Group Threat,Racial/Ethnic Animus,and Punitiveness in Latin America: A Multilevel Analysis

The Group Threat Perspective

A preeminent theoretical framework that has informed much sociological and criminological inquiry is the group threat perspective. Building largely on the work of Blumer (1958) and Blalock (1967), this theory anticipates that large and/or growing minority racial group populations can pose perceived threats to members of the majority racial group. In particular, Blalock (1967) hypothesizes that, as the minority group population increases, social control measures will be mobilized by the majority group to confront political and economic competition (see also Bobo & Hutchings, 1996; Olzak, 1992). Responses to these threats can take several forms, including political discrimination, symbolic segregation, and the proliferation of “threat-oriented ideologies” (Blalock, 1967, p. 167), which characterize minority group members in negative terms. More recently, group threat has been extended to include dimensions of “criminal threat” (Liska & Chamlin, 1984; Liska et al., 1981), where members of minority racial groups may be identified as crime prone, dangerous, or threatening to public safety. As a result, a relatively large minority population can pose a perceived criminal threat that the majority group may view as warranting increased social control via the criminal justice system (CJS; Crawford et al., 1998; Liska, 1992).

In light of the group threat perspective, much research has examined the effects of racial context on a wide variety of criminal justice outcomes, especially in the United States. Indeed, relatively large Black populations at the city, county, and state levels have been shown to be associated with increased police force size (D’Alessio et al., 2005; Kent & Jacobs, 2005; Stults & Baumer, 2007), police expenditures (Jackson & Carroll, 1981), use of force by police (Jacobs & O’Brien, 1998), complaints of police brutality (Holmes, 2000; Smith & Holmes, 2003), arrests (Liska & Chamlin, 1984; Mosher, 2001; cf. Parker et al., 2005), and incarceration rates (Bridges & Crutchfield, 1988; Eitle et al., 2002; Greenberg & West, 2001; Jacobs & Carmichael, 2001). Additionally, county- and state-level percent Black has been linked to harsher criminal court sentencing outcomes and racial/ethnic disparities therein (Jacobs et al., 2005; Johnson et al., 2008; Ulmer & Johnson, 2004; Wang & Mears, 2010). Recently, this framework has been expanded to consider Hispanics, immigrants, and other social out-groups (Holmes et al., 2008), and prior research has found Hispanic and/or foreign-born population size to be positively associated with a wide range of social control outcomes as well (Feldmeyer et al., 2015; Jacobs & Carmichael, 2001; Wang & Mears, 2010).

While the research on the consequences of group threat for the exercise of formal social control has been enlightening, there are some limitations to this body of work. Recently, scholars have begun to explore the various “micro dimensions” of racial/ethnic threat, that is, the realization of threat at the individual perceptual level. However, few studies have examined how the racial/ethnic contexts in which individuals are embedded might strengthen the relationship between animus against racial/ethnic out-group members and punitive attitudes especially among Whites. Indeed, only a few studies have explored these and other multilevel dimensions of group threat (Chiricos et al., 2014; Soss et al., 2003), and only one has examined these relationships outside the United States (Ousey & Unnever, 2012). The goal of the present study is to advance this body of work by investigating the interactive effects of group threat, out-group animus, and individual-level race/ethnicity on punitive sentiments.

Prior Research on Group Threat and Punitiveness

The group threat perspective was originally proposed to explain the proliferation and application of justice system resources as well as other forms of social control more generally. However, scholars also have posited that perceptual dimensions of threat are highly salient for the mobilization of these social control mechanisms. Indeed, as Blalock (1967) himself argued, the threat perspective is rooted in the “assumption that individual goals, motives, and needs are major causal agents in social systems” (p. 28), and group threat research should strive to take these processes into account. Chiricos et al. (2001) similarly maintain that studies which emphasize the relevance of presumably threatening macro-level racial/ethnic contexts necessarily imply the existence of “micro-processes involving the lived experiences of individuals who are situated in those social circumstances” (p. 323). Therefore, in addition to examining macro-level criminal justice outcomes such as police force size, arrest rates, and sentencing severity, tests of the group threat framework should involve the investigation of threat-related attitudes and perceptions held at the individual level.

Much previous research has demonstrated that county or neighborhood racial/ethnic context can impact individuals’ punitive sentiments as anticipated by the group threat hypothesis. In their study of support for capital punishment in the United States, Baumer and colleagues’ (2003) analyses revealed that county-level percent Black was positively associated with punitiveness. King and Wheelock (2007) likewise found that increases in county-level minority population composition were predictive of support for harsher criminal punishments. Similarly, Hispanic threat also has been shown to affect attitudes toward strict border controls (Stupi et al., 2016) and general punitive sentiments (Welch et al., 2011). Using cross-national data collected in 27 European countries, Ousey and Unnever (2012) found that ethnic fractionalization was associated with support for increasing punishment severity. In contrast with this body of work, however, several other studies have documented that—even among White respondents only—racial/ethnic composition was not related to support for “get tough” juvenile justice policies (Pickett & Chiricos, 2012) or other punitive responses to crime (Costelloe et al., 2002).

