DISORGANIZATION, CONFLICT, AND ORGANIZATIONAL CHARACTERISTICS: WHAT ARE THE STRUCTURAL CORRELATES OF VIOLENCE AGAINST POLICE OFFICERS?
DISORGANIZATION, CONFLICT, AND ORGANIZATIONAL CHARACTERISTICS: WHAT ARE THE STRUCTURAL CORRELATES OF VIOLENCE AGAINST POLICE OFFICERS?
Dale W. Willits
University of New Mexico
Draft
December, 2006
ABSTRACT
Violence against police officers is often understood as either an extension of normal violent behavior framed in terms of social disorganization and structural disadvantage or as a result of political subordination in keeping with political conflict theory. While these theories are undoubtedly important in explaining some of the violence against police officers, they tend to ignore police department organizational characteristics and/or assume that these characteristics have only minimal influence on violence against police officers. Using data from the Law Enforcement Officers Killed and Assaulted Records (LEOKA) dataset and the Law Enforcement Management and Administrative Statistics (LEMAS) survey, this study examines the independent and combined effects of disorganization, conflict, and organizational characteristics on violence against police officers. Findings indicate that all three processes contribute to violence against police officers.
Structural measures of disorganization and conflict, specifically, unemployment, segregation, and the presence of a black mayor are significant predictors
of police officer victimization. Additionally departments with protective gear regulations, more stringent education requirements, and higher proportions of black
officers experience less violence against their police officers.
INTRODUCTION
A significant number of studies have analyzed the social-structural correlates of violent crime. This research reveals statistically significant links between rates of violent crime and such structural characteristics as disadvantage, the proliferation of guns and drugs, labor market changes, and demographic shifts. It is interesting to note, however, that these factors appear to affect violence against police officers in different ways than they affect violence in general. On the national level, Kaminski and Marvel (2000) found that demographic shifts and the advent of the crack “epidemic” had a smaller impact on police homicides than on general homicides. Conversely, economic conditions and prison populations appear to have a larger role in explaining police homicides than in explaining general homicides (Kaminski and Marvel, 2002, 171-172).
These findings imply that the etiology of violence against police officers may be different than that of general violence. Previous research has examined the association between disorganization, conflict, and violence against police officers (Peterson and Bailey, 1988; Jacobs and Carmichaels, 2002). This research has found that areas with high levels of disorganization and inequality tend to have higher rates of police victimization. Previous research, however, has ignored police departments’ role in anti-police violence (Lott, 2000 is an exception). Considering the emphasis that police departments place on officer safety (Herbert, 1998),
it seems misguided to ignore the potential influence that the organizational characteristics of police departments likely have on officer safety. This study seeks to test the hypothesis that organizational characteristics of police departments influence violence against police officers.
VIOLENCE AGAINST POLICE OFFICERS
There has been both micro and macro level research on violence against police officers. The micro or situational level research focuses on how the individual
characteristics of police officers, suspects, and incidents might influence violence against police officers. Research at this level has found that minority police officers are more likely to be injured and killed than white police officers (Fyfe, 1981; Geller and Karales, 1981; Konstantin, 1984). This research does not argue that minority officers are more likely to be targeted due to racial discrimination or that minority officers are inferior to white officers. Instead, they argue that minority officers are disproportionately “assigned to high-crime inner city precincts in an effort to ameliorate the police-citizen racial conflicts characterizing these areas” (Konstantin, 1984: 32).
Furthermore, they argue that a large number of police officer homicides occur when the officers are off duty, which may be due to the fact that minority officers are more likely to live in high crime minority areas (Fyfe, 1981; Geller and Karales, 1981; Konstantin, 1984). Micro level research has also demonstrated that robbery-related service calls have the highest probability of resulting in officer injury (Geller and Karales, 1981; Konstantin, 1984). The macro level research examines how structural factors may influence violence against police officers. Some of this research focuses on national trends (Kaminski and Marvel, 2002), though most focuses on the state and city-level correlates of police victimization (Lester, 1978; Peterson and Bailey, 1988; and Jacobs and Carmichael, 2002). This research can be divided into three broad categories: disorganization, conflict, and organizational characteristics.
Social Disorganization
The social disorganization model suggests that violence against police officers is caused by the same factors that influence other types of violent crime. From this perspective communities that are structurally disadvantaged are likely to have higher rates of crime because disadvantage tends to reduce community level cohesion and social control. Traditionally, economic deprivation, family disruption, residential stability, and population heterogeneity have been key measures of social disorganization and together are thought to indicate higher levels of social disorganization (Sampson and Groves, 1989). Increases in disorganization indicate less community level cohesion, which in turn predicts more crime (Markowitz, et al., 2001).
