Assessing Racial Disparities in Homicide Sentencing: Findings From Brazil

Abstract

Over the past four decades, hundreds of studies have examined the presence of racial disparities in criminal punishment. The bulk of this research has been conducted in the United States and a few other western democracies, with limited research assessing the presence of racial disparities in criminal sentencing for homicides in South America. Using information gathered via original data collection on homicide cases from five different capital cities in Brazil, the current study examines two criminal court outcomes: whether a defendant was convicted and the length of sentence. Findings reveal the absence of racial disparities in conviction decisions, even though Black and Brown defendants received longer sentence lengths. Supplementary analyses show racial disparities in sentence length are most pronounced when the homicide victim was White. Implications for studying court outcomes in international contexts are discussed.

Keywords

homicide criminology violent offenders

Introduction

Over the past four decades, hundreds of studies have examined racial disparities in criminal punishment (Baumer, 2013; Mitchell, 2005; Spohn, 2000; Zatz, 2000 ). This research is well-established in the United States, where an abundance of empirical studies on judicial decision-making has documented racial disparities in criminal sentencing, with non-White defendants receiving the harshest criminal court outcomes (see Baumer, 2013; Mitchell, 2005). However, the vast majority of the research in this area is limited as it has primarily utilized data from the United States and occasionally from other Western countries such as Canada, Australia, and Western European nations (Liem & Eisner, 2020). Indeed, in his review of sentencing research, Ulmer (2012, p. 31) noted that a “key gap in the literature is that almost all of the research on sentencing is limited to the contemporary North American—particularly the U.S.—context.”

The relatively limited literature on sentencing disparities outside of the Global North limits knowledge regarding the presence of racial disparities in criminal justice for several reasons (Franklin & Henry, 2020). First, it is unknown whether findings and theories of criminal punishment primarily generated from data in wealthy western democracies will generalize to other countries. Indeed, psychological research has found that studies conducted in Western, Educated, Industrialized, Rich, and Democratic (WEIRD) countries often do not generalize to non-WEIRD nations (Henrich et al., 2010). Second, most studies on criminal punishment use data from countries with relatively low crime rates (i.e., North America, Western Europe, Oceania) compared to regions with the highest levels of crime and violence (i.e., Latin America). Thus, knowledge about sentencing patterns of criminal offenders—especially homicide offenders—in contexts of high levels of interpersonal violence has gone understudied. This is a critical gap, especially considering research from US counties finds more severe sentencing for homicide offenses in areas where homicide rates are lower (Cooney & Burt, 2008). Finally, much of the research on sentencing disparities is conducted in multiracial contexts with deep-seated histories of institutional racism, such as the United States and Europe. However, it remains unclear whether race disparities in sentencing would be attenuated in countries with different histories of racial inequalities and greater fluidity in defining racial groups. On this point, sentencing scholars have remarked that “more cross-national and comparative research would greatly broaden knowledge of sentencing and sentencing disparity as related to larger patterns of social stratification” (Ulmer, 2012, p. 32).

This study aims to expand upon racial disparities in criminal punishment literature by examining judicial sentencing decisions for defendants charged with homicide in a “racial democracy”: Brazil. Specifically, we draw on novel data collected from homicide defendants in five different state capital cities in Brazil: Belem (Pará), Belo Horizonte (Minas Gerais), Goiania (Goias), Recife (Pernambuco), and Porto Alegre (Rio Grande do Sul) to assess the role of the race of the defendant and victim in sentencing outcomes. Considering the lack of quantitative research on sentencing patterns in Latin America, the profound issue that homicide poses in Brazil, and the unique context of race relations in Brazil, this study provides an opportunity to assess the presence of racial disparities in criminal sentencing processes.

Race, Homicide Sentencing, and the Brazilian Context

The main area of scholarship in criminal punishment since the 1970s has been racial disparities in sentencing decisions (Baumer, 2013). In short, this body of research focuses on the fact that minority groups (i.e., non-White persons) are often oppressed by the criminal justice system (Tonry, 1995). In speculating on the reasons for such disparities, the predominant theories center around the idea that judges typically have limited time and information when considering how to punish a defendant (Ulmer & Kramer, 1996). Because of these constraints, judges may resort to stereotypes attributed to demographic characteristics such as race when weighing critical concerns related to how blameworthy the defendant is, how likely the individual is to cause future harm to the community, as well as practical constraints such as the harms of a prison sentence on the individual or their family (Steffensmeier et al., 1998, 2017). In this context, it is theorized that the most oppressed racial and ethnic minority groups are viewed as more blameworthy for their offense, as posing a greater risk to the community, and as more capable of handling punishment, and thus will be more likely to be sent to prison and be sentenced to incarceration for more significant periods (Spohn & Holleran, 2000; Steffensmeier et al., 1998, 2017). In support of this proposition, a large body of research from the United States has found that minority defendants (i.e., Black and Hispanic persons) tend to receive more severe punishments net of legally relevant case characteristics than White persons (Baumer 2013; Kutateladze et al., 2016; Mitchell, 2005; Spohn, 2000; Zatz, 2000).

While an important finding, such research is predominantly limited to Western countries, and it is unclear if it generalizes to regions where there is a lack of research on racial disparities in sentencing, such as in Brazil (Bergman, 2018). A long-standing concept, which has permeated the Brazilian self-image, is that of a “racial democracy” (Campos de Sousa & Nascimento, 2008). In short, the idea of a racial democracy describes a context of harmony between the races, at times arguing for a de facto equality where any racial disparities are not a product of contemporaneous discrimination (De la Fuente, 1999). The concept of racial democracy is one piece of a broader debate regarding racial relations in Brazil, and about how these relations are exceptional to Brazilian society (Campos de Sousa & Nascimento, 2008). That debate is nearly a century old (Freyre, 1933), and arises largely out of historical and contemporary factors, namely a large extent of miscegenation or race mixture during early settlement periods that has created a racially diverse society (Campos de Sousa & Nascimento, 2008). Unlike other countries such as the United States and South Africa, Brazil was never able to approve race-specific laws or policies such as segregation or apartheid throughout the 20th century (Telles, 2002), although many elected officials have proposed segregation laws and even two penal codes—one for Blacks and another for Whites (Alvarez, 2002).