Beyond the role that contextual-level racial/ethnic threat might play in informing punitive attitudes, much research evidence also suggests that, especially among Whites, support for punitive criminal justice policies can be influenced by individual-level perceived political and/or economic competition (King & Wheelock, 2007; Pickett, 2016; Quillian, 1995), perceived cultural threat (Buckler et al., 2009; Chiricos et al., 2014), the typification of members of social out-groups as criminals (Chiricos et al., 2004; Stupi et al., 2016; Unnever & Cullen, 2012), and general feelings of racial animus (Unnever & Cullen, 2010b). Indeed, these studies have shown that resentment against racial/ethnic minorities is independently associated with support for increased criminal justice spending (Barkan & Cohn, 2005), harsher criminal sentences (Baker et al., 2018; D. Johnson, 2001), punitive juvenile justice measures (Pickett & Chiricos, 2012), and other such policies (Peffley & Hurwitz, 1998, 2002). Most of this research has been conducted using survey data from the United States, though studies of Great Britain (Stansfield & Stone, 2018), France (Dambrun, 2007), Germany (Cochran & Piquero, 2011), Russia (Wheelock et al., 2011), other European countries (Ousey & Unnever, 2012; Unnever & Cullen, 2010a), and Israel (Pickett et al., 2014) have revealed similar patterns of findings.

While the group threat perspective anticipates that individual-level racial/ethnic intolerance can influence punitiveness, especially among Whites, it is theoretically plausible that minority group population size might further moderate this relationship. Indeed, a macro-level context in which racial/ethnic threat is more pronounced could provide the conditions under which the consequences of out-group animus for punitiveness might be amplified (Chiricos et al., 2014; Ousey & Unnever, 2012). Under such circumstances, this association between racial/ethnic prejudice and punitive sentiments might be further enhanced since it is in conditions that are characterized by a salient threat that the perceived need for social control responses among those who hold racial/ethnic animus is likely to be strongest. While empirical support for the social threat framework has been found in research on macro-level racial/ethnic context and individual-level racial/ethnic animus independently, only one study has examined the interactive influence of these potential sources of punitiveness within a multilevel framework. Using data on a national sample of Americans, Soss and colleagues (2003) observed that the effect of racial prejudice on support for the death penalty among Whites was strengthened by percent Black at the county level. The goal of the present study is to further explore this line of inquiry by investigating these interactive relationships in the context of Latin America.

Race/Ethnicity and Mano Dura in Latin America

Central and South American countries are a useful setting in which the effects of racial/ethnic threat and animus on punitive sentiments may be examined. The racial/ethnic landscape of this region is dominated by an intense mixing (i.e., mestizaje) of racial and ethnic groups, and, while Latin Americans identify themselves as belonging to specific racial/ethnic categories when prompted (e.g., White, Mestizo, Black, Mulatto, or Indigenous), most individuals are of some combination of European, African, and/or Indigenous descent (Telles & Garcia, 2013). Through the encouragement of White elites, this blending of identities has led to widespread acceptance of “the myth of multiracialism” (Nobles, 2005), that is, the notion that no social stratification, prejudice, or discrimination along the lines of race or ethnicity exists (Silva, 2016; Telles & Bailey, 2013; Wade, 1993). Nonetheless, extensive empirical evidence suggests that long-standing race-based inequalities with regard to political participation, economic success, health, and educational achievement persist in this region, with Whites having access to social advantages not available to Blacks and the Indigenous (Bailey et al., 2014; Dixon & Telles, 2017; Twine, 1998; Villarreal, 2010).

The racial and ethnic context of Latin America provides a unique setting in which to investigate the interactive relationship between macro- and microlevel dimensions of group threat as identified in prior research. It has long been the case that European ancestry and White skin are highly valued in Latin America, while Blacks, Mulattos, and the Indigenous are believed by many to be “inferior or degenerative” (Telles & Bailey, 2013, p. 1563; see also Dixon & Telles, 2017; Nobles, 2005). Indeed, some evidence suggests that Whites attribute an array of negative characteristics, including laziness and unintelligence, to minority racial/ethnic group members (Castillo et al., 2010; Uhlmann et al., 2002). Further, these groups are sometimes perceived as culturally dissimilar but also politically visible and sometimes as unfairly benefiting from government welfare assistance (Barrientos et al., 2008; Hooker, 2005; Lavinas, 2015). Finally, there are historically persistent stereotypes which characterize Blacks and the Indigenous as disproportionately likely to be involved in crime, and members of these groups have long been overrepresented in Latin American prison populations (Aguirre, 1998; Kalunta-Crumpton, 2012). In short, it may be reasonable to suggest that members of racial and ethnic minorities might be identified by Whites as posing political, economic, or criminal threats and, as a result, should be controlled via the criminal justice system (Lehmann et al., 2019).