Empirical studies on violence against
police officers have found some support for the disadvantage thesis. Lester (1978) found that poverty was significantly correlated with the murder rates of police officers, while Peterson and Bailey (1988) found that poverty and divorce rates are important predictors of violence against police officers.
Conflict
Noting that the social disorganization perspective did not implicitly account for inequality, recent research has examined the role that conflict theory might play in
explaining violence against police officers. The conflict approach to understanding violence against police officers emphasizes the role of political and social power
structures. Drawing upon neo-Marxist theory, these scholars argue that, “some of the violence directed at the police is due to primitive resistance against … unequal arrangements” (Jacobs and Carmichael, 2002: 1223).
Jacobs and Carmichael find considerable support for the conflict perspective on police homicides and assaults. Using measures of income inequality and black political power, they find that political and economic inequality account for a significant proportion of the variance in violence against police officers, even when controlling for traditional disadvantage and disorganization factors (Jacobs and Carmichael, 2002). Furthermore, they argue that when controlling for conflict, the disorganization indicators are no longer useful predictors of officer victimization. Jacobs and Carmichael argue that this is evidence that at the violence that happens against police officers is a form of political resistance (Jacobs and Carmichael, 2002, 1244-1245).
Organizational Characteristics
The social disorganization and conflict perspectives reveal many of the structural characteristics that influence violence against police officers. However, just as the social disorganization perspective ignores the political role of policing, the conflict perspective ignores the practical role of policing. The most obvious practical role of policing is to prevent and control crime and maintain public safety and order. There is reasonable evidence that police departments can have varying degrees of success in meeting these goals, and that this degree of success is not entirely contingent on the structural and political influences of a given era and location; but is the result of policing practices and characteristics at the department level. Some argue, for example that while crime, in general, declined in the 1990’s, cities that employed community policing recorded more significant crime drops than areas that employed traditional policing practices (for example, see Xu et. al.).
Previous research has defined officer safety as one of the six important values or normative orders central to policing (Herbert, 1998). It seems reasonable then to expect police leaders to make organizational level decisions that they believe will not only prevent and control crime, but also promote officer safety. Lott (2000) employed this perspective in his analysis of the impact of affirmative action hiring practices on police safety. His results indicate that departments with higher proportions of female officers, controlling for other factors, are likely to experience higher rates of assaults against police officers. Lott claims that this is the case because female police officers are physically weaker than male police officers (Lott, 2000). Drawing on Lott and other policing research, this paper examines the influence of the following characteristics of police departments on police victimization: percentage female officers, percentage black officers, education requirements of police recruits, community policing, protective gear regulations, and budget.
CURRENT RESEARCH
In general, previous research views the three theoretical models described above as separate, competing explanations of violence against police officers. In fact, much of the early research focused exclusively on disorganization and disadvantage (Lester, 1978 and Peterson and Bailey, 1988) and did not consider other possible sources of police victimization. More recent research (see Jacobs and Carmichaels, 2002), while focusing on conflict, considered the social disorganization theory when examining police victimization.
They, however, viewed disorganization as a competing theoretical explanation and argued that when controlling for conflict, indicators of disorganization
(specifically, divorce) did not influence violence against police officers. I have found no research that seriously considers social disorganization and conflict while evaluating the influence of departmental characteristics. The current research tests the separate and combined influence of social disorganization, conflict, and the organizational characteristics on violence against police officers. This is accomplished by first examining indicators of each theoretical model separately, then by examining the joint influence of all three processes in combined models. Hypotheses 1 through 5 describe the expected results:
·Hypothesis 1: Higher levels of social disorganization will be associated with higher rates of violence against police officers.
·Hypothesis 2: Higher rates of conflict will be associated with higher rates of violence against police officers.
·Hypothesis 3: The organizational characteristics of police departments influence violence against police officers.
·Hypothesis 4: Conflict mediates the influence of disorganization on violence against police officers.
·Hypothesis 5: The organizational characteristics of police departments will mediate the influence of the disorganization and conflict on violence against police officers. It is expected that this mediating effect will substantially reduce the role of structural and political variables in estimating assaults against police
officers.