As Telles (2007, p. 1) notes “for these reasons, Brazilians thought of their country as a ‘racial democracy’ from as early as the 1930s until recent years. They believed that racism and racial discrimination were minimal or non-existent in Brazilian society in contrast to the other multiracial societies in the world.” Accordingly, at the core of the debate in Brazil is the question of whether blurred racial boundaries result in reduced racial discrimination (Marteleto & Dondero, 2016). Emerging research from across disciplines have provided evidence that relative to White, Black, and Brown individuals (i.e., the word used to identify mixed races, or the combination between Black and White) are worse off in terms of socioeconomic status (Cano et al., 2010), education (Botelho et al., 2015; Marteleto & Dondero, 2016), health (Pavao et al., 2012), infant birth outcomes (Nyarko et al., 2013), the labor market (Arcand & D’hombres, 2004), and even crime victimization (Truzzi et al., 2021).

In addition to the historical racial contexts, Brazil is a unique country within which to advance understanding of court processes and the determinants of sentencing as violent crime has traditionally been higher there than in Western countries, and as homicide rates have remained steadily high in the country even while rates have precipitously declined throughout much of the Western world since the 1990s (LaFree et al., 2015; Santos & Testa, 2018; Weiss et al., 2016). For instance, in 2018, the homicide rate in Brazil was 27.8 per 100,000, a rate over six times greater than that in the United States (Ribeiro & Diniz, 2020). Moreover, between 1990 and 2015, while the global homicide rate decreased by approximately 25%, the homicide rate in Brazil increased 32% during this same period (Santos & Testa, 2018; Truzzi et al., 2021). The high rate of homicide in Brazil has a significant impact on the population, as research has found that nearly 50 million Brazilians over the age of 16 (about one-third of the population) have lost friends or relatives to murder (Brazilian Forum of Public Safety, 2019). Racial disparities in homicide are also particularly pronounced, with homicide victimization rates for Black persons (37.8 homicides per 100,000) exceeding the rate of non-Blacks (13.9 homicides per 100,000 in 2018 [Truzzi et al., 2021; see also Murray et al., 2013]).

In addition, Brazil’s court system provides unique qualities that make it an important focus for researchers. The Penal Code of Brazil (1940) adopts an indeterminate sentencing structure, setting the maximum and minimum length that should be considered for each crime; but leaving a judge ultimately responsible for discerning the final sentence length. In addition, this criminal code is federal legislation and therefore regulates all sentences in Brazil. That scope, combined with the absence of specific guidelines for sentencing may be conducive to producing disparities in sentencing. Yet, the very existence of these disparities remains a question for research. Finally, courtroom actors in Brazil, including attorneys and judges, are civil servants. Similar to other countries, such as the United States, they are highly educated individuals who have completed formal training in law. However, their positions are obtained through competitive exams. These jobs are highly sought after as they entail prestige, high salaries, and job stability (Ribeiro et al., 2022).

Only a few studies have assessed sentencing disparities in the Brazilian context (Ribeiro & Diniz, 2020). This existing literature often uses data from only one major city and has yielded mixed results. Historical literature from the late 19th and early 20th centuries finds that Black individuals were more likely than White persons to receive a criminal conviction in São Paulo (Fausto, 1984) and Rio de Janeiro (Costa Ribeiro, 1995). Some contemporary research has provided evidence in line with these findings. In a study of murder cases in São Paulo between 1991 and 1998, Ribeiro (2010) found Black individuals had higher odds of criminal conviction than White individuals. However, in a sample of convictions of robbery, homicide, and drug trafficking, imposed on offenders in the cities of São Paulo and Rio de Janeiro in 2000 and 2001, Cano et al. (2010) found no evidence of racial differences in the sentence length. Similarly, Ribeiro & Diniz (2020) did not find racial differences in homicide convictions when assessing the cases tried in Belo Horizonte (the capital of the state of Minas Gerais) between 2003 and 2013.

Given the lack of sentencing research in Brazil and the mixed findings in the literature regarding the presence or absence of racial disparity in sentencing outcomes, the current study aims to examine the racial disparities in criminal punishment among a sample of homicide defendants across five cities in Brazil.

Data

Over two decades into the 21st century, no databases enable tracking criminal sentencing outcomes in Brazil (Ribeiro & Diniz, 2020). Accordingly, most research on criminal punishment in the country has focused on just one out of the two most populous cities in the country (i.e., São Paulo or Rio de Janeiro), limiting the generalizability of any findings (Costa Ribeiro, 1995; Fausto, 1984; Ribeiro, 2010). To fill this gap, the Ministry of Justice (Grant #30543) supported the collection of data to generate a dataset of a representative sample of homicide proceedings in 5 capital cities (out of 27), each from a distinctive administrative region. The cities selected were: Belém (capital city of Pará, northern region); Recife (capital city of Pernambuco, northeast region); Goiânia (capital city of Goiás, central-west region); Belo Horizonte (capital city of Minas Gerais, southeastern region), and Porto Alegre (capital city of Rio Grande do Sul, southern region).