In addition to the complexities surrounding racial and ethnic identities, Latin America recently has experienced a notable movement toward punitive forms of crime control. Fear of crime in this region is pervasive due to increases in violent crime and gang activity, and these anxieties have been fueled by sensationalized news reports about the crime problem (Dammert, 2012; Malone, 2010; Peetz, 2011). As a result, many Latin Americans support punitive and authoritarian crime control strategies, some of which have been implemented in Argentina, Brazil, Costa Rica, Chile, El Salvador, Mexico, and other countries (Hume, 2007; Nivette, 2016; Wolf, 2017). These mano dura (i.e., “iron fist”) policies have included a wide range of criminal justice responses including the constriction of offenders’ due process rights, zero tolerance policing strategies, and harsh criminal court sentences (Holland, 2013; Krause, 2014; Swanson, 2013). While some prior research has suggested that public support for these policies is linked to criminal victimization experiences, crime-related fears, and perceived institutional illegitimacy (Malone, 2010; Nivette, 2016; Singer et al., 2019), the correlates of punitive sentiments in contemporary Latin America have remained largely unexplored.

The Present Study

In summary, within the large body of extant research on punitiveness, several studies on international contexts have reported that racial resentment, animus, and negative crime-related stereotypes are positively associated with punitive attitudes (Cochran & Piquero, 2011; Dambrun, 2007; Ousey & Unnever, 2012; Stansfield & Stone, 2018; Unnever & Cullen, 2010a). However, none of these studies examined how this relationship might be magnified in contexts characterized by large minority populations as anticipated by the group threat perspective. While one study has explored these interrelationships using data from the United States (Soss et al., 2003), no cross-national research on these issues has yet been conducted. Given the complexities of race and ethnicity of Latin America and the important role that perceived political, economic, and criminal threat might play in Whites’ assessments of Black and Indigenous populations, the present research explores the direct and interactive influence of macro-level group threat, animus against racial/ethnic minorities, and respondents’ race/ethnicity on punitiveness. The topic of this study is particularly salient in light of recent trends in Latin American criminal justice policy that have involved the expansion of “get tough” mano dura crime control responses.

Thus, in the present research, we intend to address the following hypotheses.

Hypothesis 1: Anti-Indigenous animus will be positively associated with support for harsher punishments, and these effects will be stronger among White respondents than among other racial/ethnic groups.

Data and Methods

To address these hypotheses, we use data from the 2012 AmericasBarometer survey, which is maintained by the Latin American Public Opinion Project (LAPOP). ¹ The LAPOP has administered the AmericasBarometer survey to residents of North, Central, and South American and Caribbean countries biennially since 2004. Respondents are selected using a multistage cluster design that uses strata based on the sizes of municipalities as identified in the most recent national censuses. These municipalities are approximately equivalent to U.S. counties. This design produces probability samples that are representative of the noninstitutionalized voting-age populations of each municipality, thus allowing for the characteristics of the respondents to be aggregated to the municipal level. Researchers conduct face-to-face interviews with one individual per household and use quotas for age and sex to avoid sampling error due to differential nonresponse. The sampling design of this survey, as well as the wide range of political, economic, and criminal justice topics covered in the questionnaire items, makes these data particularly well suited to investigate this study’s hypotheses.

The 2012 AmericasBarometer survey was administered to 41,632 respondents in 26 countries. However, the questionnaires that were distributed varied somewhat by country, and the question capturing punitive attitudes that represent our dependent variable was not asked in Canada, the United States, and Suriname. Additionally, the two items used to measure racial and ethnic animus were not available in an additional 13 countries. Therefore, our study was restricted to 16,999 respondents surveyed in the following Latin American nations: Argentina, Belize, Bolivia, Brazil, Colombia, Costa Rica, Mexico, Nicaragua, Paraguay, and Uruguay. Municipalities with fewer than 10 respondents were removed to ensure that municipal-level characteristics could be reliably generated from individual-level data, resulting in a final sample of 16,782 cases nested within 633 municipalities.

Because there were data missing on some of the variables of interest, Little’s (1988) test was conducted, and the results indicated that the data are not missing completely at random. Therefore, the missing cases were imputed through multiple imputation using chained equations and 20 imputations (White et al., 2011). The descriptive statistics for the study variables from the first imputation are presented in Table 1. ²

OPEN IN VIEWER

Table 1.

