Minority threat and police brutality: Determinants of civil rights criminal complaints in U.S. municipalities

The conflict theory of law stipulates that strategies of crime control regulate threats to the interests of dominant groups. Aggregate-level research on policing has generally supported this proposition, showing that measures of minority threat are related to legal mechanisms of crime control. Police brutality (i.e., use of excessive physical force) constitutes an extra-legal mechanism of control that has yet to be examined in this theoretical framework. This study extends research in the area theoretically and substantively by testing the hypothesis that the greater the number of threatening acts and people, the greater the number of police brutality civil rights criminal complaints filed with the U.S. Department of Justice. The findings show that measures of the presence of threatening people (percent black, percent Hispanic (in the Southwest], and majority/minority income inequality) were related positively to average annual civil rights criminal complaints.

Perhaps no political tenet is held more dearly in the United States, at least among dominant group members, than the belief in equal justice. Yet some agents of the institutions of justice inevitably challenge that premise by discriminating against racial and ethnic minorities. It is hardly surprising that many minority citizens distrust the criminal justice system, just as many criminal justice agents distrust them. Nowhere is that tension more apparent than in the relations between minorities and the police. Scholars have long expressed concern about police reactions to ethnic minorities (e.g., Blauner, 1972; Feagin, 1991; Irwin, 1985; Myrdal, 1944.; Sellin,1930; Westley,1953,1970), and it is becoming increasingly clear that the roots of police-minority hostility are deeply embedded into the social structure (Jackson, 1989).

The conflict theory of law maintains that crime control is an instrument used by powerful groups to regulate threats to their interests, thereby maintaining the existing social structure (e.g., Turk, 1969). In this view, the police function to control the "dangerous classes" of immigrants, racial minorities, and the poor. Structural-level studies of police-minority relations in this tradition have addressed the issue of whether aggregate measures of minority threat (e.g., percent nonwhite) predict the use of mechanisms of crime control by the police. That research generally has supported the hypothesis that minority threat is related to crime control (e.g., Jackson, 1985; Jackson and Carroll, 1981; Jacobs, 1979; Jacobs and O'Brien, 1998; Liska and Chamlin, 1984; Liska and Yu, 1992; Liska et al., 1985; Sorensen et al., 1993). To date, however, structural-level research has focused on systematic variations in police behavior that may negatively impact minority popul! ations but still fit within the legal limits of discretion. This investigation extends that research tradition by examining the incidence of police brutality, which involves excessive physical force that falls outside the limits of legality (Locke, 1996).

Previous research has considered a variety of abusive police practices. The findings generally indicate that minorities are targeted disproportionately by the police (e.g., Bogomolny, 1976; NAACP, 1995; Pate and Fridell, 1993; Piliavin and Briar, 1964; Reiss,1971; Smith, 1986; U.S. Commission on Civil Rights, 1970; Westley, 1953, 1970; Worden, 1996). Although that research has made important contributions, some substantive and theoretical concerns have yet to be addressed adequately. Substantively, little reliable data exist on the relationship of race or ethnicity and police brutality (i.e., excessive force), which is the most serious of the various abusive police practices and the most germane to conflict theory's emphasis on coercive state power. Theoretically, the crucial question from the conflict perspective remains unanswered: Is the incidence of police brutality related to aggregate measures of minority threat?

BACKGROUND

The legitimate use of force is at the core of the police role (Bittner, 1970). For example, the use of deadly force by the police, which disproportionately involves minority victims (e.g., Binder and Scharf, 1982; Fyfe, 1981), is nearly always deemed legally justifiable (e.g., Fyfe,1980). At the same time, democratic societies sharply circumscribe police authority. The use of force, thus, may be either proper or excessive, depending on whether it is necessary or justified to accomplish a legitimate police duty (Kania and Mackey, 1977). From the perspective of law, excessive physical force most clearly constitutes police brutality, a term often applied loosely to various forms of police misconduct (Locke, 1996). Many citizens define police brutality broadly to include abusive police practices, such as the use of profane and abusive language and unnecessary searches, that do not entail excessive physical force (Locke, 1996; NAACP, 1995; Reiss, 1968). Minorities in particular ! perceive any degrading, restricting, or harassing practice as objectionable (Reiss, 1968), and a number of studies provide evidence that the police disproportionately employ these practices against them. For example, the use of abusive language, including racial slurs (Anderson, 1990; Skolnick, 1975; Westley,1953, 1970), the incidence of field interrogations (Bogomolny, 1976), and the likelihood of being frisked or searched (Black and Reiss, 1967; Piliavin and Briar, 1964; U.S. Commission on Civil Rights, 1970) have been related to race. Excessive physical force, however, is at the heart of minority concern about police misconduct (Locke,1996; NAACP,1995; Reiss,1968). Yet, findings from the few research studies of excessive force show consistently that its use is rare (Adams, 1996), and social science research remains unclear on the salience of race in the employment of excessive force (Locke, 1996).

Efforts to explain police brutality have focused on (1) sociological or situational, (2) psychological or individual, and (3) organizational factors (Freidrich, 1980; Worden, 1996). The first approach maintains that situational exigencies, such as the race, gender, and demeanor of suspects, determine the use of excessive force. Emphasis is placed on the social dynamics of police-citizen encounters and the situational cues that officers use to decide how to handle an incident. The second approach identifies the characteristics of officers, such as their racial identity, degree of prejudice, and personality factors, that predict the use of excessive force. In this view, individual variations among officers produce different responses to similar situations. The third approach maintains that organizational properties of police departments, including administrative controls and police subculture, determine the degree to which excessive force is employed. The focus is on formal a! nd informal aspects of police organization that influence officers' street-level behavior.

Elements of the various explanations are found in Westley's (1953, 1970) seminal work on police violence. Drawing on observations and interviews collected in 1949-1950, he examined a department located in a medium-size industrial city with a large slum area and a large black population. Police behavior in the community reflected officers' concerns, grounded in informal organizational norms, about the maintenance of authority and respect. The police believed that blacks are naturally prone to criminality and that they are a particular threat to police authority. Thus, situations calling for extra-legal violence frequently involved blacks perceived as disrespectful of the police. Not only were blacks particular targets of illegitimate violence, but also the police felt impunity from sanctions because blacks lacked political power.