The starting point of this research was the archive of the Court of Justice—which is responsible for retaining all documents related to the criminal proceeding—in the five cities selected for this study. Each state’s Court of Justice is responsible for archiving all documents produced by the police and the courts about each criminal case.¹ The research teams asked the Courts of Justice² to provide an excel spreadsheet with a list of all homicide cases filed in 2013 in each capital city.³ A statistician established the number of cases needed from each city to ensure a representative sample of all homicide cases in each city with adequate statistical power.

All criminal proceedings in Brazil encompass the documentation gathered in all previous stages of the criminal investigation and prosecution. Accordingly, the criminal proceedings include files with several documents, such as police records, police inquiry (with the indirect transcripts of witnesses and suspects hearing), the police indictment (the report from the head of the investigation), prosecutor’s charge, defense’s opposition to charges, indirect transcript of the evidentiary and judgment hearing, and the sentence. Fieldwork to extract this information from the court files involved the use of a structured observation instrument to gather systematized information from the existing documents, including the characteristics of victims, the context of the crime, and the police procedures deployed (such as arrest in a flagrant offense, eyewitnesses, and forensic analyses), and judicial decisions. These data were extracted and then cross-checked against the initial files to ensure accuracy.

Ultimately, a random sample of 786 cases was selected from a universe of 4,132 homicide files listed by the Courts of Justice. The following samples were produced from each city: 83 out of 281 in Belem (30%); 205 out of 1,363 in Belo Horizonte (15%); 119 out of 741 in Goiania (16%); 179 out of 793 in Porto Alegre (23%); and 200 out of 954 in Recife (21%). These sampling procedures were used to generate a representative sample of cases so that the findings could be generalized to the population of homicide cases within these cities for the year of data collection. Current analysis uses data on consummated intentional homicides, excluding homicide attempts (where the victim did not die). From 786 cases, 315 were consummated homicides and 471 were attempts. The focus of this study is on 315 consummated cases. Because at the time Brazil did not have a system of plea bargaining,⁴ consummated homicides cannot be pleaded down to a lower charge.

This study’s focus on intentional homicide offenses offers essential contributions to the criminal sentencing literature for a few reasons. First, while significant literature on criminal sentencing disparities in general, substantially fewer analyses have focused on murder cases (Johnson et al., 2010). Second, homicide offenders hold a unique moral and legal position across societies, with homicide often carrying the harshest sentences. Commenting on the special features of homicide offenses in the criminal justice system, Johnson et al. (2010, p. 983) have previously remarked, “homicide punishments engender broad moral and symbolic concerns in society, serve as a global barometer of national sentencing policy, and provide a useful analogy for the long-standing criminological tradition focusing on homicide offenders.” Finally, because of the high historical and contemporary levels of homicide in Brazil, this is a grave and widespread offense that warrants examination.

Dependent Variables

The analysis includes two dependent variables. First, a binary indicator is included for whether the outcome of the intentional homicide trial was an acquittal (0) or a conviction (1). Next, sentence length is a measure of the number of months a convicted individual is sentenced to incarceration. Regarding the sentence length, it is essential to note that Article 5 Section XLVII of the Brazilian Constitution states that there shall be no punishment: (a) of death, save in case of declared war; (b) of life imprisonment; (c) of hard labor; (d) of banishment; and (e) which is cruel. Moreover, though the Brazilian Criminal Code (1940) limits incarceration time served to a maximum of 30 continuous years, there is no limit for sentence lengths given upon conviction. For example, a convict who actually served 30 years in prison could have been sentenced to 50 years or more. Accordingly, sentence lengths are measured as the amount of time a defendant was sentenced to incarceration.⁵

Independent Variables

The independent variables measure the skin color of the defendant and victim. We used the race recorded by the police officer, reflecting the perception of this enforcement agent, which might be different from how offenders and victims perceive themselves in terms of skin color (Costa Ribeiro, 1995). In Brazil, since 1991, the Census has gathered information about race using a self-identification method divided into a five-category scheme: White (branco), Black (preto), Brown (pardo), Yellow (amarelo), and Indigenous (indígena).⁶ As the police record is available in all files produced by the police and later submitted to judges, defenders, and prosecutors, the race recorded by the police officer could guide the perception of these actors before they see the defendant in person. Because of a lack of cases in which the defendant or victim was defined as Yellow or Indigenous, these categories were not captured in the current study. Accordingly, Defendant’s skin color is classified as White (0), Brown (1), or Black (2). The victim’s skin color⁷ is similarly coded as White (0), Brown (1), or Black (2).⁸

Control Variables

Analyses accounted for several controls, including demographic characteristics of the defendant and victim, as well as focal legal factors which impact the odds of conviction and the sentence length. Demographic control variables include the sex of the defendant and victim (1 = male; 0 = female). Previous relationship is a binary indicator of whether there was a prior known relationship between the victim and defendant (1 = yes; 0 = no). Use of a firearm is a binary variable that denotes whether a firearm was used in the homicide (1 = yes; 0 = no). Multiple offenders are a binary variable indicating whether more than one offender was implicated in the homicide (1 = yes; 0 = no), and multiple victims are a variable identifying whether there was more than one victim (1 = yes; 0 = no). A categorical variable measures the type of legal counsel, including (0) no appointed representation—when the defendant is represented by whichever attorney is present at court, (1) a public defender who was assigned to the case, (2) counsel appointed before trial, and (3) a private attorney. Finally, we also controlled for several evidentiary and case characteristics, including the presence of a forensic examination (1 = yes, 0 = no),⁹ whether the defendant made a confession admitting to the crime (1 = yes; 0 = no), if there was eyewitness testimony during trial (1 = yes; 0 = no), and whether a prosecutor requested an acquittal from the jury (1 = yes; 0 = no).¹⁰

Analytic Strategy

Logistic regression analysis was used to predict the log odds of conviction for each defendant in the data on the indicator for the defendant’s skin color (β₁), the measure for the skin color of victims (β₂), in addition to all other controls (β_j).

