. (2000) found that female officers spent more time on problem solving than male officers. Snipes (2001) found no difference between male and female officers in the amount of time they spent on encounters in which a citizen requested some form of assistance, and DeJong (2000) found that females were significantly more likely to comfort female citizens than males were to comfort male citizens. Cross- gender pairings of officer and citizen were not significantly different from the female officer-female citizen pairing. But Mastrofski et al. (2000) found that female officers were less likely to grant citizens' requests to control others who were causing trouble, regardless of the degree of control re- quested (ranging from advice to arrest). Snipes (2001) found a similar rela- tionship in another city, but the sample of cases was much smaller and the

152 FAIRNESS AND EFFECTIVENESS IN POLICING results not significant. Finally, Novak (1999) found no difference between men and women in their proclivity to initiate order maintenance activities on their own. Thus, two of the studies do suggest that female officers are more likely to engage in the caregiving sorts of community building and problem-solving aspects of community policing, while the evidence is less clear on the kinds of behaviors that constitute a projection of the more intrusive and coercive aspects of police authority. Future research exploring the possibility that officers' gender influences practice would benefit from a number of improvements. First, more refined measures of police practice should be considered. For example, whether or not an officer makes an arrest may not be linked to an officer's gender, but how that officer treats the people involved (whether or not an arrest was made) might be influenced by the officer's gender. Whether female officers tend to listen more to both sides of the story in a dispute is a question worth answering. Second, an examination of how officers spend their time that is free from assignments from the dispatcher and supervisors might reveal sex-linked differences. That is, there are few formal constraints determining when and where officers choose to mobilize, so any sex-linked inclinations would be most likely to be revealed in these patterns of behavior. IMPLICATIONS The committee explored research relevant to a number of propositions about the proximate influences of police behavior; both characteristics of the situation and of the officers. Virtually all of the literature reviewed fo- cused on patrol officers. We found that the evidence available to test most of these propositions was inadequate to draw firm conclusions. Some impli- cations are nonetheless possible. That police practices, especially those tied to the enforcement function, are influenced far more by legal than extralegal considerations is encourag- ing news for those wishing to assess the state of policing in America. This is not necessarily cause for celebration, however, since there are a sufficient number of studies finding that race, sex, and social class influence police practice to be cause for concern. And even if these effects are much smaller than those of legal considerations, this says nothing about how much toler- ance a society should have for these influences. Indeed the mixed evidence calls for more rigorous research to determine the circumstances under which the personal characteristics of the citizen do affect police practice. The situ- ation with regard to the effects of citizen demeanor is different, inasmuch as a great deal of research has found this to be a relatively strong influence on whether a suspect is arrested. There is still debate about how much the extralegal aspects of citizen demeanor influence police enforcement prac- tices, once the legally relevant aspects of that demeanor (constituting viola-

EXPLAINING POLICE BEHAVIOR: PEOPLE AND SITUATIONS 153 tions of the law) are taken into account, an issue that future research should resolve. More important for policy purposes, however, is determining what interventions, such as training, supervision, and discipline, prove most ef- fective in reducing the scope of the extralegal aspects of citizen demeanor. Researchers still have plenty to learn about the effects of legal and ex- tralegal situational influences on police practices, but future research will make especially valuable contributions to improving policing if it can deter- mine the sources of variation in these effects. For example, when legal con- siderations are not as influential as desired, is this due to a lack of police knowledge about legal requirements? Difficulties in applying the law to specific situations? A negative attitude about the law? The impact of com- peting priorities, such as the need to husband resources or deliver substan- tive justice? Are extralegal factors, such as revealed in race effects, less likely in departments with more active systems for detecting, correcting, or pun- ishing racially biased officer practices? Despite the considerable effort police leaders have devoted to winning the hearts and minds of police officers, the available evidence is not encour- aging about the prospects of changing officer behavior by changing their outlook. Research on general outlooks (authoritarian personality, police culture, cynicism, and job satisfaction) has for the most part not tested the effects of these constructs, due in part to formidable measurement difficul- ties. The small body of research that has tested the influence of specific attitudes has shown at most only weak linkages between attitude and be- havior, a finding that is consistent with findings in the field of social psy- chology generally. One implication is that attitudes and attitudinal change are poor proxies for actual practice when evaluating the impact of policy interventions on actual police practices. A second implication is that poli- cies and management practices designed to shape officers' philosophies about their work appear to be unfruitful. However, an encouraging feature of this pattern of results is that police agencies may be fairly effective in breaking the link between individual beliefs and preferences on one hand and practice on the other. What remains to be shown is how successful organizations are--and through what mechanisms--in getting officers to pattern their practices consistent with the goals of the organization. While the improvement of knowledge, skills, and abilities of police of- ficers seems an unassailable objective, available research leaves largely un- tested the degree of influence these things exert on actual police practice, so policy makers and the public remain uninformed on the actual return re- ceived from these investments. This is also the case for two specific strate- gies that have been the mainstays of professional reform: increasing the quality and quantity of education and training for police. The small number of studies and the methodological limitations of most studies mean that particular programs to enhance police training and education are developed

154 FAIRNESS AND EFFECTIVENESS IN POLICING and offered without scientific evidence of their likely effects. Given the im- portance of these tools to those striving to improve policing, the committee cannot overstate the importance of developing a comprehensive and scien- tifically rigorous program to learn what is and is not effective in the educa- tion and training of police officers. Such evaluations should measure out- comes in terms of actual policing practices rather than tests and other proxies. The absence of effects on police behavior related to the officer's race suggests that it may be irrelevant to actual practice on the street, although it leaves untested the impact of a more racially diverse police workforce on the legitimacy of the police (a topic for Chapter 8). Because the available evidence on the effects of an officer's gender is inadequate to draw conclusions, it is difficult to draw specific implications in this area. Certainly there is a need for research that looks with greater care for areas of police practice in which differences between the sexes are most likely. Such studies will prove of limited practical value, however, unless they are able to determine the source of those differences. To what extent are they based on physiology, sex-role expectations, work environ- ment differences, and so on? Ultimately, the search for the causes and control of police behavior must extend beyond the limited domains of the situations and individual officers who handle them. Policing is shaped by the organizational and com- munity contexts in which these events occur. These are the focus of the following chapter.

Next: 5 Explaining Police Behavior: Organizations and Context »
Fairness and Effectiveness in Policing: The Evidence Get This Book
×
Buy Paperback | $54.95 Buy Ebook | $43.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Because police are the most visible face of government power for most citizens, they are expected to deal effectively with crime and disorder and to be impartial. Producing justice through the fair, and restrained use of their authority. The standards by which the public judges police success have become more exacting and challenging.

Fairness and Effectiveness in Policing explores police work in the new century. It replaces myths with research findings and provides recommendations for updated policy and practices to guide it. The book provides answers to the most basic questions: What do police do? It reviews how police work is organized, explores the expanding responsibilities of police, examines the increasing diversity among police employees, and discusses the complex interactions between officers and citizens. It also addresses such topics as community policing, use of force, racial profiling, and evaluates the success of common police techniques, such as focusing on crime “hot spots.” It goes on to look at the issue of legitimacy—how the public gets information about police work, and how police are viewed by different groups, and how police can gain community trust.

Fairness and Effectiveness in Policing will be important to anyone concerned about police work: policy makers, administrators, educators, police supervisors and officers, journalists, and interested citizens.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!