Subsequent research has relied on a variety of data sources to investigate how situational, individual, and organizational variables relate to police brutality. To date, observational studies have provided the most systematic and reliable evidence (Worden, 1996:33). The first large-scale observational study was conducted for the President's Commission on Law Enforcement and Administration of Justice (Black and Reiss, 1967). Data were collected during the summer of 1966 in Boston, Chicago, and Washington, D.C., by 36 observers who accompanied police officers on patrol in high-crime precincts. Data from that study revealed that whites were more likely than blacks to be the victims of excessive force and that individuals having experienced excessive force were more likely to be victimized by an officer of their own race. Those individuals regarded as being in deviant offender roles or as challenging police authority were most likely to experience misuse of force (Reiss, 1968, ! 1971). A multivariate reanalysis of the data by Freidrich (1980), which treated the distinction between reasonable and excessive force as a mat38, no. 2 (May 2000): p. 343-367ter of degree, showed that neither the officer's nor the citizen's race had an effect on the use of force. Other situational variables, however, notably indicators of demeanor toward the police, predicted use of force. Among white officers, prejudice toward blacks also influenced its use. Overall, situational factors were better predictors than either individual or organizational variables.

Observational data collected for the Police Services Study (PSS) by researchers from the Workshop in Political Theory and Policy Analysis at Indiana University and the Center for Urban and Regional Studies at the University of North Carolina have provided another source of data for research on excessive force. The data were collected during the summer of 1977 for 24 police departments located in the metropolitan areas of Rochester, N.Y., St. Louis, Mo., and Tampa-St. Petersburg, Fl. Smith (1986) used these data in a multivariate analysis of police use of coercive authority (nondangerous encounters involving use of physical force or threat of force, arrest, or surveillance), a dependent variable that included both excessive physical force and lesser abuses. Coercive authority was related to individual racial identity in interaction with neighborhood racial composition; police were most likely to use coercive authority against black suspects in neighborhoods that were primari! ly black. Other situational factors, including suspect antagonism, also were found to be related to the use of coercive authority. In a subsequent analysis of the PSS data, Warden (1996) distinguished between the police use of reasonable and improper (excessive or unnecessary) force. Multivariate analyses showed, consistent with Smith's earlier study, that black or antagonistic suspects were more likely to be targets of improper force. The findings parallel Freidrich's insofar as situational factors were important predictors of the improper use of force, whereas individual and organizational variables generally were not.

Taken together, the findings of these studies provide the greatest support for the situational explanation, but are mixed in regard to the argument that minorities are the primary targets of police brutality. Even these large-scale observational studies, however, contain relatively few incidents of excessive force observed in a limited number of jurisdictions. In the Black and Reiss data, approximately 42 incidents of its use are recorded in 5,012 police-citizen transactions (Reiss, 1971:142). The PSS data contain information on 5,688 police-citizen encounters, of which only 23 involved force judged to be unnecessary or excessive (Worden, 1996:36). Although multivariate statistical studies of observational data have provided the most sophisticated analyses of the causes of police brutality, this strategy of data collection has not generated sufficient cases of excessive force for generalizing reliably. This limitation and the inconsistent findings leave open the question of! whether racial and ethnic minorities are the primary victims of excessive force.

Police brutality complaints filed with law enforcement agencies have provided another source of data on the issue. A national study of citizen complaints was conducted recently for the Police Foundation by Pate and Fridell (1993). In 1992, a mail survey was sent to a sample of municipal and county police departments, county sheriffs departments, and state police agencies, of which 1,111 (67.2%) responded. Analysis of excessive force complaints revealed that a substantial majority (89%) were filed with municipal police departments. Blacks were overrepresented among complainants against municipal police officers, but their complaints were less likely to be sustained. Still, the proportion of sustained complaints involving black citizens (27.3%) was somewhat greater than their representation in the general population (21.4%). Hispanics were represented proportionately in the pool of sustained complaints, and other minorities were underrepresented. Whites filed fewer complaints! , but theirs were more likely to be sustained. Still, they represented a smaller fraction of sustained complaints (56.3%) than their representation in the general population (66.2%). An examination of the racial/ethnic identity of officers named in complaints revealed that the distribution was approximately the same as for the population of sworn officers.

The researchers caution that these findings should not be generalized beyond the responding agencies because of the low overall response rate (only three-fourths of the 67.2% of agencies that responded to the survey provided data on excessive force complaints). They further warn that the data comprise self-reports of a sensitive nature and that some agencies may not have provided complete and accurate information. It also should be noted that the analyses are descriptive and do not include systematic multivariate models of the factors that may predict complaints.

Evidence from other data sources offers additional support for the conclusion that minorities are disproportionately the victims of excessive force. Surveys of public opinion generally have shown that minorities are more likely than whites to perceive police abuses, including the use of excessive physical force (Flanagan and Vaughn, 1996). Another source of information has been investigations sponsored by government agencies (e.g., Independent Commission on the Los Angeles Police Department, 1991; U.S. Commission on Civil Rights, 1970) and private organizations (e.g., NAACP, 1995). The findings of such investigations indicate that minorities are the primary targets of police brutality. Although surveys of public opinion and findings from various commissions are the most consistent with the possibility that minorities are the victims of police brutality, they are also the most problematic methodologically. Perceptions of excessive force do not necessarily correspond to legal! definitions (Pate and Fridell, 1993); few citizens have the knowledge to recognize the distinction between force that is reasonable and force that is excessive. The data collection procedures of independent commissions, although providing an array of descriptive evidence, have not involved rigorous methodological approaches that allow for reliable conclusions about the issue.