\log [\frac{p}{1 - p}] = β_{0} + β_{1} Defendant Skin {Color}_{i} + β_{2} Victim Skin {Color}_{i} + β_{j} X_{i}

Next, sentence length is assessed using negative binomial regression models among the sub-sample of convicted defendants, focusing on the impact of the defendant’s skin color (β₁) and the skin color of victims (β₂), net of all other controls (β_j). This modeling strategy is justified by the fact that the sentence length is a count variable that is heavily right-skewed with evidence of overdispersion (Hilbe, 2011).¹¹

\log λ_{i} = β_{0} + β_{1} Defendant Skin {Color}_{i} + β_{2} Victim Skin {Color}_{i} + β_{j} X_{i} + σ ε_{i}

The coefficients across all models were converted to proportional effects using an exponential into odds ratios for the logit model and incident-risk ratios for the negative binomial models. We use predicted probabilities and counts to facilitate the visualization of our results.

Results

Table 1 contains the descriptive statistics of the analytic samples. About 43% of the 315 cases resulted in convictions. Among those 135 convictions, the average sentence length was 132.56 months (about 11 years), ranging from a minimum of 12 months to a maximum of 754.3 months.¹² Defendants were overwhelmingly males (96%) and victims (88%). Among the defendants, 33% were White, 49% were Brown, and 18% were Black. The skin color distribution of victims was similar, at 29% Whites, 53% Brown, and 18% Black.¹³

Table 1.

Descriptive Statistics.

Variables	Obs.	Mean	SD	Min	Max
Dependent variables
Conviction (binary)	315	0.43	0.50	0.00	1.00
Sentence length (months)	135	132.56	92.22	12.17	754.3
Defendant’s skin color
Defendant White (binary)	315	0.33	0.47	0.00	1.00
Defendant Brown (binary)	315	0.49	0.50	0.00	1.00
Defendant Black (binary)	315	0.18	0.39	0.00	1.00
Victims’ skin color
Defendant White (binary)	315	0.29	0.46	0.00	1.00
Defendant Brown (binary)	315	0.53	0.50	0.00	1.00
Defendant Black (binary)	315	0.18	0.38	0.00	1.00
Controls
Defendant male (binary)	315	0.96	0.19	0.00	1.00
Victim male (binary)	315	0.88	0.33	0.00	1.00
(Presence of) previous relationship (binary)	315	0.82	0.39	0.00	1.00
(Presence of) firearm (binary)	315	0.73	0.45	0.00	1.00
(Presence of) multiple offenders (binary)	315	0.39	0.49	0.00	1.00
(Presence of) multiple victims (binary)	315	0.15	0.35	0.00	1.00
Type of counsel
No appointment (0)	315	0.35	0.48	0.00	1.00
Public defender (1)	315	0.40	0.49	0.00	1.00
Appointed counsel (2)	315	0.05	0.21	0.00	1.00
Private attorney (3)	315	0.21	0.41	0.00	1.00
(Presence of) forensic analysis (binary)	315	0.14	0.34	0.00	1.00
(Presence of) defendant’s confession (binary)	315	0.34	0.47	0.00	1.00
(Presence of) eyewitness testimony (binary)	315	0.34	0.47	0.00	1.00
(Presence of prosecutorial) request for acquittal (binary)	315	0.16	0.37	0.00	1.00

Conviction

Table 2 displays the set of logit models predicting conviction. Models 1 through 4 use the same sample but gradually include more controls. As the results were consistent, we will focus on our interpretation of Model 4, which includes all control variables. Neither the skin color of the defendant nor the victim had a significant association with the chances of conviction. Generally, defendants were significantly more likely to be convicted when they were males (OR = 5.183; p < .05), had a previous relationship with the victim (OR = 1.956; p < .10), in cases with multiple victims (2.659; p < .05), when they relied on a private attorney (OR = 2.118; p < .05) relative to a public defender, and if the defendant confessed to the crime (OR = 2.656; p < .001). A request for acquittal by the prosecutor dramatically reduces the odds of conviction (OR = 0.0379; p < .01).

Table 2.

Logit Regression Model of Conviction.

Variables	Model 1	Model 2	Model 3	Model 4
Defendant’s skin color
Defendant Brown	0.817 (0.209)	0.865 (0.245)	0.844 (0.243)	1.043 (0.340)
Defendant Black	0.596 (0.203)	0.622 (0.216)	0.589 (0.208)	0.745 (0.291)
Victim’s skin color
Victim Brown		0.871 (0.253)	0.855 (0.252)	1.010 (0.339)
Victim Black		0.789 (0.280)	0.816 (0.294)	0.879 (0.349)
Controls
Defendant male			4.279^* (3.415)	5.183^** (4.313)
Victim male			0.512^* (0.182)	0.532 (0.224)
(Presence of) previous relationship				1.956^* (0.697)
(Presence of) firearm				1.034 (0.340)
(Presence of) multiple offenders				0.683 (0.201)
(Presence of) multiple victims				2.659^** (1.045)
Type of counsel (no appointment = 0)
Public defender				1.077 (0.338)
Appointed counsel				2.015 (1.269)
Private attorney				2.118^** (0.794)
(Presence of) forensic analysis				0.830 (0.322)
(Presence of) defendant’s confession				2.656^*** (0.831)
(Presence of) eyewitness testimony				0.794 (0.232)
(Presence of prosecutorial) request for acquittal				0.0379^*** (0.0286)
Constant	0.907 (0.179)	0.982 (0.229)	0.444 (0.376)	0.134^** (0.136)
Observations	315	315	315	315
Pseudo R²	0.00550	0.00659	0.0253	0.192

Note. Coefficients are in odds ratios. Standard errors are in parenthesis.

