Race, Place, and Police-Caused Homicide in U.S. Municipalities

Abstract

We test structural hypotheses regarding police-caused homicides of minorities. Past research has tested minority threat and community violence hypotheses. The former maintains that relatively large minority populations are subjectively perceived as threats and experience a higher incidence of police-caused homicide than whites do, the latter that higher rates of violent crime among minorities create objective threats that explain these disparities. That research has largely ignored some important issues, including: alternative specifications of the minority threat hypothesis; the place hypothesis, which maintains highly segregated minority populations are perceived as especially threatening by police; and police-caused homicide in the Hispanic population. Using data for large U.S. cities, we conducted total-incidence and group-specific analyses to address these issues. A curvilinear minority threat hypothesis was supported by the Hispanic group-specific findings, whereas the place hypothesis found strong support in both total and group-specific analyses. These results provide new insights into patterns of police-caused homicide.

Keywords:

• Police homicide
• race
• place
• racial segregation

Acknowledgements

The authors thank Christopher J. Holmes and anonymous reviewers for helpful comments on drafts of this article. They remain fully responsible for all analyses and interpretations presented herein.

Notes

1 Recent efforts to develop new datasets have sought to overcome the limitations of the SHR data, specifically the problem of undercounting in national estimates. However, these data collections involve potentially untrustworthy media accounts and web-based crowd-sourcing data. While generally resulting in higher police-caused homicide counts than the SHR data, that is largely attributable to the inclusion of many small cities that do not report these data to the UCR, as well as the use of more inclusive definitions of police killings (e.g. adding unintentional and accidental deaths) in some datasets than used in the SHR (intentional police action). These new data sources provide quite different estimates of the incidence and trends in police killings (see http://www.cbc.ca/news/trending/number-of-black-people-killed-by-us-police-still-no-stats-1.3670513). Apart from definitional concerns, there are one or more issues with these data collections that make them very problematic or impossible to use in research that examines patterns of police-caused homicide involving independent police agencies in large cities. Notably, these include: unsystematic data collection and verification procedures; incomplete verification; limited methodological documentation; missing and sometimes inaccurate reporting of victim race/ethnicity; lack of geographical and/or law enforcement agency identification; and/or a political agenda that is clearly incompatible with the norms of scientific research.

2 http://www.s4.brown.edu/us2010/segregation2010/Default.aspx.

3 The North and Midwest regions followed the U.S. Census designations. The South category corresponded with the U.S. Census category of South with the exception of Texas, which was included in the Southwest category. The Northwest included the Census category of West, with the exception of Arizona, California and New Mexico, which were included in the Southwest region along with Texas (see Holmes, 2000).

4 Notably, the bivariate relationships of the theoretically specified predictor variables to police-caused homicide were consistent with theory and largely follow the pattern of findings in the multivariate analyses presented below. The positive correlation coefficients for the segregation variables were fairly large and statistically significant (r_bwseg = .41; r_Hwseg = .43); the positive percent minority coefficients were smaller and statistically significant (r_bpct = .15; r_Hpct = .11); and the coefficients for the three community violence variables were positive and (with the exception of arrest rate, p = .13) statistically significant (r_vcrt = .25; r_arrt = .10; r_polkil = .43).

5 Dividing a continuous predictor variable into a set of dummy variables is a procedure that allows identification of various forms of non-linearity. We divided the segregation variables into quintiles because that is enough categories to identify common forms of non-linearity while retaining enough cases in each category to provide robust parameter estimates.

6 All reported significance tests of differences among dummy-variable categories were obtained using the SAS PROC MIANALYZE F-test procedure.

7 We included the percent black variable because, irrespective of the relative size of the black population, cities with larger numbers of blacks would be expected to have a higher of incidence of police-caused homicide of blacks. We retained the overall violent crime rate because the UCR does not report crimes known to the police by race/ethnicity of alleged offenders.

8 All predictor and control variables included in every reported analysis in this study were set at their means; setting the variables at their means (rather than 0) makes the intercept more realistic for graphical presentation of findings. The predicted value of the first quintile is negative because of the negative intercept in the model.

Additional information

Notes on contributors

Malcolm D. Holmes

Malcolm D. Holmes is a Professor of Sociology at the University of Wyoming. He has published widely on racial/ethnic disparities in criminal justice outcomes. Police violence is the central concern of his current research.

Matthew A. Painter

Matthew Painter is an Associate Professor of Sociology at the University of Wyoming. He completed his PhD at The Ohio State University. His research interests include immigration, race/ethnicity, and wealth inequality.

Brad W. Smith

Brad W. Smith is a Professor in the Department of Criminal Justice at Wayne State University. His research focuses on policing, including studies of police discretionary behavior, citizens’ attitudes toward the police, police-minority relations and police use of force.