Descriptive Statistics: Latin American Public Opinion Project (LAPOP) Americas Barometer Survey, 2012.

Variables	Percentage	Mean	SD	Minimum	Maximum
Dependent variable
Support for harsher punishments	48.51	—	—	0	1
Individual-level variables
Anti-Indigenous animus	3.85	—	—	0	1
Anti-Black animus	3.75	—	—	0	1
Mestizo	49.78	—	—	0	1
Indigenous	6.03	—	—	0	1
Black/Mulatto	8.76	—	—	0	1
Other	4.77	—	—	0	1
Male	49.82	—	—	0	1
Age	—	39.73	15.93	15	94
Education	—	9.02	4.44	0	18
Unemployed	4.99	—	—	0	1
Government assistance	21.06	—	—	0	1
Products and services	—	7.15	3.19	0	14
Personal economic insecurity	24.19	—	—	0	1
Marital status	70.61	—	—	0	1
Parental status	73.88	—	—	0	1
Political ideology	—	5.56	2.56	1	10
Religious salience	—	3.28	0.94	1	4
Religious attendance	—	2.98	1.40	1	5
Personal crime victimization	19.41	—	—	0	1
Vicarious crime victimization	20.13	—	—	0	1
Fear of crime	—	2.16	0.88	1	4
News media consumption	—	4.37	1.02	1	5
Trust in the CJS	—	3.84	1.77	1	7
Trust in the police	—	3.83	1.87	1	7
Trust in the judicial system	—	2.27	0.99	1	4
Municipal-level variables
Percent Indigenous	—	0.06	0.13	0	0.91
Percent Black/Mulatto	—	0.09	0.18	0	1.00
Unemployment rate	—	0.05	0.05	0	0.50
Percent government assisted	—	0.21	0.20	0	1.00
Community products/services	—	7.15	2.26	0.25	12.08
Political climate	—	5.56	0.87	2.08	9.06
Victimization rate	—	0.19	0.12	0	0.75
Community fear of crime	—	2.16	0.36	1.00	3.67
Community economic insecurity	—	0.24	0.12	0	0.83

Note. N (Level 1) = 16,782; N (Level 2) = 633. Descriptive statistics from the first imputation are presented. Country dummy variables are not displayed. CJS = criminal justice system.

Results

In light of our first hypothesis, the analyses begin with an assessment of the independent, direct effect of anti-Indigenous animus on support for harsher punishments, and we then assess the possibility of an interaction between this variable and respondents’ race/ethnicity. The findings from the multilevel logistic regression model examining these relationships are shown in Table 2. The unstandardized logistic regression coefficients are presented, and these can be exponentiated to obtain the odds ratios. The results from Model 1, which estimates the main effect of animus against the Indigenous on punitiveness net of the individual- and municipal-level covariates, indicate that respondents who hold anti-Indigenous animus have a 34.3% greater odds than those who do not of supporting harsher criminal punishments, and this effect is statistically significant (b = 0.295, Exp(b) = 1.343, p < .01). In contrast with our hypothesis, however, the results from Model 2, which includes four multiplicative interaction terms between this variable and race/ethnicity, indicate that the effect of anti-Indigenous animus on punitiveness is not significantly stronger among Whites than among respondents of other racial/ethnic groups.

OPEN IN VIEWER

Table 2.

Multilevel Logistic Regressions of Support for Harsher Punishments: Main and Individual-Level Interaction Effects.

Variables	Model 1	Model 2	Model 3	Model 4
Individual-level interactions
Anti-Indigenous animus × Mestizo	—	.049	—	—
Anti-Indigenous animus × Indigenous	—	−.554	—	—
Anti-Indigenous animus × Black/Mulatto	—	.058	—	—
Anti-Indigenous animus × Other	—	.303	—	—
Anti-Black animus × Mestizo	—	—	—	.133
Anti-Black animus × Indigenous	—	—	—	−.210
Anti-Black animus × Black/Mulatto	—	—	—	.128
Anti-Black animus × Other	—	—	—	.506
Individual-level variables
Anti-Indigenous animus	.295***	.273*	—	—
Anti-Black animus	—	—	.208**	.126
Mestizo	−.106**	−.108**	−.109**	−.114**
Indigenous	−.089	−.072	−.096	−.083
Black/Mulatto	.120	.117	.120	.116
Other	−.114	−.127	−.113	−.134
Male	.001	.001	.001	.001
Age	−.010****	−.010****	−.010****	−.010****
Education	−.078****	−.078****	−.078****	−.078****
Unemployed	.188**	.187**	.188**	.188**
Government assistance	.009	.008	.009	.009
Products and services	−.005	−.005	−.004	−.004
Personal economic insecurity	.069*	.069*	.069*	.069*
Marital status	−.108**	−.110**	−.109**	−.109**
Parental status	.079	.079	.078	.079
Political ideology	.022***	.022***	.022***	.022***
Religious salience	.014	.014	.014	.015
Religious attendance	−.047***	−.047***	−.047***	−.047***
Personal crime victimization	.122***	.121***	.122***	.122***
Vicarious crime victimization	−.023	−.022	−.019	−.019
Fear of crime	.111****	.111****	.112****	.113****
News media consumption	.004	.004	.004	.004
Trust in the CJS	−.039***	−.038***	−.039***	−.039***
Trust in the police	.011	.011	.011	.011
Trust in the judicial system	.020	.020	.021	.021
Level 1 intercept	−.003	−.004	.001	−.002
Municipal-level variables
Percent Indigenous	.421	.412	.412	.415
Percent Black/Mulatto	.183	.180	.183	.183
Unemployment rate	−.535	−.540	−.543	−.541
Percent government assisted	.432*	.433*	.429*	.427*
Community products/services	.006	.006	.005	.005
Political climate	.058*	.058*	.059*	.060*
Victimization rate	−.378	−.381	−.376	−.377
Community fear of crime	.098	.100	.100	.101
Community economic insecurity	−.648***	−.646***	−.648***	−.647***
Level 2 intercept	.243****	.243****	.243****	.242****