***

p < .001. **p < .05. *p < .10 (two-tailed).

Figure 1 illustrates the results of Model 4 in Table 2 by plotting the predicted probability of conviction for each defendant’s skin color. Though none of the probabilities in the plot are significantly different from each other, Blacks are 5.6% points less likely to be convicted (37.9% of cases) than similarly situated Whites (at 43.5%) and Browns (44.3%). While the influence of skin tone was not statistically significant, it is essential to note that the magnitude of the difference of nearly 6% points is sizeable, considering just 43% of the sample received a conviction. Because of the relatively small sample size compared to much of sentencing research in the United States that uses statewide or federal administrative datasets (Baumer, 2013), the lack of statistical significance is partly attributed to larger standard errors.

Figure 1.

Predicted probability of conviction by homicide defendant’s skin color.

Sentence Length

Table 3 contains the negative binomial models predicting the sentence length of the subset of convicted cases (n = 135).¹⁴ Controlling for all independent variables in Model 4, skin color significantly predicts sentence length. Brown defendants have sentences that are 27.1% (p < .05) longer than the sentences of White defendants, and Black defendants have sentence lengths 44.3% (p < .05) longer than White defendants. Regarding victim’s skin colors, in the partial Model 2, we initially find that offenses with Black victims have an expected sentence 33.7% (p < .05) shorter than the sentences for homicides with White victims. Some of that difference is explained by the legal characteristics of the offense, declining to a difference of 21.6% (p > .10), which is substantively relevant but statistically non-significant. As noted above, the magnitude of this effect is relatively sizeable; however, the larger standard error because of the small sample contributes to results that are not statistically significant. Finally, cases with multiple offenders increased the expected sentence length by 26.9% (p < .05).

Table 3.

Negative Binomial Model of Sentence Length.

Variables	Model 1	Model 2	Model 3	Model 4
Defendant’s skin color
Defendant Brown	1.214 (0.144)	1.252^* (0.155)	1.275^** (0.158)	1.271^** (0.151)
Defendant Black	1.613^*** (0.268)	1.641^*** (0.268)	1.640^*** (0.266)	1.443^** (0.232)
Victim’s skin color
Victim Brown		0.983 (0.125)	0.982 (0.124)	0.959 (0.115)
Victim Black		0.663^** (0.110)	0.678^** (0.116)	0.784 (0.130)
Controls
Defendant male			1.802 (0.837)	1.460 (0.646)
Victim male			0.909 (0.132)	0.804 (0.119)
(Presence of) previous relationship				0.892 (0.133)
(Presence of) firearm				1.174 (0.158)
(Presence of) multiple offenders				1.269^** (0.145)
(Presence of) Multiple victims				1.129 (0.162)
Type of counsel (no appointment = 0)
Public defender				1.056 (0.140)
Appointed counsel				1.222 (0.279)
Private attorney				0.970 (0.128)
(Presence of) forensic analysis				0.796 (0.120)
(Presence of) defendant’s confession				0.819 (0.100)
(Presence of) eyewitness testimony				0.861 (0.0987)
(Presence of prosecutorial) request for acquittal				0.595 (0.250)
Constant	0.389^*** (0.0458)	0.371^*** (0.0439)	0.366^*** (0.0433)	0.304^*** (0.0366)
Ln(alpha)	111.0^*** (9.903)	116.5^*** (12.23)	69.65^*** (35.03)	98.02^*** (51.99)
Observations	135	135	135	135
Pseudo R²	0.00556	0.0100	0.0113	0.0285

Note. Coefficients are in incidence-rate ratios. Standard errors are in parenthesis.

***

p < .001. **p < .05. *p < .10 (two-tailed).

Figure 2 illustrates the sentence lengths predicted by Model 4 for defendants of each skin color while holding all other variables at their respective means. White defendants have an average predicted sentence of 110.3 months, against 140.2 months for Brown defendants and 159.2 months for Black defendants.

Figure 2.

Predicted sentence length by defendant’s skin color.

Race of Defendant/Race of Victim Dyad Supplemental Analysis

As a supplementary analysis, we assessed the interactive relationship between the defendant’s and the victim’s skin color. Though informative, a caveat of these analyses is the low number of observations per intersection, which makes coefficients overly sensitive and may negatively impact the efficiency of estimates. Thus, results should be interpreted cautiously. The coefficients for each of the interactions are in the regression table in Supplemental Appendix A. Supplemental Appendix B displays plots with the predicted probability of conviction (panel A) and the predicted sentence length (panel B) for each combination between the skin color of defendants and victims. In the conviction model, Black defendants generally have a lower probability of conviction, particularly when victims are also Black (21.2%).

Regarding the predicted sentence lengths, Panel B shows much more evidence of racial disparities. Brown defendants have about the same predicted sentence lengths, regardless of the skin color of victims, at about 140 months. White defendants have lower predicted sentence lengths when victims are Black, at 71.6 months, compared to 112.8 months when the victim is White and 126.2 months when the victim is Brown. Importantly, all predicted sentence lengths for White defendants are lower than the predicted sentence lengths for defendants of other skin colors. Finally, Black defendants have lower sentence lengths when the victim is Black (129 months) or Brown (137.4 months) but have much higher predicted sentence lengths when the victim is White, at 210.3 months.

Discussion

The current study investigates the influence of race on sentencing outcomes in a sample of intentional homicide defendants from five major Brazilian capitals. We sought to understand the presence of racial disparities in Brazilian society—a high-violence country—and to examine the existence of racial disparities in sentencing outside of the United States. Overall, this study produced several findings that provide insights into the intricacies of homicide sentencing in Brazil and the role that race plays in that process.

