Racial bias in police use of lethal force in Brazil

Abstract

This research attempts to test the existence of racial bias in the use of lethal force by police in Brazil. Data from Rio de Janeiro and São Paulo reveal that the proportion of blacks and mulattos among lethal victims of police intervention is higher than their respective share in the population. The paper further analyses the ratio of individuals killed vs. those wounded by police in Rio and shows that the likelihood of being killed is higher for blacks than for whites. Overall findings support the hypothesis of racial bias in police use of lethal force.

Keywords:

• police
• lethal force
• ethnicity
• racial bias
• racial discrimination

Notes

1. This research is part of a project conducted thanks to a grant of the Ford Foundation in Brazil. The research team included James Cavallaro, Ariel Alves, and Cristina Jakimiak, apart from the author.

2. In Brazil, homicide and intentional crimes against the life of others are the only crimes tried by a jury.

3. The ‘Ouvidoria de Polícia’ is a civilian police oversight body whose mandate can be summarized as follows: (a) collecting citizens’ complaints related to abuses and crimes allegedly committed by police agents while protecting the identity of the witnesses; (b) following up the investigations of such cases carried out by Internal Affairs Units; and (c) keeping society informed on these issues.

4. Opponents are individuals against whom police shot intentionally.

5. There were also another 19 fatal and 33 non‐fatal civilian victims in cases where it was not possible to determine whether the shots had been intentional or accidental.

6. b = 0.35; std. error = 0.182; df = 1; p = 0.055.

7. Accidental victims may have been wounded or killed from police as well as from suspects’ bullets.

8. According to estimations based on a sample carried out by the Official Bureau of the Census (IBGE) together with the census itself in 1991, whites comprised 64% of the population in the rest of the city but only 38% in the favelas (Preteceille & Valladares, 1999).

9. (Likelihood Ratio Chi‐square test = 12.66; df = 2; p = 0.0018).

10. (Likelihood Ratio Chi‐square test = 0.476; df = 2; p = 0.788).

11. (Likelihood Ratio Chi‐square test = 4.364; df = 2; p = 0.1128).

12. (Likelihood Ratio Chi‐square test = 12.66; df = 2; p = 0.0018).

13. (Likelihood Ratio Chi‐square test = 42.86; df = 1; p < 0.0001).

14. Applying the model just to blacks and browns, the race by outcome interaction yields the following results: Likelihood Ratio Chi‐square test = 0.705; df = 1; p = 0.401.

15. Actually, the unit of analysis for weapons apprehended is the episode, not the individual. Thus, all the victims of one episode will be reporting the same guns. Obviously, one episode can have victims of different races involved. So the data for each race group are not completely independent, but they are still valuable in order to have an idea of the situation.

16. (F = 2.78; df = 2 & 1001; p = 0.063).

17. (F = 0.22; df = 2 & 1005; p = 0.803).

18. (Chi‐square = 6.14; df =6; p = 0.407).