DATA
Data for the current study come from three primary sources: LEOKA, LEMAS, and the US Census. The LEOKA (Law Enforcement Officers Killed and Assaulted)
dataset, which is part of the larger Uniform Crime Report (UCR) data collection effort, provides data on violence against police officers. The LEOKA data are released annually and report monthly totals on the number of officers killed and assaulted. Data on the characteristics of police departments come from the year 2000 Law Enforcement Management and Administrative Survey (LEMAS). The LEMAS dataset is released every 2 to 3 years. In addition, the current study draws indicators of city level characteristics from the US Census Bureau and the Lewis Mumford Center. I omitted cities from the dataset that had populations less than 100,000 in the year 2000.
Previous research has employed similar filtering methods and thus this selection basis was required to maintain comparability to previous research. I also omitted cities that were missing LEOKA or LEMAS data1, as the LEMAS data is required to examine the influence of organizational characteristics on violence against police officers. The final sample in this study includes 201 American cities with populations of over 100,000. The number of injurious assaults of police officers is the dependent variable in these analyses. I use LEOKA data to operationalize the concept of violence against police officers. Homicides are often a favored measure of violent crime, as this crime does not generally suffer from the same underreporting problems of other crime types.
However, homicides, both in general and of police officers, are statistically rare events. The small number of homicides presents a significant problem for any estimation procedure that uses officer homicides as a measure of violence against police officers. The independent variables for this study fall into three categories. These categories, which correspond to the theoretical perspectives outlined in the previous section, are disorganization, conflict, and organizational variables.
Social Disorganization
Data for the disorganization variables come from the U.S. Census Bureau. Social disorganization is often measured with indicators of residential stability, racial and ethnic heterogeneity, and family disruption (Bursik Jr., 1988). I measure residential stability using the percentage of population that has moved in the last five years. I measure racial and ethnic heterogeneity using a racial heterogeneity variable (1 Р̴ 2 i p ), where pi is the proportion of the population of racial/ethnic group i (Wadsworth and Kubrin, 2004: 656).
1 The LEMAS data for 2000 is a sample of American cities, and thus, did not include all cities that were included in the LEOKA dataset. Furthermore, several police departments in cities populations over 100,000 did not report LEOKA and/or LEMAS data for unknown reasons.
2 Alternative models were estimated using the number of officer homicides as the dependent variable. The lack of variance in officer homicides rendered these results uninteresting, as there appeared to be no consistent predictors of officer homicides. Lott (2000) reported similar difficulties in analyzing officer
homicides. Other research has addressed this issue by using time series data, but the use of the LEMAS dataset (which is not released annually) to capture department level factors prevents this approach from being utilized in this study.
This variable, which ranges from 0 (completely homogenous) to 1 (completely heterogeneous), indicates the diversity of a city’s population. Divorce rates are included to measure family disruption. I also include unemployment as an indicator of economic disadvantage, which has been linked to social disorganization (see Osgood 2000).
Political Conflict
The conflict variables include measures of political and economic inequality. The guiding research for this model is the recent work done by Jacobs and Carmichaels
(Jacobs and Carmichaels, 2002: 1240-1242). Following their lead, the conflict variables include racial segregation, income inequality and the presence of a non white mayor. Racial segregation is measured via the index of dissimilarity (http://mumford.albany.edu/census/data.html). Income inequality is measured by dividing
white median income by African American median income; with higher numbers thus suggesting more inequality. The final conflict variable is the presence of a minority mayor. Jacobs and Carmichaels (2002) found that cities with African American mayors, when controlling for other factors, had less violence against police officers.
According to Jacobs and Carmichaels (2002), cities that elect African American mayors are less racially divided, which leads to less conflict and fewer crimes of resistance. The logic behind this measure is that cities that elect African American mayors are politically more equal, which leads to less conflict and thus fewer crimes of resistance. David Jacobs provided a list of cities that had black mayors in the year 2000. This list was checked against information in a database maintained by the Joint Center for Political and Economic Studies.