It is essential to highlight that the conviction rate for homicide cases in Brazil is lower than in other countries (Liem & Eisner, 2020). The current study found the conviction rate to be only 43% for homicide cases. According to Brazilian scholars, this low conviction rate results from the jury trial’s features. Notably, criminal trials last just 1 day and do not involve cross-examination (Kant de Lima, 1995). Instead, the trial has been described as a theatrical dispute between the prosecutor and the defense regarding how the crime was committed and the defendant’s responsibility for the offense (Schritzmeyer, 2019). The outcome is based on votes of yes or no by the jurors, with jurors deciding how to vote on their own, with no deliberations occurring amongst jurors before a vote (Kant de Lima, 2004). For every question the judges present to the jurors, they remind them that any doubt should favor the defendant, increasing the odds of no votes, as there is little evidence of the defendants’ responsibilities (Schritzmeyer, 2019).

The core question of this study is whether blurred racial lines within a “Racial Democracy” of Brazil produce less racial discrimination (Marteleto & Dondero, 2016). Contemporary research has revealed that racial disparity does exist within institutions in Brazil, such as the economy (Arcand & D’hombres, 2004), education (Botelho et al., 2015; Marteleto & Dondero, 2016), and health outcomes (Nyarko et al., 2013; Pavao et al., 2012). Statistics also demonstrate the presence of racial disparities in criminal victimization (Murray et al., 2013; Truzzi et al., 2021), although existing studies have presented mixed results regarding how race influences the decision-making for homicide cases brought to the criminal justice system (Adorno, 1995; Cano et al., 2010; Costa Ribeiro, 1995; Fausto, 1984; Ribeiro, 2010; Ribeiro & Diniz, 2020). Adding to the above empirical findings, the current study finds no direct effects of race on jurors’ decisions about conviction but finds racial disparity in judicial decisions of sentence length. Although there were no significant differences in the probability of conviction by race, Black and Brown defendants were given longer average sentence lengths by judges compared to White defendants.

Regarding the probability of conviction, the negative, although non-significant, findings show that Blacks have smaller odds of conviction than either White or Brown defendants (37.9% for Blacks, 43.5% for Whites, and 44.3% for Browns). This result might reflect differential prosecutorial practices since prosecutors are more prone to bring forth criminal charges against Black defendants even when they have weaker cases or less evidence. Some research on prosecutorial decision-making practices in the United States have explained this phenomenon with similar ideas that police arrest non-White individuals more often on weaker evidence (Kutateladze et al., 2014; Petersilia, 1983). Providing some support for this possibility, ancillary analyses using our data show that relative to White defendants, Brown defendants receive acquittal requests at a 4.5 times higher rate and Black defendants at a 3.9 times higher rate. Furthermore, in Brazil, a case is considered cleared once it is charged by the prosecutor, meaning that all cases brought by the police to the courts have to be charged, and prosecutors have no discretion to drop them (Ribeiro & Diniz, 2020). Based on the results obtained for conviction models, we might argue that the process is the punishment for Black defendants: even though their cases are weaker, the prosecutor presents charges against them, and, at the final stage, they invite jurors to acquit the defendant. Confirmation of this pattern, however, would require analyzing other data related to prosecutorial practices in homicide cases.

Regarding sentence lengths, although White defendants have the lowest predicted sentences regardless of the victim’s race, the shortest sentences are in cases in which the defendant was White and the victim was Black, and the longest sentences are in cases in which the defendant was Black, and the victim was White. These interactions demonstrate that the victim’s race is also an important predictor of judicial sentencing practices in homicide cases but that outcomes are differentially impacted depending on the dyad between the race of the defendant and the victim’s race. These latter findings indicate a pattern of racial disparity in sentencing outcomes in Brazil, which is suggestive that net of a variety of factors related to the case, judges are more likely to sentence non-White defendants to longer periods of incarceration, especially in cases when the victim is White. Moreover, these findings expand upon prior research, which has assessed racial disparities in Brazil by focusing on a broader sample of cities and assessing the role of the victim’s race.

The current study also broadened the literature on the determinants of sentencing in international contexts. These data are unique because they were obtained through original data collection from court case files across five Brazilian cities. Future research should continue to explore whether findings regarding court outcomes from mostly Western rich and industrialized countries (WEIRD) are generalizable to other contexts (non-WEIRD). Indeed, our data proved to be a fruitful source to explore and uncover disparities according to the extra-legal defendant and victim characteristics that had not been done previously. Future studies should gather more information related to crimes other than homicide and collect data on demographic characteristics of judges and prosecutors to provide a further understanding of the nature of criminal case processing and to add to current knowledge of disparities and how they manifest in other social structural contexts. Finally, our results provide renewed evidence against the idea of a racial democracy in Brazil where de facto equality exists, and racial discrimination is minimal or non-existent.

Despite the study’s contributions, the findings should be interpreted in the context of limitations that can be expanded upon in future research. First, the sample in this study focuses on homicide defendants, and therefore findings cannot speak to racial disparities in other types of crime in Brazil. Second, the data come from five capital cities in Brazil and may not generalize to other cities outside of this context. Research that examines patterns of criminal sentencing, especially in rural contexts, would be especially usual as a future direction to expand knowledge of sentencing disparities in Brazil. Third, the measure of race was derived from police records. While these records are part of the trial and are provided to judges, we cannot confirm whether the race matches the defendant’s or victim’s racial classifications. Fourth, while the data provide a large amount of detail about the defendant and case characteristics, specific potentially relevant measures were unavailable, such as the socioeconomic status of the defendant, as well as characteristics of the prosecutor or judge. Fifth, because the sample size is small relative to other sentencing studies, at times, relatively meaningful differences in magnitude are not statistically significant because of larger standard errors. It would be helpful to invest further in collecting larger-scale data sets on criminal sentencing outcomes in Brazil and other countries in South America to generate more precise estimates of sentencing outcomes. Finally, it is essential to note that any findings in the current study should be considered associations rather than causal evidence of the influence of race on sentencing outcomes.