Note. N (Level 1) = 16,782; N (Level 2) = 633. Unstandardized logistic regression coefficients are presented. Country dummy variables are not displayed. CJS = criminal justice system.

*p < .10. **p < .05. ***p < .01. ****p < .001 (two-tailed).

Presented in Model 3 of Table 2 are the multilevel logistic regression estimates of the effect of anti-Black animus on punitiveness net of the controls as anticipated by our second hypothesis. These findings show that, as hypothesized, this variable is a robust predictor of support for harsher punishments, and respondents with anti-Black animus have a 23.1% greater odds than those without such animus of reporting favorable attitudes toward this form of punitive crime control (b = 0.208, Exp(b) = 1.231, p < .05). However, as indicated in Model 4 in which the multiplicative interaction terms between anti-Black animus and respondent race/ethnicity are included, the positive effect of this variable is not significantly stronger among Whites than respondents belonging to any of the other four race/ethnicity categories. Thus, our findings provide partial support for our second hypothesis. The effects of the control variables remain largely consistent across the four models shown in Table 2, and being unemployed, personal economic insecurity, political conservatism, personal crime victimization, fear of crime, and a municipal-level politically conservative climate are positively associated with support for harsher punishments. The covariates that are negatively associated with this outcome include being Mestizo, being older, having more education, being married, religious attendance, trust in the CJS, and municipal-level economic insecurity.

The next stage of the analyses tests our third hypothesis that expects that larger Indigenous populations at the municipal level will strengthen the independent effect of being White on support for harsher punishments as well as the interactive effects of race/ethnicity and holding anti-Indigenous animus. Presented in Model 1 of Table 3 is a multilevel logistic regression of punitive sentiments which includes cross-level interactions between percent Indigenous and respondents' race/ethnicity. In this and all models in Table 3, the full list of controls is included, but the coefficients of these variables are not displayed for the sake of parsimony. In the proceeding discussion, the unstandardized logistic regression coefficients are divided by 100 before exponentiating because this variable reflects a proportion and thus has maximum possible value of 1. As a result, a one-unit increase in this variable is not intuitively meaningful, and the estimated coefficients can appear abnormally large. Dividing the unstandardized logistic regression coefficient by 100 before exponentiating allows the odds ratios to be interpreted as predicted changes in the odds of being coded 1 on the dependent variable due to a 1% increase in the independent variable. This strategy is equivalent to multiplying all values on this variable by 100 before estimating the regression.

OPEN IN VIEWER

Table 3.

Multilevel Logistic Regressions of Support for Harsher Punishments: Cross-Level Interaction Effects.

Variables	Model 1	Model 2	Model 3	Model 4
Cross-level interactions
Percent Indigenous × Mestizo × Anti-Indigenous animus	—	−3.418	—	—
Percent Indigenous × Indigenous × Anti-Indigenous animus	—	−2.391	—	—
Percent Indigenous × Black/Mulatto × Anti-Indigenous animus	—	2.942	—	—
Percent Indigenous × Other × Anti-Indigenous animus	—	−7.163	—	—
Percent Indigenous × Anti-Indigenous animus	—	2.904	—	—
Percent Indigenous × Mestizo	−1.458*	−1.313*	—	—
Percent Indigenous × Indigenous	−0.995	−0.905	—	—
Percent Indigenous × Black/Mulatto	−2.443**	−2.444*	—	—
Percent Indigenous × Other	−1.868*	−1.709	—	—
Percent Black/Mulatto × Mestizo × Anti-Black animus	—	—	—	−6.842**
Percent Black/Mulatto × Indigenous × Anti-Black animus	—	—	—	2.461
Percent Black/Mulatto × Black/Mulatto × Anti-Black animus	—	—	—	−2.744
Percent Black/Mulatto × Other × Anti-Black animus	—	—	—	−13.663
Percent Black/Mulatto × Anti-Black animus	—	—	—	3.800
Percent Black/Mulatto × Mestizo	—	—	0.261	0.439
Percent Black/Mulatto × Indigenous	—	—	−0.183	−0.272
Percent Black/Mulatto × Black/Mulatto	—	—	−0.136	−0.109
Percent Black/Mulatto × Other	—	—	0.041	0.254
Individual-level interactions
Anti-Indigenous animus × Mestizo	—	0.156	—	—
Anti-Indigenous animus × Indigenous	—	−0.520	—	—
Anti-Indigenous animus × Black/Mulatto	—	−0.006	—	—
Anti-Indigenous animus × Other	—	0.483	—	—
Anti-Black animus × Mestizo	—	—	—	0.438*
Anti-Black animus × Indigenous	—	—	—	−0.259
Anti-Black animus × Black/Mulatto	—	—	—	0.136
Anti-Black animus × Other	—	—	—	1.315*
Individual-level variables
Anti-Indigenous animus	0.294***	0.193	—	—
Anti-Black animus	—	—	0.208**	−0.076
Mestizo	−0.070	−0.075	−0.118**	−0.131**
Indigenous	−0.146	−0.124	−0.090	−0.079
Black	0.186**	0.182**	0.165	0.163
Other	−0.053	−0.070	−0.120	−0.156
Municipal-level variables
Percent Indigenous	1.652**	1.530**	0.401	0.412
Percent Black/Mulatto	0.155	0.156	0.234	0.194

Note. N (Level 1) = 16,782; N (Level 2) = 633. Unstandardized logistic regression coefficients are presented. Models include the full list of control variables.

*p < .10. **p < .05. ***p < .01. ****p < .001 (two-tailed).

As in the models shown in Table 2, White is used as the reference category, and the main effect of percent Indigenous in this model can be interpreted as the effect of this variable on support for harsher punishments among White respondents. This coefficient is positive and statistically significant, and among Whites, a 1% increase in percent Indigenous is associated with a 1.6% increase in the odds of supporting harsher punishments (b = 1.652, Exp(b/100) = 1.016, p < .05). Further, the interaction terms indicate that the effect of being White relative to other groups varies by municipal-level percent Indigenous, with increases in the latter variable associated with a weakening of the effects of being Mestizo (b = −1.458, Exp(b/100) = 0.984, p < .10), Black/Mulatto (b = −2.443, Exp(b/100) = 0.975, p < .05), or Other (b = −1.868, Exp(b/100) = 0.981, p < .10). In this model, anti-Indigenous animus remains positively associated with support for harsher punishments (b = 0.294, Exp(b) = 1.341, p < .01).

Shown in Model 2 of Table 3 are the three-way cross-level interactions respondent race/ethnicity, anti-Indigenous animus, and percent Indigenous. Hypothesis 3b expects that municipal-level percent Indigenous will condition the interactive association between being White and holding out-group animus on punitive sentiments. However, the findings from this model indicate no such relationship. Indeed, the evidence does not suggest that the positive effect of anti-Indigenous animus on punitiveness is conditioned by both percent Indigenous and race/ethnicity, nor is the interactive relationship between percent Indigenous and race/ethnicity observed in Model 1 further conditioned by anti-Indigenous sentiments. Thus, in sum, these findings provide partial support for our third hypothesis.

The last stage of the analyses tests Hypotheses 4a and 4b and involves an examination of the potential interactive relationship between municipal-level percent Black/Mulatto, respondent race/ethnicity, and anti-Black animus. The findings from Model 3, which estimates a multilevel logistic regression model predicting support for harsher punishments which includes multiplicative interaction terms between race/ethnicity and percent Black/Mulatto, provides no support for this theoretically anticipated relationship. However, as in Model 3 of Table 2, anti-Black animus has a significant and positive effect on punitive sentiments (b = 0.208, Exp(b) = 1.231, p < .05). In Model 4 of Table 3, however, a different pattern of findings emerges, and there is some evidence of a three-way interaction between municipal-level race/ethnicity, anti-Black animus, and percent Black/Mulatto. Indeed, the negative three-way interactive relationship between percent Black/Mulatto, Mestizo, and anti-Black animus suggests that municipal-level racial context strengthens the interactive association between being White (relative to Mestizo) and anti-Black animus (b = −6.842, Exp(b/100) = 0.933, p < .05). Interestingly, the interaction term for percent Black/Mulatto, anti-Black animus, and Indigenous is positive though not statistically significant. In summary, then, we find some support for our fourth hypothesis, with the effects of being White and holding animus against Blacks on punitiveness emerging particularly notably in municipal-level contexts with greater proportions of Black/Mulatto residents. ⁵