Department Organizational Characteristics
To measure the organizational characteristics of police departments I include indicators of the racial composition, the operational budget, education requirements,
equipment regulations, and participation in community policing activities for police departments. I use the LEMAS survey to develop indicators for these characteristics. In summary, I use the following indicators: the percentage of sworn officers that are black measures the racial composition of a police department; the police department’s operating budget (logged) measures available funds; the minimum number of semester hours of new recruits measures education requirements; and a dummy variable that indicates whether the use of protective gear was required by the department. Community policing is measured with an index of 22 yes or no questions in the LEMAS survey (Cronbach’s alpha = 0.802). These 22 questions determine the degree to which police departments interact with the communities they serve, and include questions about patrol varieties, implementation of problem solving strategies, and interactions with community
organizations. Other organizational variables were considered for analysis, but are excluded for a variety of reasons3
METHODS
Variables measured as integer counts should not (in general), be estimated with the traditional ordinary least squares (OLS) regression. The OLS approach assumes normality of error terms, which is often not the case in count data. The Poisson distribution is suited for count variables and thus is a better choice for estimating violence 3 Police presence was tested both by the log of police officers and by the police to citizen ratio in early models and found to statistically not significant and moderately to highly correlated with several disorganization, conflict, and other organizational factors.
Indicators of police were dropped from subsequent models to prevent potential collinearity. A variety of communities policing measures were tested in various preliminary analyses, including the percentage of officers assigned to community policing roles and various dummy variable constructs. These alternative indicators provided substantively similar results as the measure used. Measures of training regulations were also considered, but these values were not independent across observations, due to state regulations on the minimum number of training hours for new recruits against police officers than OLS. Because injurious assaults against police officers are over-dispersed, models were estimated using the negative binomial distribution (Osgood, 2000: 28-29). In this instance, a zero-inflated negative binomial model fit better than a standard negative binomial model. Table 1 summarizes the five models examined in this study.
TABLE 1
RESULTS
The results of the zero-inflated negative binomial regressions are displayed in
table 2.
TABLE 2
The results from model 1 provide some support for hypothesis 1. Three of the indicators of disorganization demonstrate the expected sign, indicating that areas with higher levels of disorganization are also likely to have higher counts of injurious assaults of police officers. However, unemployment is the only variable in model 1 that is a statistically significant predictor of injurious assaults against police officers. The positive coefficient implies that areas with higher unemployment rates are also likely to have higher counts of injurious assaults of police officers. The z-statistic for percent divorced is borderline statistically significance (z=1.78, p=0.075).
Model 2 reveals general support for hypothesis 2. Segregation, as measured by the white to black index of dissimilarity, is a statistically significant positive predictor of injurious assaults. Therefore, when controlling for other inequality measures, cities with more segregation are likely to have more violence against police officers. In addition to this, the dummy variable mayor is also a statistically significant predictor of injurious assaults. Therefore, when controlling for other inequality measures, cities with a black mayor are likely to have fewer injurious assaults of police officers than cities without a black mayor.
Controlling for other organizational factors, model 3 shows that education requirements and the use of protective gear are significant predictors of injurious assaults. Departments with more demanding education standards and departments that require officers to wear protective gear are likely to report fewer injurious assaults against police officers. The presence of significant variables indicates that at least some of the organizational characteristics of police departments are important predictors of violence against police officers, thus demonstrating moderate support for hypothesis 3. The gender and racial composition of a police department, as well as community policing practicesand operating budget are not found to be significant predictors of violence against police officers.
Controlling for all other variables, unemployment, segregation, and the presence of a black mayor are significant predictors of violence against police officers in both
models 4 and 5. Protective gear is a significant predictor of violence against police officers in models 3 and 5. Education requirements are no longer statistically significant in model 5; however, the percentage of black police officers is statistically significant.
Controlling for all other measured factors, we find that the cities with more black police officers are likely to report fewer injurious assaults of police officers than cities with fewer black police officers. The magnitude of the coefficient for unemployment decreases when including conflict indicators. However, the decrease (from 11.72 to 10.33) is not substantial, thus revealing no support for hypothesis 4. Similarly, the magnitude of the significant disorganization and conflict variables from previous models did not dramatically change when controlling for the organizational characteristics of police departments, thus revealing no support for hypothesis 5.
The organizational characteristics of police departments do not appear to mediate the effects of social disorganization or conflict. The fact that there are no clear mediating effects implies that the three models tested are more complementary than competing. Log likelihood tests confirm this result, showing that models that include indicators of disorganization, conflict, and department-level characteristics have better fit than models that only consider one or two of the theoretical
perspectives explored in this paper.
DISCUSSION
In general, the results from the zero-inflated negative binomial models demonstrate considerable support for each of the theoretical perspective tested in this
study. Support was found for hypotheses 1, 2, and 3; but not for hypotheses 4 and 5. Including conflict with disorganization improves model fit; but does not diminish the importance of disorganization. Similarly, including organizational characteristics in a model with conflict and disorganization improves model fit; but does not diminish the importance of disorganization and conflict. It appears that the three theoretical models tested in this paper are best described as complementary rather than competing explanations of violence against police officers.