In sum, the findings of empirical studies tentatively support the argument that minorities are victimized disproportionately by police brutality. Methodological concerns about reliability preclude any definitive conclusion, however. Substantive and theoretical matters also remain unaddressed. Notably, existing research has focused on blacks to the virtual exclusion of other ethnic minorities. This lack is particularly apparent in regard to the rapidly growing Hispanic population. Data from a civil rights commission (U.S. Commission on Civil Rights, 1970) and surveys (Bayley and Mendelsohn, 1968; Holmes, 1998) suggest that Hispanics in the Southwest have experienced abusive police practices, but national data on excessive force complaints show that Hispanics have a rate of sustained complaints proportional to their population (Pate and Fridell, 1993). Another issue is the lack of attention to community characteristics in explanations of police brutality. The sociological exp! lanation has focused on situational factors to the exclusion of structural-level variables. By building on existing theories, particularly the situational approach, predictions can be made regarding the structural-level influences of sociodemographic characteristics of cities on the incidence of police brutality. Relying on the threat hypothesis of crime control, the present study begins to address these issues in an analysis of civil rights criminal complaints against municipal police departments filed with the Civil Rights Division of the U.S. Department of Justice.

MINORITY THREAT AND CRIME CONTROL

Scholars have developed the threat hypothesis to test empirically the ideas of conflict theory. It stipulates that "the greater the number of acts or people threatening to the interests of the powerful, the greater the level of deviance and crime control" (Liska, 1992:18). Whites perceive racial and ethnic minorities in particular as criminal threats (Liska et al., 1981), and their fear of crime is greater when culturally dissimilar minority groups are present (Chiricos et al., 1997; Liska et al., 1981; Liska et al., 1982). The perception of threat associated with the presence of minorities leads to greater crime control efforts.

Tests of the threat hypothesis that focus on policing have used structural-level data to examine how percent nonwhite and economic inequality affect allocations of police resources (e.g., Jackson, 1985; Jackson and Carroll, 1981; Jacobs, 1979), arrests (e.g., Liska and Chamlin, 1984; Liska et al., 1985), and homicides by the police (Jacobs and O'Brien, 1998; Liska and Yu,1992; Sorensen et al., 1993). The findings from that research have tended to support threat hypothesis predictions, although they also reveal complexities in the relationships between measures of threat and measures of crime control. For example, a study of race-specific arrest rates showed that percent nonwhite was related negatively to nonwhite arrest rates (Liska and Chamlin, 1984). The researchers argued that intraracial crimes increase as percent nonwhite increases, and that the police and minority victims alike see such crimes as personal or family matters in which formal intervention is not required.! Such crimes pose no threat to dominant group members or police officers. More consistent support for the threat hypothesis has been provided by studies of homicides by the police, which have shown that such killings increase as percent minority increases (Jacobs and O'Brien,1998; Liska and Yu,1992; Sorensen et al.,1993). The disproportionate rate of minority victimization in homicides by the police may reflect reactions to situations in which officers personally perceive minority threat (Liska and Yu, 1992).

The most plausible explanation of the divergent findings is that threat has multiple dimensions, involving the interests of both dominant group members and the police. Work on the threat hypothesis has focused primarily on the mobilization of social control to protect the dominant classes. Resource allocations and departmental policies may reflect perceived threats to dominant group interests, but police officers are hardly automatons blindly following dominant group imperatives. Moreover, police behavior is characterized by a high degree of discretion and a low degree of visibility and may thus be open to extra-legal influences (e.g., Smith and Visher, 1981). The salience of threats perceived directly by the police should be more important than distal threats to the dominant group in predicting their street-level responses to minorities. These considerations suggest that police brutality is particularly likely to be related positively to the presence of minorities.

Within the police subculture, extra-legal force is considered a normal and essential instrument of control for handling those individuals perceived as threatening to the police or otherwise discredited (Hunt, 1985; Skolnick and Fyfe, 1993; Van Maanen, 1974; Westley, 1970). Given the extra-legal quality of police brutality, it is available in situations in which legal responses are deemed inappropriate or insufficient. Excessive force can be employed against threatening persons when no probable cause to arrest exists or when "justice" calls for informal sanctions in addition to any formal ones.

Minority attitudes and actions in particular may be seen as threatening to an officer's well-being or challenging to an officer's authority (Chevigny, 1969; Skolnick and Fyfe, 1993; Westley, 1970). The police pejoratively characterize and differentially respond to those individuals in the poorer minority areas of a city (Irwin, 1985; Sampson, 1986, Westley, 1970). Their working personality in such locales often consists of stereotypes, cynicism, mistrust, suspicion, and hostility (Anderson, 1990; Irwin, 1985; Skolnick, 1975; Westley, 1970). Although white police officers have been identified as the primary source of police-minority tensions (e.g., Feagin, 1991; NAACP, 1995), research findings indicate that the use of excessive force is unrelated to an officer's racial or ethnic identity (Freidrich, 1980; Pate and Fridell, 1993). Minority officers also may perceive tower class members of their group as a threat and a challenge to their authority (Alex, 1969). Further, the pe! er culture of policing may pressure minority officers into misconduct against minority citizens (Alex, 1969; Locke, 1996).

Minority citizens, in turn, distrust the police, whom they view as threatening symbols of oppression (Chamlin, 1989; Feagin, 1991; Locke, 1996; NAACP, 1995). They may be more antagonistic to the police, which increases the severity of both formal and informal police sanctions against them (Smith, 1986; Smith and Visher, 1981). Minority antagonism also may precipitate extra-legal violence, including deadly force, against the police (Chamlin, 1989). The climate of mutual threat means that the mere presence and visibility of minorities may amplify the perception of risk among police officers, irrespective of their racial/ethnic identity. Therefore, the police should feel particularly threatened, and hence more likely to employ coercive controls, in cities with a high percentage of minorities (Liska and Yu, 1992).

RESEARCH PROBLEM

This study analyzes civil rights criminal complaints alleging police brutality by officers in municipal police departments that were investigated by the Federal Bureau of Investigation (FBI) and reported to the Civil Rights Division of the U.S. Department of Justice (1991; DOJ hereafter).1 In a study of patterns of police brutality, the DOJ used its Civil Rights Division's computer database of official misconduct complaints (the vast majority of which involve allegations of brutality) to calculate the average number of police brutality complaints for agencies with 2+ complaints annually for the period 1985 to 1990, inclusive. These agencies represent a small fraction (.01%) of all law enforcement agencies in the United States, but they generated almost half of the 15,279 complaints of police misconduct received by the Criminal Section of the Civil Rights Division during this period. The DOJ concluded that no discernible pattern exists in the civil rights criminal complaints! data, but that study included no variables measuring characteristics of minority populations.

This investigation reexamines the DOJ's data for municipal police departments with a more complete set of predictor variables specified by the threat hypothesis. Following the DOJ (1991) research, average annual civil rights violations criminal complaints are operationalized as an indicator of underlying patterns of police brutality. The number of excessive force complaints that citizens file with municipal police departments, which involve a wide range of force, may be far greater than the number that the FBI investigates (Pate and Fridell, 1993), but the majority (nearly 85%) of such complaints are not sustained. The DOJ is selective in its investigation of police brutality criminal cases, pursuing only those of sufficient substance and strong evidence (Skolnick and Fyfe, 1993). Consequently, the DOJ data most validly indicate patterns of relatively serious police brutality (e.g., severe beatings and unjustifiable shootings) supported by relatively strong evidence.

METHOD

The data for this study were obtained from three sources: the DOJ Police Brutality Study FY 1985-FY 1990, Uniform Crime Reports for 1985-90, and U.S. Census reports for 1990. Of the 187 agencies included in the DOJ study, 106 were police departments of various types. This study is limited to municipal police departments, which are represented in sufficient numbers to be analyzed systematically.2

Two subpopulations of cities were included in the analysis. First, the statistical analysis was conducted using the data for the municipal police departments (n = 99) included in the DOJ study. Then, it was conducted for all U.S. cities with populations of 150,000 or more (n = 115), of which 64 were contained in the DOJ data. More than half (55.6%) of the U.S. cities of 150,000+ population are contained in the DOJ data, whereas a tiny fraction of smaller cities are included 3 The first analysis provides an understanding of the variables that predict police brutality complaints in those municipalities with a relatively large number of civil rights criminal filings; however, it provides no comparison to municipal police departments with fewer complaints. The second analysis provides that comparison, including in it the larger municipalities that have produced the bulk of civil rights criminal complaints. Also, large cities have been the focus of research on police brutality a! nd on the threat hypothesis.

The dependent variable in the analyses was the average annual number of civil rights violations criminal complaints. When available, this figure was coded from the data provided in the DOJ report; those jurisdictions had a minimum average of two complaints annually. The figure was set at one per year for the 150,000+ population cities that were not included in the DOJ study.4

The independent variables included city population, index crime rate, percent black, percent Hispanic, majority/minority income inequality, and region. City population is important partly because larger cities have more law enforcement personnel and more arrests than do smaller ones.s The populations of larger cities also may be characterized by greater fear of crime and greater perception of minority group threat (Jackson, 1992).

The index crime rate, percent black, percent Hispanic, and majority/ minority income inequality variables measured the central elements of the threat hypothesis. The first of these variables is a measure of threatening acts. The mean index crime rate ((mean annual index crimes/1990 population] x 100,000) was calculated for the six-year period under study. Data were missing on this variable for seven cities in the DOJ study. The mean was substituted for those cities, and a dummy variable denoting missing data cases (1 = missing; 0 = nonmissing) was included in the equations. None of the cities of 150,000+ population had missing data on this variable.

Percent black and percent Hispanic were measures of threatening people. Blacks and Hispanics are by far the largest minority populations in the United States, and they are seen by the public, policy makers, and police as posing the greatest threat because they comprise the majority of urban gangs and are thought to represent a major part of the drug problem in the United States (Jackson, 1992). Moreover, blacks and Hispanics are considered the primary targets of police brutality (NAACP, 1995).6

The other measure of threatening people used here was majority/minority income inequality. The greater the gap between minorities' and dominants' incomes, the greater the threat minorities are seen as posing (Jackson and Carroll, 1981). Majority/minority income inequality was defined as the ratio of the Anglo (non-Hispanic white) median household income to the weighted mean of black and Hispanic median household incomes (the higher the score on this variable, the greater the gap between Anglo and minority household incomes)? Data were unobtainable for this variable in ten of the cities included in the DOJ study; the mean was used for those cases, and a dummy variable indicating missing data cases (1 = missing; 0 = nonmissing) was included in the analyses. No missing data existed in the cities with populations of 150,000+.

A set of dummy variables indicating region (South, Southwest, Northwest, and North [necessarily deleted comparison category]) was also included in the statistical analyses g Differences between regions may have important implications for policing. For example, regional differences in ethnic hostilities and systems of ethnic stratification may influence efforts at crime control (Jackson, 1989, 1992).

ANALYSIS AND FINDINGS

As discussed above, the statistical analysis was performed separately for cities with 2+ civil rights complaints and for those with populations of 150,000+. Two models were estimated for each set of cities. In the first, the average annual number of civil rights criminal complaints was regressed on the independent variables in an additive model. In the second model, terms were added for the statistical interactions of the region dummy variables with percent black, percent Hispanic, and majority/ minority income inequality. Regional variations in ethnic relations may influence degree of perceived threat (Jackson, 1989, 1992); if degree of threat does in fact vary by region, that should be revealed in interactions between the region variables and the measures of threatening people.

The equations were estimated using linear ordinary least squares (OLS) regression. Count-dependent variables, such as used here, tend to be skewed to the lower end of the distribution with large outcomes being rare events (Neter et al., 1996). Although linear OLS provides unbiased estimates of model parameters for count variables, inferences may be unreliable because of heteroscedasticity. Accordingly, all equations were reestimated assuming conditional Poisson distributions of the dependent variable (Weisberg, 1985:133-135). With count responses that follow the Poisson, a regression model with square-root-transformed responses will be homoscedastic. The equations using the Poisson transformations agreed closely with the estimates obtained from the linear equations, indicating that heteroscedasticity is not a problem in the linear models. Therefore, the parameters from the linear OLS equations are reported, consistent with the theoretical model.

Additionally, statistical models estimated for aggregate-level data are susceptible to problems of multicollinearity and influential cases. Although a perusal of the correlation matrices for the two sets of data (see the Appendix) revealed no obvious problems of multicollinearity, variance inflation factors (VIFs) were calculated for each variable in each equation. The VIFs for the additive (< 2.2) and interaction (< 5.2) equations for both data sets are well below the generally accepted limit of ten (Neter et al., 1996:387), which indicates that the OLS estimates are not unduly influenced by multicollinearity.

The presence of influential cases may produce unstable parameter estimates. Cook's D was therefore calculated for all cases. One case in each equation-New York City-has a value of Cook's D (1.1 to 2..3) in excess of the level (1.0) at which changes in estimates may occur (Weisberg, 1985), with no other case approaching that level (D < .50; p > .90). Accordingly, the equations were reestimated with New York City deleted from the data. Each regression coefficient was compared with its counterpart from the equations for the full data sets using the DFBETAS measure of influence (Neter et al., 1996). The coefficients change slightly across models, and none of the DFBETAS surpasses its limit (2/square root of n ). Therefore, all cases were retained.

CITIES WITH 2+ CIVIL RIGHTS CRIMINAL COMPLAINTS ANNUALLY

The municipal police departments contained in the DOJ data were analyzed first. As to be expected, the additive equation presented under model 1 in Table 1 shows that the city population exerted a strong positive influence on the average annual civil rights criminal complaints. It may also be seen that percent Hispanic and majority/minority income inequality were associated positively with civil rights complaints, consistent with the threat hypothesis. The effects of index crime rate and percent black were small and nonsignificant, however, as were those of the region variables.9

Before testing for statistical interaction, percent black squared and percent Hispanic squared terms were added to the equation estimated in model 1 to test the relative explanatory power of the linear threat hypothesis (Liska, 1992) versus the nonlinear power-threat hypothesis (Blalock, 1967). That analysis (not presented in tabular form) initially suggested a possible nonlinearity involving percent Hispanic. However, the nonlinear relationship was not consistent with the predictions of the power-threat hypothesis, and it was found to be nonsignificant when the model was specified to permit interaction. Accordingly, only the interaction results will be discussed here.

Table 1.

The equation derived from the examination of statistical interaction is presented under model 2. It may be seen that the effects of the population and majority/minority income inequality variables remained essentially the same as in model 1. Consistent with previous research suggesting that southwestern Hispanics are perceived as threatening (Jackson, 1985), the only significant interaction term that surfaced was the positive effect of Southwest x percent Hispanic. All other interaction terms were negligible and therefore were excluded from the equation.10 The additive coefficient for percent Hispanic was small and nonsignificant after the inclusion of the Southwest x percent Hispanic term, indicating that percent Hispanic was not related to civil rights complaints except in the Southwest. As percent Hispanic increased in the Southwest, the number of civil rights complaints increased. That relationship was very strong; the standardized coefficient for the Southwest x percen! t Hispanic term is the largest in the equation, and the inclusion of this single interaction term resulted in a substantial increase in explained variance from model 1 (31.8%) to model 2 (39.3%).

The negative coefficient for the Southwest dummy variable indicates that the predicted number of complaints in southwestern communities would be lower than in other regions if percent Hispanic is set at zero. The predicted value of complaints converges at about 20% Hispanic because of the slope difference, and on average there are 25.6% Hispanics in the southwestern cities and 7.2% elsewhere. Therefore, the negative coefficient for the Southwest dummy variable is not substantively important.

Taken together, the findings for the municipal police departments with 2+ civil rights criminal complaints annually provide partial support for threat hypothesis predictions regarding threatening people. They also refute the DOJ conclusion that no discernible pattern is evident in the data.

CITIES WITH POPULATIONS OF 150,000+

The findings for the additive equation presented under model 1 show that city population, percent black, percent Hispanic, and majority/minority income inequality were related positively and strongly to the average number of civil rights criminal complaints in cities of 150,000+ population, net of the effects of other variables in the equation. The index crime rate and region variables had no effects. The findings are very similar to those obtained in the analysis of the 2+ complaints cities presented above, with the notable exception that here percent black exhibited a strong positive relationship to average annual complaints, consistent with the prediction of the threat hypothesis. The positive relationship between percent black and complaints appears here because 2+ complaints cities of 150,000+ had proportionately larger black populations (27.4%), on average, than did other cities of 150,000+ (15.9%). Apparently, the effect of the variable was attenuated in the first an! alysis because cities with 2+ civil rights criminal complaints annually also have relatively large black populations.

The same procedure employed in the analysis of the 2+ complaints cities was used to obtain the equation reported under model 2. First, nonlinearity was tested (none was detected), and then statistical interaction was examined. The pattern of findings observed in the 2+ complaints cities again emerged; the statistically significant variables in model 1 exhibited similar effects in model 2, with the positive relationship between percent Hispanic and civil rights complaints still restricted to the Southwest. The Southwest x percent Hispanic interaction effect was not as strong in these data (the unstandardized regression coefficient was approximately one-half as large as in the model 2 equation for the 2+ complaints cities), yet its standardized regression coefficient was the second largest in the equation. The Southwest dummy variable was not significant. It is particularly noteworthy that percent Hispanic remained an important predictor of complaints in the Southwest, becaus! e 16 southwestern cities of 150,000+ population were added to the 2+ complaints cities data and 4 smaller ones were deleted.

The analysis of the cities of 150,000+ population has provided a test of the threat hypothesis in large cities, which have been the focus of past research on both the threat hypothesis and police abusive practices. It has also provided a basis of comparison lacking in the original DOJ data set. The findings are consistent with threat hypothesis predictions regarding threatening people and, with exception of the percent black effect, the results from the analysis of the cities with 2+ complaints.

SUMMARY AND DISCUSSION

This study has sought to determine, based on the predictions of the threat hypothesis, the factors that influenced the average annual civil rights criminal complaints alleging police brutality by officers of municipal police departments for the period 1985-1990. The results for measures of threatening people (percent black, percent Hispanic, and majority/minority income inequality) are generally consistent with threat hypothesis predictions. Those results for threatening acts (index crime rate) are not, but previous findings regarding this variable have been inconsistent, perhaps because serious crimes tend to be intraracial.

Two considerations speak to the validity of the findings. Notably, the effects of the independent variables are similar across two separate analyses encompassing substantially different subpopulations of cities, one defined by frequency of civil rights criminal complaints (2+ annually), the other by population size (150,000+). Further, various diagnostics indicate the statistical appropriateness and robustness of the linear threat model for both analyses, though inferences must be limited to the two important subpopulations included in them.

Percent black is the only variable that has different effects in the two sets of data. The variable does not have an effect in the 2+ complaints cities, apparently because those cities have relatively large black populations, but percent black has a strong positive relationship to civil rights criminal complaints in the data for cities of 150,000+ population. This finding corresponds to observations from earlier studies of abusive police practices, which suggest urban blacks are disproportionately targeted by the police (NAACP, 1995; Smith, 1986; Westley, 1953, 1970; Worden, 1996). America's urban black population is deeply impoverished and hypersegregated, conditions that produce crime and social disorder (Massey and Denton, 1993; Skogan, 1990). Also, more unrest may exist where the income gap between the dominant and minority groups is greatest (Jackson and Carroll, 1981). The findings for percent black, as well as for majority/minority income inequality, support the argu! ment that the conditions of urban life amplify the police's perception of minority threat and increase the likelihood of their using coercive control (Liska and Yu, 1992).

Also of key interest is the finding that, in both subpopulations of cities, percent Hispanic is a powerful predictor of average annual civil rights criminal complaints in the southwestern United States. Tensions between Anglos and Hispanics (nearly all of whom are of Mexican origin) in the Southwest have persisted throughout this century, and those of Mexican origin continue to be seen as threatening dominant group interests (Calavita, 1996). Surprisingly, given the historically poor relations between Anglos and Hispanics in the Southwest, comparatively little research has examined police-minority relations in the region. Still, the Southwest by percent Hispanic interaction observed here is similar to Jackson's (1985) findings in regard to the effect of percent Hispanic on urban financial commitments to policing in western and southern cities (cities in southwestern states were not separated from those two regions). Moreover, field investigations and hearings by the U.S. Co! mmission on Civil Rights (1970) surfaced evidence of abusive police practices in the Southwest, and attitudinal studies have shown that people of Mexican origin perceive more police abuses than do Anglos (Bayley and Mendelsohn, 1968; Holmes, 1998). The findings here provide more reliable data to support the conclusion that southwestern Hispanics are targets of such crime control strategies, including police brutality.

Overall, the results of this study have theoretical, substantive, and policy implications. Generally, police brutality has special significance to conflict theory, which holds that the essential component of state authority is its monopoly of force (Jacobs and Britt, 1979). More specifically, the investigation is important theoretically for two reasons. First, it extends work on the threat hypothesis from formal mechanisms of crime control by the police to an informal one. Second, it suggests that proximal threats perceived by agents of social control, not just more distal threats to the dominant group, determine the use of coercive power. This study and the existing body of work on the threat hypothesis indicate that both legal and extra-legal mechanisms are used to control minorities perceived as threatening the dominant group or the police.

At the same time, an important issue remains to be addressed. Research on minority threat has assumed implicitly that political and law enforcement power structures are controlled by whites and that the ranks of police departments are predominantly white. Although their presence may curb police violence against minorities (e.g., Jacobs and O'Brien, 1998), few cities have minority mayors or chiefs of police. Perhaps more significantly, even though minorities remain underrepresented in the ranks of municipal police departments, their representation is increasing and cities with relatively large minority populations generally have a relatively large number of minority police officers (Walker, 1992:246-249). Yet, police officers and more affluent citizens, irrespective of their racial/ethnic identity, may perceive threat largely from minority underclass populations. For example, middle-class black police officers may maintain cordial relations with more affluent black citizens,! but they, like white officers, may perceive proximate threats from more antagonistic lower class blacks in poorer neighborhoods perceiving them as representatives of an oppressive power structure (Alex,1969). Research suggests that it is black citizens in predominantly black neighborhoods who are most likely to experience coercive police authority (Smith, 1986). These considerations indicate a need for studies that examine minority threat as it attaches specifically to the economically marginalized and segregated segments of minority populations. Research is needed to examine the relationships of minority threat and crime control measures at the neighborhood level within cities rather than at the community level between cities, which has been the focus of previous research in this theoretical tradition.

Substantively, the findings of this study add a new dimension to empirical work on police brutality. Studies have indicated that minorities are the primary targets of excessive force, but their generalizations have been limited for a variety of reasons (Locke, 1996; Pate and Fridell, 1993). This study validates and extends the findings of earlier ones by demonstrating a broad pattern of civil rights criminal complaints alleging police brutality, particularly in large cities with relatively large and relatively poor minority populations. It also provides much needed data on the rapidly growing Hispanic population, which has not been a focus of previous research on either police brutality or minority threat.

Finally, policy implications of the findings exist. Clearly, the most immediate issue is identifying ways of ameliorating the ongoing tensions between minorities and the police. The solutions that typically have been suggested, such as training, community-oriented policing, and civilian review (see, e.g., NAACP, 1995), may prove useful stopgaps. Still, they may have limited value because they overlook the larger problem, namely, the segregation and deprivation of America's minority populations. The findings of this study support the argument of conflict theory that policeminority relations symbolize the deeply rooted social divisions separating dominants and minorities. Even efforts to increase minority representation within police departments seem likely to have relatively little success because of class distinctions and antagonisms within minority populations. Popular policy proposals to address the problem will likely have little effect because they focus on altering the! individual and interpersonal dynamics of police-minority relations while failing to recognize that those dynamics are produced by the underlying structural divisions of interest within society.

Appendix.

The assistance of Amy Johnson, Michele Mintling, and Joy Thompson during the collection of data for this project is greatly appreciated. Thanks to former Senator Alan Simpson's staff for helping me obtain a copy of the U.S. Department of Justice Police Brutality Study FY 1985-FY 1990. I appreciate the helpful comments of Richard Anderson-Sprecher, Audie Blevins, Burke Grandjean, Garth Massey, William Taggart, and Bryan Vila on drafts of the manuscript. The statistical advice of Richard Anderson-Sprecher, Burke Grandjean, and William Taggart was very helpful. Thanks also for the thoughtful comments provided by the Editor, Robert J. Bursik, Jr., and reviewers of Criminology. The author remains fully responsible for all analyses and interpretations presented herein.

1. The DOJ's authority to prosecute police misconduct is based on Title 18, Section 242, of the U.S. Code, which "applies to anyone who, while acting under color of law conferred by his or her position of authority, willfully interferes with any of the constitutional or federal statutory rights of any U.S. inhabitant" (DOJ, 1991:5). Note that the DOJ's authority to prosecute police brutality cases is not limited by a complainant's race.

2. Other types of agencies were excluded because of their jurisdictional and operational differences, and because too few of them existed to permit comparisons to other agencies with fewer than two civil rights criminal complaints annually. Five of the 106 police departments included in the DOJ data were not included in this analysis because they were not municipal police departments. Two others could not be included in the analysis because census data were unavailable for their minority populations. All data used in this study were for cities (plus communities under the jurisdiction of their police departments), rather than MSAs, to ensure comparability of data sources.

3. Two cities from the DOJ data with populations of 149,000+ were included in this analysis, because a substantial break in the population distribution for the cities included in the DOJ data occurs at roughly that point (149,377 to 122,899). All equations were reestimated after deleting them, with virtually identical results being obtained in each.

4. The number of complaints in the non-DOJ cities was set at one because the average annual complaints was related positively and strongly to city population (as shown below). The largest cities in the United States would be expected to-have some complaints lodged against their police departments, and setting the number of complaints at zero for the non-DOJ study cities would have artificially maximized the difference between them and the cities with 2+ complaints annually. When the equations presented below were reestimated with the number of complaints set at zero for the non-DOJ cities, the regression coefficients were slightly larger, but the results were otherwise identical.

5. The DOJ study, as well as much of the research on the threat hypothesis, included the number of law enforcement employees and the number of arrests in the analysis. These variables were not used here because of the collinearity between them and city population (r > .90), which consistently produced minuscule coefficients for the number of employees and number of arrests variables and standardized coefficients of greater than one for the city population variable when all three were entered simultaneously in the equations. Statistically, the appropriate solution was to include only the city population variable in the equations.

6. The measure that has generally been used in research on the threat hypothesis is percent nonwhite, but it places most Hispanics in the white category simply because they are so recorded in census data. That approach is problematic because it combines a minority that has experienced significant discrimination and that has been perceived as threatening, at least in some regions (Jackson, 1985), with the dominant group.

7. Majority/minority income inequality was used, rather than a more general

measure of income inequality (such as the Gini coefficient), because minorities are recognized as the most salient source of threat perceived by the police (e.g., Jacobs and O'Brien, 1998). The majority/minority income inequality variable could not be separated into black income inequality and Hispanic income inequality variables because that measurement procedure produced intractable multicollinearity problems.

8. The region variables were constructed so that the "North" category included those states defined as being in the North Central or Northeast by the U.S. Census. The "South" category included all states so defined in the census, except for Texas. The "Northwest" category included those states defined as in the West by the census, except for Arizona, California, and New Mexico. The "Southwest" category included Arizona, California, New Mexico, and Texas, the states in which the majority of the Mexican origin population in the United States reside.

9. The coefficients indicating the effects of the missing crime rate and missing minority income inequality cases are small and nonsignificant, as they are in the interaction equation reported below, revealing no problems with the inclusion of the missing data cases in the analyses.

10. The findings for the interaction terms, both here and in the next analysis, are the same, irrespective of whether the variables are entered simultaneously or with a forward stepwise procedure, providing evidence of the robustness of the interaction estimates.

REFERENCES

Adams, Kenneth

1996 Measuring the prevalence of police abuse of force. In William A. Geller and Hans Toch (eds.), Police Violence: Understanding and Controlling Police Abuse of Force. New Haven, Conn.: Yale University.

Alex, Nicholas

1969 Black in Blue: A Study of the Negro Policeman. New York: AppletonCentury-Crofts.

Anderson, Elijah

1990 Streetwise: Race, Class, and Change in an Urban Community. Chicago, Ill.: University of Chicago.

Bayley, David H. and Harold Mendelsohn

1968 Minorities and the Police: Confrontation in America. New York: Free Press.

Binder, Arnold and Peter Scharf

1982 Deadly force in law enforcement. Crime and Delinquency 28:1-23.

Bittner, Egon

1970 The Functions of Police in Modern Society. Washington, D.C:: U.S. Government Printing Office.

Black, Donald and Albert J. Reiss, Jr.

1967 Studies of Crime and Law Enforcement in Major Metropolitan Areas, Vol. 2, Field Surveys III. Section I: Patterns of Behavior in Police and Citizen Transactions. Washington, D.C.: U.S. Government Printing Office.

Blalock, Hubert M., Jr.

1967 Toward a Theory of Minority-Group Relations. New York: alley. Blauner, Robert

1972 Racial Oppression in America. New York: Harper and Row. Bogomolny, Robert

1976 Street patrol: The decision to stop a citizen. Criminal Law Bulletin 12:544-582.

Calavita, Kitty

1996 The new politics of immigration: "Balanced budget conservatism" and the symbolism of proposition 187. Social Problems 43:284-305. Chamlin, Mitchell B.

1989 Conflict theory and police killings. Deviant Behavior 10:353-368. Chevigny, Paul

1969 Police Powers: Police Abuses in New York City. New York: Vintage. Chiricos, Ted, Michael Hogan, and Marc Gertz

1997 Racial composition of neighborhood and fear of crime. Criminology 35:107-131.

Feagin, Joe R.

1991 The continuing significance of race: Antiblack discrimination in public places. American Sociological Review 56:101-116.

Flanagan, Timothy J. and Michael S. Vaughn

1996 Public opinion about police abuse of force. In William A. Geller and Hans Toch (eds.), Police Violence: Understanding and Controlling Police Abuse of Force. New Haven, Conn.: Yale University.

Freidrich, Robert J.

1980 Police use of force: Individuals, situations, and organizations. Annals of the American Academy of Political and Social Science 452:82-97.

Fyfe, James J.

1980 Police Shootings in Philadelphia: A Systems Model Analysis. Report to the U.S. Department of Justice, Civil Rights Division.

1981 Race and extreme police-citizen violence. In R. L. McNeely and Carl E. Pope (eds.), Race, Crime, and Criminal Justice. Beverly Hills, Calif.: Sage.

Holmes, Malcolm D.

1998 Perceptions of abusive police practices in a U.S.-Mexico border community. Social Science Journal 35:107-118.

Hunt, Jennifer

1985 Police accounts of normal force. Urban Life 13:315-341. Independent Commission on the Los Angeles Police Department

1991 Report of the Independent Commission on the Los Angeles Police Department. Los Angeles: Unpublished report.

Irwin, John

1985 The Jail: Managing the Underclass in American Society. Berkeley, Calif.: University of California.

Jackson, Pamela Irving

1985 Ethnicity, region, and public fiscal commitment to policing. Justice Quarterly 2:167-195.

1989 Minority Group Threat, Crime, and Policing: Social Context and Social Control. New York: Praeger.

1992 Minority group threat, social context, and policing. In Allen E. Liska (ed.), Social Threat and Social Control. Albany, N.Y.: State University of New York.

Jackson, Pamela Irving and Leo Carroll

1981 Race and the war on crime: The sociopolitical determinants of municipal police expenditures in 90 non-southern cities. American Sociological Review 46:390-405.

Jacobs, David

1979 Inequality and police strength: Conflict theory and coercive control in metropolitan areas. American Sociological Review 44:913-925. Jacobs, David and David W. Britt

1979 Inequality and the police use of deadly force. Social Problems 26:403-412.

Jacobs, David and Robert M. O'Brien

1998 The determinants of deadly force: A structural analysis of police violence. American Journal of Sociology 103:837--862.

Kania, Richard R. E. and Wade C. Mackey

1977 Police violence as a function of community characteristics. Criminology 15:27-48.

Liska, Allen E.

1992 Introduction to the study of social control. In Allen E. Liska (ed.), Social Threat and Social Control. Albany, N.Y.: State University of New York. Liska, Allen E. and Mitchell B. Chamlin

1984 Social structure and crime control among macrosocial units. American Journal of Sociology 90:383-395.

Liska, Allen E. and Jiang Yu

1992 Specifying and testing the threat hypothesis: Police use of deadly force. In Allen E. Liska (ed.), Social Threat and Social Control. Albany, N.Y.: State University of New York.

Liska, Allen E., Mitchell B. Chamlin, and Mark D. Reed

1985 Testing the economic production and conflict models of crime control. Social Forces 64:119-138.

Liska, Allen E., Joseph J. Lawrence, and Michael Benson

1981 Perspectives on the legal order: The capacity for social control. American Journal of Sociology 87-413-426.

Liska, Allen E., Joseph J. Lawrence, and Andrew Sanchirico

1982 Fear of crime as a social fact. Social Forces. 60:760-770.

Locke, Hubert G.

1996 The color of law and the issue of color: Race and the abuse of police power. In William A. Geller and Hans Toch (eds.), Police Violence: Understanding and Controlling Police Abuse of Force. New Haven, Conn.: Yale University.

Massey, Douglas S. and Nancy A. Denton

1993 American Apartheid: Segregation and the Making of the Underclass. Cambridge, Mass.: Harvard University.

Myrdal, Gunnar

1944 An American Dilemma. New York: Harper and Brothers.

National Association for the Advancement of Colored People and The Criminal Justice Institute at Harvard Law School

1995 Beyond the Rodney King Story: An Investigation of Police Conduct in Minority Communities. Boston, Mass.: Northeastern University.

Neter, John, Michael H. Kutner, Christopher J. Nachtsheim, and William Wasserman 1996 Applied Linear Statistical Models. 4th ed. Chicago, Ill.: Irwin.

Pate, Anthony M. and Lorie A. Fridell

1993 Police Use of Force: Official Reports, Citizen Complaints, and Legal Consequences. Washington, D.C.: Police Foundation.

Piliavin, Irving and Scott Briar

1964 Police encounters with juveniles. American Journal of Sociology 70:206-214.

Reiss, Albert J., Jr.

1968 Police brutality - Answers to key questions. Transaction 5:10-19.

1971 The Police and the Public. New Haven, Conn.: Yale University.

Smpson, Robert J.

1986 Effects of socioeconomic context on official reaction to juvenile delinquency. American Sociological Review 51:876-885.

Sellin, Thorsten

1930 The negro and the problem of law observance nad administration in the light of social research. In Charles S. Johnson (ed.), The Negro in American Civilization. New York: Holt.

Skogan, Wesley G.

1990 Disorder and Decline: Crime and the Decay in American Neighborhoods. New York: Free Press.

Skolnick, Jerome H.

1975 Justice Without Trial. 2d ed. New York: Wiley.

Skolnick, Jerome H. and James J. Fyfe

1993 Above the Law: Police and the Excessive Use of Force. New York: Free Press.

Smith, Douglas A.

1986 The neighborhood context of police behavior. In Albert J. Reiss, Jr., and Michael Tonry (eds.), Crime and Justice: A Review of Research. Vol. 8. Chicago, Ill.: University of Chicago.

Smith, Douglas A. and Christy A. Visher

1981 Street-level justice: Situational determinants of police arrest decisions. Social Problems 29:167-177.

Sorensen, Jonathan R., James W. Marquart, and Deon E. Brock

1993 Factors related to killings of felons by police officers: A test of the community violence and conflict hypotheses. Justice Quarterly 10:417-440.

Turk, Austin

1969 Criminality and the Legal Order. Chicago, Ill.: Rand McNally. U.S. Commission on Civil Rights

1970 Mexican Americans and the Administration of Justice in the Southwest. Washington, D.C.: U.S. Government Printing Office.

U.S. Department of Justice, Civil Rights Division, Criminal Section

1991 Police Brutality Study FY 1985-FY 1990. Unpublished report. Van Maanen, John

1974 Working the street: A developmental view of police behavior. In Herbert Jacob (ed.), The Potential for Reform of Criminal Justice. Beverly Hills, Calif.: Sage.

Walker, Samuel

1992 The Police in America: An Introduction. 2d ed. New York: McGrawHill.

Weisberg, Sanford

1985 Applied Linear Regression. 2d ed. St. Paul, Minn.: University of Minnesota.

Westley, William A.

1953 Violence and the police. American Journal of Sociology 59:3441. 1970 Violence and the Police: A Sociological Study of Law, Custom, and Morality. Cambridge, Mass.: Massachusetts Institute of Technology. Worden, Robert E.

1996 The causes of police brutality: Theory and evidence on police use of force. In William A. Geller and Hans Toch (eds.), Police Violence: Understanding and Controlling Police Abuse of Force. New Haven, Conn.: Yale University.