Finally, a set of supplementary analyses were conducted in which respondents who reported that they support both harsher punishments for criminals and the implementation of preventative measures were coded as 1 instead of 0 on the outcome measure. Multilevel logistic regression models were estimated, and these findings show that both anti-Indigenous animus (b = 0.168, Exp(b) = 1.183, p < .10) and anti-Black animus (b = 0.164, Exp(b) = 1.178, p < .10) are positively associated with support for harsher punishments, but these effects are weaker and only marginally statistically significant. Further, none of the previously observed cross-level interactions emerged. Thus, these results suggest that respondents who indicated a preference for harsher criminal punishments only are distinct from those who supported both crime control strategies with regard to the relevance of out-group animus and group threat.

Discussion and Conclusion

A dominant focus of much recent criminological research has been testing the precepts of the group threat perspective (Blalock, 1967), which anticipates that community-level minority group composition can pose political, economic, and criminal threats to the majority racial group that can be responded to through the exercise of objective social control (Crawford et al., 1998; Liska, 1992). Much prior research also has demonstrated that, in the United States, contextual-level Black and Hispanic population size is positively associated with support for various punitive crime control measures among members of the public and especially among Whites (Baumer et al., 2003; King & Wheelock, 2007; Stupi et al., 2016). However, net of racial/ethnic context, a large body of work has uncovered evidence that, particularly for Whites, individual-level perceptions of political, economic, or cultural threat, the typification of minorities as criminals, and animus against social out-groups are consistently predictive of punitive sentiments (e.g., Barkan & Cohn, 2005; Chiricos et al., 2004; D. Johnson, 2001; Pickett, 2016; Unnever & Cullen, 2010b, 2012), and this relationship has been documented in international contexts as well (Cochran & Piquero, 2011; Ousey & Unnever, 2012; Unnever & Cullen, 2010a).

Despite these important advancements to the literature connecting group threat and punitive sentiments, notable gaps in this body of research remain. Specifically, very little work on public punitiveness has examined possible interactions between macro-level racial/ethnic context and Whites’ individual-level racial attitudes (Soss et al., 2003), and this research has focused exclusively on the United States. Such a relationship is strongly theoretically expected, as embeddedness in a macro-level context that is characterized by a large minority presence is apt to strengthen feelings of animosity and resentment toward minority groups and amplify the extent to which these groups are considered to be politically, economically, culturally, or criminally threatening and in need of social control. No prior study has investigated these issues in Latin America, which is surprising given the region’s complex history of racial and ethnic conflict (Nobles, 2005; Telles & Bailey, 2013), the existence of long-standing social disadvantages experienced by Blacks and the Indigenous (Kalunta-Crumpton, 2012; Silva, 2016), and recent publicly supported movements embracing punitive crime control strategies (Holland, 2013; Nivette, 2016; Swanson, 2013; Wolf, 2017). Making use of survey data collected in 633 municipalities within 10 Latin American countries, the present research is the first study to test these tenets of the group threat perspective using a multilevel framework within this theoretically relevant context. Several key findings emerged from the analyses.

First, the present study shows that, after accounting for an extensive list of individual- and community-level control variables, animus against the Indigenous and Blacks is positively associated with punitive sentiments, and these effects are similar across racial and ethnic groups. This relationship has been well-documented in prior studies of the United States, and racialized conceptions of crime, resentment arising from perceived political and economic competition with minority groups, and general feelings of anger or bitterness toward minorities have been linked to increased support for a wide range of punitive criminal justice policies (e.g., Baker et al., 2018; Chiricos et al., 2004; King & Wheelock, 2007; Unnever & Cullen, 2012). This finding also corresponds with the results from prior work examining a variety of international contexts, including Canada and European countries (Cochran & Piquero, 2011; Dambrun, 2007; Ousey & Unnever, 2012; Stansfield & Stone, 2018), Russia (Wheelock et al., 2011), and Israel (Pickett et al., 2014). Indeed, this empirical consistency has led Unnever and Cullen (2010a) to theorize that “racial-ethnic intolerance as a source of punitiveness is a cultural universal in countries with conflicted minority relations” (p. 849). The results from this study provide further evidence of this relationship and suggest that Latin Americans might perceive members of minority groups to be political, economic, cultural, or criminal threats and deserving of social control (Aguirre, 1998; Kalunta-Crumpton, 2012; Lehmann et al., 2019). That the direct effects of out-group animus on punitiveness are not conditioned by respondents’ race/ethnicity suggests that, unlike in the United States, this relationship is not necessarily exclusive to Whites.

The second key finding from this study is that, as suggested by the cross-level interaction analyses, Whites’ punitive sentiments are moderated by municipal-level racial/ethnic context. Indeed, while we observe that White respondents are generally not more punitive than members of other racial/ethnic groups, Whites demonstrate increased punitive sentiments in municipalities with relatively high concentrations of Indigenous residents. Similarly, punitive sentiments among White respondents who hold animus against Blacks also are heightened in municipal-level contexts characterized by larger Black/Mulatto compositions. These findings are consistent with the group threat framework (Blalock, 1967; Liska, 1992) and its more recent developments (King & Wheelock, 2007; Ousey & Unnever, 2012; Stupi et al., 2016), which anticipate that minority population size will aggravate racial animus among members of the majority group and, as a result, will lead to an increased desire for punitive social control. Indeed, White respondents in Latin American communities with higher minority population concentrations may perceive members of these groups to be particularly politically, economically, culturally, or criminally threatening, and, as a result, they may be more inclined to support harsher criminal punishments as a means of responding to these threats (Chiricos et al., 2014; Soss et al., 2003; Welch et al., 2011). Further, the finding from this research that an interactive relationship exists between Whites’ anti-Black animus and Black/Mulatto population size points to the relevance of macro-level factors for informing, aggravating, and reinforcing antiminority attitudes and thereby strengthening the already robust effects that these sentiments have on Whites’ support for punitive criminal justice policies.

Finally, the results from these analyses emphasize the global relevance of the group threat perspective for understanding public attitudes. Indeed, not only does this research lend further credence to Unnever and Cullen’s (2010a) “cultural universal” thesis, but also the finding that antiminority attitudes, particularly among members of the privileged racial group, are conditioned by community-level population demographic characteristics highlights the broad applicability of this framework to diverse social conditions. Not only might out-group animus increase support for social control, but in contexts of racial and ethnic fractionalization, conflict, and historically persistent social inequalities a relatively large minority population size might represent political or economic competition (King & Wheelock, 2007; Ousey & Unnever, 2012; Quillian, 1995) or symbolize a viable criminal threat to community safety (Baumer et al., 2003; Stupi et al., 2016). Thus, the empirical support for the group threat perspective provided by this study, which engages data from the largely unexamined context of Latin America, highlights the profound importance of these social forces for fostering punitive sentiments and, in so doing, reifies this theory’s far-reaching relevance for criminological research.

Although this study provides additional insights into the sources of public punitiveness in an understudied setting, it is not without limitations. First, future research should strive to expand upon the survey measure used as the dependent variable in these analyses through the inclusion of additional items capturing support for specific criminal justice policies or strategies. For instance, it is unclear from the findings presented in this study whether the nuanced patterns of public attitudes observed in the United States might also be replicated in Latin America, such as simultaneous support for punitive and reintegrative criminal sanctions, pragmatic considerations of the economic consequences of correctional policies, and preferences regarding the differential punishment of violent versus nonviolent offenders (Cullen et al., 2000; Unnever et al., 2010). Second, our measure of antiminority animus is likewise limited since it relates only to explicit racial/ethnic resentment and does not capture attributions of negative characteristics to members of minority groups or perceptions that they pose political, economic, or criminal threats (Chiricos et al., 2004; King & Wheelock, 2007; Stupi et al., 2016; Unnever & Cullen, 2012). Finally, due to the cross-sectional design of the study, we were unable to capture dynamic group threat, that is, changes in municipal-level minority population composition over time, which some prior research has emphasized as more important for social control outcomes than static indicators of threat (e.g., Chiricos et al., 2014; Feldmeyer et al., 2015; King & Wheelock, 2007).

In summary, the present study analyzed survey data collected in the understudied context of Latin American nations, which are characterized by notable racial and ethnic inequality, intergroup conflict and resentment, and the recent adoption of punitive crime control policies. In these analyses, we report that animus directed against racial/ethnic minorities is consistently predictive of punitive sentiments, and Whites’ support for harsher criminal punishments is conditioned by a heightened community-level minority presence. These results, therefore, provide support for the group threat perspective and underscore similar observations made in previous research from the United States that antiminority animus, race/ethnicity, and macro-level racial/ethnic context interact to influence punitive attitudes. More specifically, these findings raise concerns regarding the effects of racial attitudes on support for mano dura criminal justice responses in Latin America, and they suggest that anti-Indigenous and anti-Black sentiments might contribute to this shift in public opinion. Future criminological inquiry should strive to investigate these and other complex relationships among the social sources of punitiveness in this region as well as the consequences of these attitudes for the implementation and legitimation of “get tough” crime control measures.