The conflict indicators have the most success in predicting violence against police officers. This is true both in the number terms of the number of significant conflict
variables and in terms of model strength. Segregation and the presence of a black mayor are significant predictors of violence against police officers in all of the models that include these variables. The log likelihood value reported in table three for the conflict model is closer to 0 than the log likelihood value for the social disorganization or organizational models. Jacobs and Carmichaels (2002) found similar results.
Areas characterized by higher levels of inequality are likely to also have higher rates of violence against police officers. These results lend credit to the idea that violence against police officers is often conflict driven. While it seems unlikely that the individual assaulting a police officer is striking out due to the political nature of his or her inequality, it may be the case that individuals in these areas feel less trust and are, in general, more antagonistic towards police officers than people in other, more integrated areas. The organizational characteristics, which have been mostly ignored in previous literature, also proved to be important predictors of violence against police officers.
Protective gear regulations, the percentage of black police officers, and education requirements all have a significant role in predicting violence against police officers. Furthermore, protective gear is one of the most substantively important variables employed in this study. In model five, for example, controlling for other factors, departments that required protective gear for officers had 20% fewer injurious assaults than departments that did not require protective gear for officers. Only the presence of a black mayor seemed to have a larger impact on the expected count of injurious assaults. One important limitation of this research is that this cross sectional analysis cannot provide any causal claims. It is possible that other unmeasured factors influence the amount of violence against police officers. Given this limitation, it is more appropriate to view the results of this study as descriptive in nature. I argue, however, that this limitation does not prevent these results from being useful for policy makers. Until longitudinal data analysis is conducted (which is complicated, given the nature of the LEMAS dataset) or until quasi-experiments are funded, it is probably still useful for police leaders to refer to cross-sectional results when making policy decisions.
The results found here provide, at least, a possible model through which we can understand how violence against police officers happens and provide details on characteristics of police departments seem to be associated with more or less violence against police officers. Furthermore, while these results are not definitive, none of the policy implications seem dangerous and there does not appear to be any obvious downsides to implement these changes.
References
Bursik, Robert Jr. “Social Disorganization and Theories of Crime and Delinquency:
Problems and Prospects” Criminology, Vol. 26., No. 4 (1988): 519-551.
Eck, John and Edward McGuire. “Have Changes in Policing Reduced Violent Crime?
An Assessment of the Evidence.” In The Crime Drop in America, ed. Alfred
Blumstein and Joel Wallman. New York: Cambridge Press, 2000: 207-265.
Fyfe, James J. “Who Shoots? A Look at Officer Race and Police Shooting.” Journal of
Police Science and Administration, No. 9 (1981): 367-382.
Geller, William A. and Kevin J. Karales. Split-second Decisions: Shootings of & by
Chicago Police. Chicago (1981): Chicago Law Enforcement Study Group.
Herbert, Steve. “Police Subculture Reconsidered.” Criminology, Vol. 36, No. 2 (May
1998): 343-370.
Jacobs, David and Jason T. Carmichael. “Subordination and Violence against State
Control Agents: Testing Political Explanations for Lethal Assaults against the
Police.” Social Forces, Vol. 80, No. 4 (June 2002): 1223-1251.
Kaminski, Robert J. and Thomas B. Marvell. “A Comparison of Changes in Police and
General Homicides: 1930-1998.” Criminology, Vol. 40, No. 1 (Nov. 2002): 171-
190.
Konstantin, David N. “Homicides of Law Enforcement Officers, 1978-1980.” Justice
Quarterly, Vol. 1, No. 1 (March 1984): 29-45.
Kubrin, Charis E. and Tim Wadsworth. “Identifying the Structural Correlates of African
American Killings.” Homicide Studies, Vol. 7, No. 1 (Feb. 2003): 3-35.
Lester, David. “Predicting Murder Rates of Police Officers in Urban Areas.” Police Law
Quarterly, Vol. 7 (1978): 20-25.
17
Levitt, Stephen D. “Understanding Why Crime Fell in the 1990s: Four Factors that
Explain the Decline and Six that Do Not.” Journal of Economic Perspectives,
Vol. 18, No. 1 (Winter 2004): 163-190.