Supplemental Material

sj-docx-1-jiv-10.1177_08862605221135143 – Supplemental material for Assessing Racial Disparities in Homicide Sentencing: Findings From Brazil

Supplemental material, sj-docx-1-jiv-10.1177_08862605221135143 for Assessing Racial Disparities in Homicide Sentencing: Findings From Brazil by Alexander Testa, Mateus Rennó Santos, Ludmila Ribeiro and Richard Hartley in Journal of Interpersonal Violence

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interests with respect to the authorship and/or publication of this article.

Funding

The author(s) received no financial support for the research and/or authorship of this article.

ORCID iD

Alexander Testa

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biographies

Alexander Testa, PhD is an Assistant Professor in the Department of Management, Policy and Community Health at the University of Texas Health Science Center at Houston. His research focuses on the influence of criminal justice system contact and childhood adversity for health and development over the life course.

Mateus Rennó Santos, PhD is an Assistant Professor in the Department of Criminology at the University of South Florida. His research focuses on the drivers of crime and criminal justice trends, on the effects of policy, and on the application of quantitative methods and data science to research of crime and justice.

Ludmila Ribeiro, PhD is an Associate Professor in the Department of Sociology and a Researcher in the Center for Crime and Public Safety Studies, both at the Federal University of Minas Gerais. Her research interests include sentencing disparities, homicide trials, and the penitentiary system. Her most recent published research appears in Homicide Studies, Victims & Offenders, and Law & Society Review.

Richard Hartley, PhD is Professor and Chair in the Department of Criminology and Criminal Justice at the University of Texas at San Antonio. His research interests include contextual disparities in sentencing practices, prosecutorial and judicial discretion, and evaluation research. His most recent published research appears in Sexual Abuse, Criminal Justice Policy Review, and Trauma, Violence, & Abuse.

References

Adorno

(1995). Discriminação racial e justiça criminal em São Paulo. Novos Estudos CEBRAP, 43, 45–63.

Alvarez

M. C.

(2002). A criminologia no Brasil ou como tratar desigualmente os desiguais. Dados, 45(4), 677–704.

Arcand

J. L.

D’hombres

(2004). Racial discrimination in the Brazilian labour market: Wage, employment and segregation effects. Journal of International Development: The Journal of the Development Studies Association, 16(8), 1053–1066.

Baumer

E. P.

(2013). Reassessing and redirecting research on race and sentencing. Justice Quarterly, 30(2), 231–261. https://doi.org/10.1080/07418825.2012.682602

Bergman

(2018). More money, more crime: Prosperity and rising crime in Latin America. Oxford University Press.

Boateng

F. D.

Dzordzormenyoh

M. K.

(2021). Capital punishment in Brazil: Exploring factors that predict public support for the death penalty. Journal of Contemporary Criminal Justice, 38(1), 56–71. https://doi.org/10.1177/10439862211034345

Botelho

Madeira

R. A.

Rangel

M. A.

(2015). Racial discrimination in grading: Evidence from Brazil. American Economic Journal: Applied Economics, 7(4), 37–52.

Brazilian Forum of Public Safety. (2019). Anuário Brasileiro de Segurança Pública 2019. Brazilian Forum of Public Safety.

Campos de Sousa

Nascimento

(2008). Brazilian national identity at a crossroads: The myth of racial democracy and the development of black identity. International Journal of Politics, Culture, and Society, 19(3), 129–143.

10.

Cano

Ribeiro

Meireles

(2010). Race, crime and criminal justice in Brazil. In Kalunta-Crumpton

(Ed.), Race, crime and criminal justice (pp. 207–241). Palgrave Macmillan.

11.

Cooney

Burt

C. H.

(2008). Less crime, more punishment. American Journal of Sociology, 114(2), 491–527.

12.

Costa Ribeiro

C. A

. (1995). Cor e criminalidade: estudo e análise da justiça no Rio de Janeiro, 1900–1930. Editora UFRJ.

13.

De la Fuente

(1999). Myths of racial democracy: Cuba, 1900–1912. Latin American Research Review, 34(3), 39–73.

14.

Fausto

(1984). Crime e cotidiano: A criminalidade em São Paulo, 1880–1924. Edusp.

15.

Franklin

T. W.

Henry

T. K. S.

(2020). Racial disparities in federal sentencing outcomes: Clarifying the role of criminal history. Crime & Delinquency, 66(1), 3–32.

16.

Freyre

(1933). Casa-Grande & Senzala: Formação da Economia Brasileira sobre o Regimen de Economic Patriarchal [The masters and the slaves]. Maya & Schmidt.

17.

Henrich

Heine

S. J.

Norenzayan

(2010). The weirdest people in the world? Behavioral and Brain Sciences, 33(2–3), 61–83.

18.

Hilbe

J. M.

(2011). Negative binomial regression. Cambridge University Press.

19.

Johnson

B. D.

Betsinger

(2009). Punishing the “model minority”: Asian-American criminal sentencing outcomes in federal district courts. Criminology, 47(4), 1045–1090.

20.

Johnson

B. D.

Van Wingerden

Nieuwbeerta

(2010). Sentencing homicide offenders in the Netherlands: Offender, victim, and situational influences in criminal punishment. Criminology, 48(4), 981–1018.

21.

Kant de Lima

(1995). Bureaucratic rationality in Brazil and in the United States: Criminal justice systems. In Hess

Da Matta

(Eds.), The Brazilian puzzle: Culture on the borderlands of the Western world (pp. 241–269). Columbia University Press.