Lott Jr., John R. “Does a Helping Hand Put Others at Risk?: Affirmative Action, Police
Departments, and Crime.” Economic Inquiry, Vol. 38, No. 2 (April 2000): 239-
277.
Markowitz, Fred E. and Paul E. Bellair and Allen E. Liska and Jianhong Liu. “Extending
Social Disorganization Theory: Modeling the Relationships between Cohesion,
Disorder, and Fear.” Criminology, Vol. 39, No. 2 (May 2001): 293-320.
Osgood, D. Wayne. “Poisson-Based Regression Analysis of Aggregate Crime Rates.”
Journal of Quantitative Criminology, Vol. 16, No. 1 (2000): 21-43.
Osgood, D. Wayne. “Social Disorganization Outside the Metropolis: An Analysis of
Rural Youth Violence.” Criminology, Vol. 38, No. 1 (Feb. 2000): 81-116.
Peterson, Ruth D. and William C. Bailey. “Structural Influences on the Killing of Police;
A Comparison with General Homicides.” Justice Quarterly, Vol. 5, No. 2 (June
1988): 207-233.
Quinney, Richard. “Class, State, and Crime.” In Criminological Theory:
Past to Present, ed. by Frances Cullen and Robert Agnew. Los Angeles: Roxbury
Publishing Company, 2006: 312-317.
Sampson, Robert J. and W. Byron Groves. “Community Structure and Crime: Testing
Social-Disorganization Theory.” American Journal of Sociology, Vol. 94, No. 4
(Jan. 1989): 774-802.
Terrill, William and S.D. Mastrofski. “Situational and Officer-based Determinants of
Police Coercion.” Justice Quarterly, Vol. 19, No. 2 (2002): 215-248.
18
Tyler, Tom R. and Cheryl J. Wakslak. “Profiling and Police Legitamacy: Procedural
Justice, Attributions of Motive, and Accept of Police Authority.” Criminology,
Vol. 42, No. 2 (2004): 253-281.
U.S. Dept. of Justice, Bureau of Justice Statistics. LAW ENFORCEMENT
MANAGEMENT AND ADMINISTRATIVE STATISTICS (LEMAS): 2000
SAMPLE SURVEY OF LAW ENFORCEMENT AGENCIES [Computer file].
ICPSR version. Washington, DC: U.S. Dept. of Commerce, Bureau of the Census
[producer], 2001. Ann Arbor, MI: Inter-university Consortium for Political and
Social Research [distributor], 2003.
U.S. Dept. of Justice, Federal Bureau of Investigation. UNIFORM CRIME REPORTING
PROGRAM DATA [UNITED STATES]: POLICE EMPLOYEE (LEOKA)
DATA, 2000 [Computer file]. Compiled by the U.S. Dept. of Justice, Federal
Bureau of Investigation. ICPSR ed. Ann Arbor, MI: Inter-university Consortium
for Political and Social Research [producer and distributor], 2002.
Semyonov, Moshe and Yitchak Haberfeld and Yinon Cohen and Noah Lewin-Epstein.
“Racial Composition and Occupation Segregation and Inequality across American
Cities.” Social Science Research, Vol. 29 (2000): 175-187.
Shihadeh, Edward S. and Graham C. Ousey. “Industrial Restructuring and Violence: The
Link between Entry-Level Jobs, Economic Deprivation, and Black and White
Homicide.” Social Forces, Vol. 77, No. 1 (Sept. 1998): 185-206.
Wadsworth, Tim and Charis Kubrin. “Structural Factors and Black Interracial Homicide:
A New Examination of the Causal Process.” Criminology, Vol. 42, No. 3 (2004):
647-672.
Weitzer, Ronald and Steven A. Tuch. “Reforming the Police: Racial Differences in
19
Public Support for Change.” Criminology, Vol. 42, No. 2 (May 2004): 391-416.
White, Michael D. and James J. Fyfe and Suzanne P. Campbell and John S. Goldkamp.
“The Police Role in Preventing Homicide: Considering the Impact of Problem-
Oriented Policing on the Prevalence of Murder.” Journal of Research in Crime
and Delinquency, Vol. 40, No. 2 (May 2003): 194-225.
Xu, Yilu, and Mora L. Fielder and Karl H. Flaming. “Discovering the Impact of
Community Policing: The Broken Windows Thesis, Collective Efficacy, and
Citizen Judgment.” Journal of Research in Crime and Delinquency, Vol. 42, No.
2 (May 2005): 147-186.
22