22.

Kant de Lima

(2004). Direitos civis e direitos humanos: Uma tradição judiciária pré-republicana? São Paulo Em Perspectiva, 18(1), 49–59.

23.

Kutateladze

B. L.

Andiloro

N. R.

Johnson

B. D.

Spohn

C. C.

(2014). Cumulative disadvantage: Examining racial and ethnic disparity in prosecution and sentencing. Criminology, 52(3), 514–551

24.

Kutateladze

B. L.

Andiloro

N. R.

Johnson

B. D.

(2016). Opening Pandora’s box: How does defendant race influence plea bargaining? Justice Quarterly, 33(3), 398–426.

25.

LaFree

Curtis

McDowall

(2015). How effective are our ‘better angels’? Assessing country-level declines in homicide since 1950. European Journal of Criminology, 12(4), 482–504.

26.

Liem

Eisner

(2020). From homicide to imprisonment: Mapping and understanding the flow of homicide cases. Homicide Studies, 24(3), 207–219.

27.

Lovell

P. A.

Wood

C. H.

(1998). Skin color, racial identity, and life chances in Brazil. Latin American Perspectives, 25(3), 90–109.

28.

Marteleto

L. J.

Dondero

(2016). Racial inequality in education in Brazil: A twins fixed-effects approach. Demography, 53(4), 1185–1205.

29.

Mitchell

(2005). A meta-analysis of race and sentencing research: Explaining the inconsistencies. Journal of Quantitative Criminology, 21(4), 439–466.

30.

Murray

Cerqueira

D. R.

Kahn

(2013). Crime and violence in Brazil: Systematic review of time trends, prevalence rates and risk factors. Aggression and Violent Behavior, 18(5), 471–483.

31.

Nyarko

K. A.

Lopez-Camelo

Castilla

E. E.

Wehby

G. L.

(2013). Explaining racial disparities in infant health in Brazil. American Journal of Public Health, 103(9), 1675–1684.

32.

Pavao

A. L. B.

Ploubidis

G. B.

Werneck

Campos

M. R.

(2012). Racial discrimination and health in Brazil: Evidence from a population-based survey. Ethnicity & Disease, 22(3), 353–359.

33.

The Penal Code of Brazil. Brazilian Criminal Code. (1940). https://presrepublica.jusbrasil.com.br/legislacao/91614/codigo-penal-decreto-lei-2848-40

34.

Petersilia

(1983). Racial disparities in the criminal justice system. Rand Corporation.

35.

Ribeiro

(2010). A produção decisória do sistema de justiça criminal para o crime de homicídio: Análise dos dados do estado de São Paulo entre 1991 e 1998. Dados, 53, 159–194.

36.

Ribeiro

Diniz

A. M.

(2020). The flow of murder cases through the criminal justice system in a Brazilian city. Homicide Studies, 24(3), 242–267.

37.

Ribeiro

Diniz

A. M. A.

Lages

L. B.

(2022). Decision-making in an inquisitorial system: Lessons from Brazil. Law & Society Review, 56(1), 101–121.

38.

Santos

M. R.

Testa

(2018). Global trends in homicide. In Deflem

(Ed.), Homicide and violent crime (Vol. 23, pp. 199–222). Emerald Publishing Limited.

39.

Schritzmeyer

A. L. P.

(2019). Jogo, ritual e teatro: Um estudo antropológico do Tribunal do Júri. Editora Terceiro Nome.

40.

Spohn

C. C.

(2000). Thirty years of sentencing reform: The quest for a racially neutral sentencing. Criminal Justice: The National Institute of Justice Journal, 3(1), 427–501.

41.

Spohn

Holleran

(2000). The imprisonment penalty paid by young, unemployed black and Hispanic male offenders. Criminology, 38(1), 281–306.

42.

Steffensmeier

Ulmer

Kramer

(1998). The interaction of race, gender, and age in criminal sentencing: The punishment cost of being young, black, and male. Criminology, 36(4), 763–798.

43.

Steffensmeier

Painter-Davis

Ulmer

(2017). Intersectionality of race, ethnicity, gender, and age on criminal punishment. Sociological Perspectives, 60(4), 810–833.

44.

Telles

E. E.

(2002). Racial ambiguity among the Brazilian population. Ethnic and Racial Studies, 25(3), 415–441.

45.

Telles

E. E.

(2007). Race in a miscegenation context: The Brazil experience. United Nations Chronicle, No. 3. https://www.un.org/en/chronicle/article/racial-discrimination-and-miscegenation-experience-brazil

46.

Tonry

(1995). Malign neglect: Race, crime, and punishment in America. Oxford University Press.

47.

Truzzi

Lirio

V. S.

Cerqueira

D. R.

Coelho

D. S.

Cardoso

L. C.

(2021). Racial democracy and black victimization in Brazil. Journal of Contemporary Criminal Justice, 38(18), 10439862211038448.

48.

Ulmer

J. T.

Kramer

J. H.

(1996). Court communities under sentencing guidelines: Dilemmas of formal rationality and sentencing disparity. Criminology, 34(3), 383–408.

49.

Ulmer

J. T.

(2012). Recent developments and new directions in sentencing research. Justice Quarterly, 29(1), 1–40. https://doi.org/10.1080/07418825.2011.624115

50.

Weiss

D. B.

Santos

M. R.

Testa

Kumar

(2016). The 1990s homicide decline: A western world or international phenomenon? A research note. Homicide Studies, 20(4), 321–334.

51.

Zatz

M. S.

(2000). The convergence of race, ethnicity, gender, and class on court decision making: Looking toward the 21st century. Criminal Justice: The National Institute of Justice Journal, 3(1), 503–552.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB