Political Elites,Crime,and Trust in the Police in Latin America

Abstract

This article examines the effect of crime on support for criminal justice systems in Latin America. Scholars empirically demonstrate a strong negative effect of crime on support for institutions and satisfaction with democracy. Others provide thick descriptions of the prevalence of creeping authoritarianism in response to crime—the infamous “mano dura” or “iron fist.” I test the effectiveness of elite political messaging across different countries. In other words, do politicians that promote “iron-fist” policies reassure their intended audience and shore up support for the police and the criminal justice system? Analyzing survey data from the Latin American Public Opinion Project and Wiesehomeier and Benoit’s expert survey of Latin American political party platforms, I find that elite political opinion about insecurity conditions the effect of crime victimization and fear of crime on mass support for the police and the justice system as well as on perceptions of police and justice system effectiveness. When political elites emphasize mano dura (“iron fist”) solutions, fearful citizens and victims are less critical of the police and the justice system in general.

Keywords

crime elites Latin America police

As violent crime rates and impunity continue to be unacceptably high in many parts of Latin America, what effect does this have on citizen support for democratic institutions? Politicians in the region often respond to concerns about crime with promises of tough and swift action, but in relatively few cases have there been what can be described as successful implementations of crime fighting policy; see Cruz (2011) and Malone (2012) for exceptions. While there is strong evidence that crime victimization and fear of crime undermine citizen support for institutions and satisfaction with democracy (Booth & Seligson, 2009; Carreras, 2013; Ceobanu, Wood, & Ribeiro, 2011; Cruz, 2015; Fernandez & Kuenzi, 2010), I analyze the extent to which this varies over different political contexts, expanding on previous work related to Central America (Cruz, 2011; Malone, 2010, 2012). By examining the conditional effect that crime-induced fear and anxiety has in different cultures of elite political discourse, I find that politicians can reduce citizens’ negative assessment of the police and the larger criminal justice system (CJS), which may have important implications for the citizens’ commitment to a democratic regime (Sarsfield & Echegaray, 2005).

On average, citizens victimized by crime have low levels of what Easton (1965, 1975) calls “specific” support for politicians and local government performance, and this may extend to diffuse levels of support as well (Carreras, 2013). The inclusion of political elites to understand how citizens develop their attitudes toward the CJS improves understanding of the extent to which there is meaningful variation across national contexts. I find that in areas where political elites are promoting more “iron-fist” crime control policies, citizens anxious about crime are less critical of their criminal justice institutions, controlling for other factors. Though perhaps normatively disappointing, this finding implies that politically speaking, elites are wise to promote iron-fist policies in the face of high crime to mitigate declining trust in police and the larger justice system. An important caveat, however, is that in the process of weakening this tide of criticism of the courts, the police, and other institutions, elites may also be increasing preferences for practices that often violate civil liberties, at least in their implementation (Garzón, 2015).

Using insights from the elite framing and political psychology literature on affective intelligence, I describe how crime victimization heightens the exposure of citizens to political information and consequently their propensity to be influenced by elite messaging. Since citizens receive different elite cues, I argue that elite messaging conditions the effect of crime-induced anxiety on trust in justice institutions. This is an important phenomenon deserving of inclusion in our explanations of how crime-induced anxiety affects institutional trust.

Background: Violent Crime in Latin America

The issue of violent crime in Latin America is of increasing interest to social scientists (Bateson, 2012; Carreras, 2013; Ceobanu et al., 2011; Cruz, 2015; Fernandez & Kuenzi, 2010; Krause, 2014; Malone, 2010, 2012, 2013; Merolla, Mezini, & Zechmeister, 2013). According to the Latin Barometer surveys covering all 18 Latin American democracies, citizens often rank insecurity among the worst problems in their respective countries (Lagos & Dammet, 2012). Homicide statistics from the World Health Organization indicate that the region as a whole has a murder rate 3–4 times that of the United States. Among the worst cases according to the United Nations Office on Crime and Drugs are Honduras and Venezuela, each with murder rates over 50 per 100,000 (United Nations, 2014). More recent estimates give El Salvador this dubious distinction as well (see http://www.insightcrime.org/news-analysis/insight-crime-2016-homicide-round-up). Crimes like kidnapping and assault are also increasing throughout the region, with street crime being both commonplace and seldomly reported (Bergman, 2006; Bergman & Whitehead, 2009).

Police Reform and Institutional Trust in Latin America

Latin America is not only the region with one of the highest rates of violence but also has among the lowest levels of trust in its police forces. The connection between these two is no doubt mutually reinforcing. On a scale of 0–100, citizens in 11 of Latin America’s 18 democracies place their level of trust in the police below 50¹. Citizens are the least trusting in Honduras (28.7), Guatemala (35.2), and Venezuela (37.6), which also has among the highest murder rates in the world. In eight Latin American countries, over half of the citizens believe that the police are also regular participants in crime. High levels of crime and low levels of trust are common despite years of attempted reform in the region. This includes organizational improvements through streamlined hierarchies, devolution of authority, and increased professionalism as well as efforts at speeding up criminal processing through each stage of the justice system and expanding international cooperation (Johnson, Forman, & Bliss, 2012; Ungar, 2011). Many areas of the region have also expanded community-based policing (Fruhling, 2012; Ungar, 2011). Nevertheless, numerous challenges to these reforms have greatly undermined their successful implementation.

A major obstacle to reform is political. There are often strong calls to take a zero tolerance or iron-fist approach to crime, even if the evidence for its effectiveness is weak (Swanson, 2013). When more moderate and sophisticated approaches are formally adopted, cooperation from the police and other agents necessary for implementation is often slow coming if not largely resisted and undermined (Ungar, 2011). There are also significant legal and institutional obstacles, as new edicts often lack the funding for implementation and contradict older laws that greatly empower police. The police themselves are often poorly paid and receive little training. This creates weak incentives to implement sometimes complicated reforms and strengthens incentives to engage in corruption (Bergman & Whitehead, 2009; Ungar, 2011).

There have nevertheless been a number of successful, if isolated, police reforms in Latin America, and it is important to keep in mind that crime in countries like Costa Rica, Chile, Uruguay, Panama, and Nicaragua is relatively low and trust in the police tends to be higher. Furthermore, even in countries that continue to suffer high rates of violent crime, some progress can be highlighted. This includes cooperative strategies with social welfare programs like Fica Vivo in Brazil, expanding surveillance and geographic technology in Colombia, increasing the number and concentration of police in high-risk neighborhoods via Barrio Seguro in the Dominican Republic, and reducing violence against women by working with nongovernmental organizations to provide education, victim services, needs assessment, and leadership training in Bolivia (Muggah, Szabo de Carvalho, Alvarado, Marmolejo, & Wang, 2016).

Despite some successes, many attempts at implementing reforms are thwarted by lack of resources, insufficient training, corruption, and poor communication with the public (Wazed & Akhtar, 2015). Mano dura policies have a long history in the region and still enjoy fairly widespread support. Rare efforts at improving upon them—such as in Central America—are often not given the commitment necessary to succeed (Schuberth, 2016).

Trust in the Police and CJS

Decades of research examine the causes and consequences of confidence and satisfaction in the police and the CJS in the United States, and more recently, this agenda has been greatly expanded in Latin America as well. These studies explore both individual-level demographic factors as well as neighborhood and other structural influences. At the individual level, important predictors of trust in the police tend to be age (Reisig & Parks, 2000; Weitzer & Tuch, 2002, 2005), gender (Gabbidon & Higgins, 2009; Ivkovic, 2008; Stack & Cao, 1998), and race and ethnicity (Sampson & Bartusch, 1998; Schuck, Rosenbaum, & Hawkins, 2008; Weitzer & Tuck, 2005, 2006). Older citizens belonging to racial/ethnic majority groups tend to have higher assessments of the CJS and the police on average. Findings for gender are somewhat inconsistent, however, as are those for education (Cao & Zhao, 2005; Gabbidon & Higgins, 2009; Stack & Cao, 1998) and economic status (Sampson & Bartusch, 1998; Weitzer, 2000). In the Latin American context, Sabet (2012) finds that income has no effect on satisfaction with the Mexican police. He also reports that satisfaction decreases with education and increases with age. In the Central American region, Malone (2010) shows no consistent effects of age, gender, education, and income on trust in the justice system. Looking at the region as a whole, Ahmad, Hubickey, and McNamara (2011) find that age increases trust, while secondary and tertiary education decrease it. Gender has no effect. Cao and Zhao (2005) indicate that age also has no effect, while income, social class, and education all correlate with decreasing trust in the police.

Other individual factors predictive of trust in the CJS in Latin America are victim status, fear of crime, and police bribery. Those who identify as having been a victim of crime in the past year report lower levels of trust in the police and institutions in general (Ahmad, Hubickey, & McNamara, 2011; Carreras, 2013; Romero, Magaloni, & Diaz-Cayeros, 2016). Fear of crime in one’s neighborhood produces similar results (Ahmad et al., 2011; Boateng, 2016; Carreras, 2013; Malone, 2010). Another strong predictor of lower trust is the experience of having been approached for a bribe by a police officer (Ivkovic, 2008), which is a common phenomenon in Latin America (Cruz, 2015; Sabet, 2012).

Finally, several contextual factors can influence citizen trust and satisfaction regarding the police and the justice system. Crime rates in particular are found to have a negative effect both in the United States and international contexts (Carreras, 2013; Jang, Joo, & Zhao, 2010; Reisig & Parks, 2000; Sampson, Raudenbush, & Earls, 1997). Relevant to Latin America, democratic stability and institutional strength can also have an effect (Booth & Seligson, 2009; Diamond, 1999; Jang et al., 2010; Malone, 2010), which may partly explain why police in the United States are more trusted by their fellow citizens than those in Latin America (Cao & Zhao, 2005).

Affective Intelligence and Support for Criminal Justice Institutions

To fully appreciate how crime affects citizen trust in the police and other justice institutions, we should take into account how individuals acquire and use knowledge to develop their political opinions. Citizens typically have low levels of political interest and information (Delli-Karpini & Keeter, 1996), and their responses to survey questions are often sampled from “top of the head” considerations (Lodge & Taber, 2013; Zaller, 1992). Affective models of political psychology, however, identify conditions when citizens are more attentive to their environment and are therefore more exposed to political messaging. Affective intelligence theory describes two emotional systems—dispositional and surveillance—that regulate our thoughts and actions. The dispositional system largely guides our normal, everyday behavior via the emotions of enthusiasm or aversion. The surveillance system, however, reacts to surprises and menacing situations, cuing the need to revisit normal patterns of thought (Marcus, Neuman, & MacKuen, 2000). Typically, citizens can rely on habitual patterns of behavior with little conscious consideration. For example, citizens normally engage in politics by using heuristics like party and ideology, which is a manifestation of the dispositional system reducing complexity (Marcus et al., 2000).

When new situations or unexpected developments challenge the wisdom of routine thinking, however, the surveillance system triggers emotions like fear and anxiety. We then increase the attention paid to our environment and question the effectiveness of prior strategies (Maestas & Atkeson, 2012; Marcus et al., 2000; Merolla & Zechmeister, 2009). Shocking occurrences like terrorism or natural disasters commonly motivate this behavior, but less dramatic changes can also alter political and social considerations, sometimes toward a more authoritarian character (Feldman & Stenner, 1997; Huddy, Feldman, Taber, & Lahav, 2005; Sales, 1972). When activated, the surveillance system concentrate’s citizens’ attention on their surroundings in order to evaluate the ability of prior strategies to effectively address a new challenge. If needed, citizens seek more data and update their beliefs and modes of thought.

The trauma and surprise of crime victimhood should start the surveillance system, increasing the attention citizens pay to their environment. Minimally, it will lead to new thoughts about how to avoid future victimhood. The consequent anxiety may also increase attention to the political world, particularly when it is relevant to crime. This is supported by the observation that crime victims participate more politically than nonvictims (Bateson, 2012). In this way, victims and other anxious citizens are not simply passive vessels for political entrepreneurs to manipulate but use their experience to motivate political action.

In addition to victimization, the fear from exposure to others’ experience with crime, either via personal contact or through the media, can activate the surveillance system, and consequently a heightened sense of attentiveness (Marcus, Sullivan, Theiss-Morse, & Stevens, 2005). Someone who is frightened or nervous studies the environment more carefully and pays more attention to leaders who address the causes of this anxiety. It is important to consider that this is not just a temporary disposition. Once fear and anxiety have been triggered, the surveillance system may be operative over long periods of time and more intensely activated whenever the subject of crime emerges, either directly or indirectly.

There is empirical evidence from Latin America to support this theory that crime-induced anxiety will trigger the surveillance system. For example, using survey data from the Latin American Opinion Project (LAPOP), we can observe that crime victims are more likely to pay attention to media, express higher levels of political interest, report more frequent conversations about politics, and demonstrate higher levels of political knowledge. In difference of means tests conducted for this analysis², crime victims are more likely than nonvictims to read newspapers and Internet news, watch television, and listen to the radio. The results are similar for those who fear crime in their neighborhoods. Victims, on average, are also able to answer 2.9 questions (of four) about current events correctly, while nonvictims score only 2.6 (p < .001). Finally, nearly 10% of all crime victims “discuss politics with other people” daily, while less than 7% of nonvictims do.³

After crime-induced anxiety promotes closer scrutiny of the political environment, the information encountered may condition individual attitudes. If the informational environment provides alternate strategies, anxious citizens may have a greater propensity to accept them (Marcus, MacKuen, & Neuman, 2011; Marcus et al., 2005). This is not to suggest that citizens are not actively involved in this negotiation of beliefs—merely that greater exposure to the message of political elites will on average make them more influential. The extent to which this occurs will vary depending on other factors such as partisanship, political polarization, and personal efficacy.⁴ Nevertheless, if consistently exposed to certain attitudes and ideas, these may well become their own (Marcus et al., 2005, 2011; Merolla & Zechmeister, 2009). In this respect, while crime may undermine support for the police and the justice system, the increased attention from anxious citizens provides greater opportunities for politicians to promote iron-fist policies and project an image of strength and resolve (Bergman & Whitehead, 2009).

Elite Messaging and Mano Dura

It is therefore useful to examine the relationship between insecurity and trust in the CJS within the context of political elites and elite framing (Perez, 2004). Although mano dura crime-control policies and their relationship to political support is examined in specific countries (Bergman & Whitehead, 2009; Holland, 2013; Uang, 2013), there is a need for empirical studies analyzing how variance in elite framing of physical insecurity influences public support across the region. Most prior research gives little to no attention to the effect that elite framing has on public opinion in the Latin American context. Malone (2010) indicates that the effect of crime on support for the justice system depends importantly on national context. She notes that in particular, the variation in terms of crime rates and institutional effectiveness leads to differential effects in terms of fear’s influence on trust in the CJS in Central America. This leaves out an important part of the story in terms of how citizens interact with the informational environment influenced by elites to form particular attitudes. I expand on this work by providing evidence that elite discourse modifies the effect of crime on how citizens view their criminal justice institutions.

Political elites⁵ can affect public opinion by heightening the political salience of crime and violence (Druckman, 2004; Jacobs & Shapiro, 1994) using their resources to influence how much citizens regard and prioritize social problems (Sotirovic, 2003). Natural disasters and acts of terror, for example, are sometimes opportunities for elites to weaken citizens’ commitment to civil liberties and other democratic principles (Merolla & Zechmeister, 2009). Crime-induced fear and anxieties are traumatic in similar ways and may therefore shape citizen attitudes about their institutions. Elites in these contexts therefore often turn to seemingly simple solutions, and most often this means supporting “tough-on-crime” style policies (known as “mano dura” in Spanish or “iron fist”; Bergman & Whitehead, 2009). In particular, this has meant using the military to fight crime, drastic roundups of suspected criminals, draconian punishments, and sweeping antigang laws which at the extreme can even criminalize appearance (Garzon-Vergara, 2015; Swanson, 2013; Wolf, 2011)

Given the nature of affective intelligence and the influence of elite cuing, political elites posturing as tough on crime may reassure citizens suffering from crime-related anxiety. Typically, lacking the financial and institutional capability to solve the root causes of crime, they compete to be the “toughest” in rhetoric, hoping to reassure an anxious public that help is on the way (Chevigny, 2003). It is common for Latin American political elites to boast about being tough on crime, while disparaging their opponents for their lack of similar resolve. In El Salvador, for example, mano dura policies were used by the conservative National Republican Alliance to help unify their party and maintain political control for two decades by appealing to business elites and crime-fearing citizens in rural areas (Holland, 2013). As another example, in the run-up to the 2011 presidential election in Peru, candidate Keiko Fujmori boasted to supporters, “I will be firm in imposing drastic and harsher sanctions; an iron fist against crime!” (El Peruano, 2011). But the embrace of iron-fist or zero-tolerance policies is not uniform. The rhetoric of mano dura tends to be much stronger where crime rates are higher, while in countries like Costa Rica, Uruguay, and Chile, politicians tend to avoid it. Left-wing politicians in high-crime countries have also tried to soften the mano dura rhetoric, but they still face public pressure to take zero-tolerance approaches to crime. For example, the “mano amiga” efforts of Funes in El Salvador were weakly implemented, and his left-wing government largely kept previous iron-fist policies in place. In other left-wing countries, the rhetoric of mano dura has largely been muted—such as in Venezuela and Nicaragua. Under the rule of Ortega, Nicaragua is unique for its low levels of crime despite high levels of poverty and other similarities with the Northern triangle (Cruz, 2011; Malone, 2010), which may be the result of its efforts to ban former soldiers in the country’s civil war from transitioning to policing careers (Cruz, 2011) and its holistic approach to law enforcement (Malone, 2012). Nevertheless, even in countries not strongly associated with mano dura, leaders can be tempted to use strong anticrime rhetoric (such as Chinchilla’s campaign slogan to be “firme” in Costa Rica) and borrow policies from the iron-fist playbook (like Ortega’s use of the military in fighting crime).

Fearful citizens will likely be more attentive to this strong elite rhetoric (Marcus et al., 2005, 2011), causing citizens to more firmly support their criminal justice institutions compared to environments of relatively weak resolve among the political leaders (Zaller, 1992). This is not a claim that individuals have more positive assessments of the CJS in mano dura promoting countries. These areas are more apt to have weak institutions in general, thereby generating less support. Controlling for other factors, however, elites advocating a more authoritative (even authoritarian) response to crime can mitigate these negative effects caused by crime-related fears. Anxious citizens already have a tendency toward greater authoritarianism (Huddy et al., 2005; Sales, 1972) and consequently are more responsive to elites promoting the “iron fist.”

Expectations

The mechanisms described above lead to two empirical propositions regarding elite rhetoric and citizen trust in the CJS in Latin America. Elite framing influences public attitudes in a number of situations (Chong & Druckman, 2007), many of which are similar to crime (Maestas & Atkeson, 2012; Marcus et al., 2000, 2005; Merolla & Zechmeister, 2009). Variation among Latin American political elites in terms of “iron-fist” policy promotion (Chevigny, 2003; Uang, 2013) will cause differences in mass attitudes toward the police, particularly when crime-related fear and anxiety cause citizens to consume more elite rhetoric (Marcus et al., 2000, 2005, 2011). Consequently, elite framing should condition the effect of victimization and fear of crime on support for criminal justice institutions.

Hypothesis 1a: Crime-induced anxiety will have a more weakly negative or null effect on citizen support for the police and the CJS as elite messaging on crime is more resolute and reflective of an “iron-fist” approach.

Hypothesis 1b: Crime-induced anxiety will have a more weakly negative or null effect on citizen belief in the effectiveness of police and the CJS as elite messaging on crime is more resolute and reflective of an “iron-fist” approach.

Research Design and Variables

I use data from the 2008 LAPOP surveys and Nina Wiesehomeier and Kenneth Benoit’s (2009) expert survey on political parties in 12 Latin American nations (conducted in 2007) to test these hypotheses. Although these data are now approximately a decade old, the latter is unique data capturing more specific attitudes toward crime control of political elites. Other survey data of elite actors in Latin America, such as the Parliamentary Elites of Latin America surveys produced by the University of Salamanca do not include as specific a measure of policy preference as it relates to crime control. Using the 2008 round of the LAPOP is appropriate given that the surveys were administered shortly after the measure of elite attitudes, thus mitigating the problem of changing political environments over time. Although the hypotheses may apply to other regions of the world, restricting the analysis to Latin America avoids potentially invalid inferences when analyzing countries with particularly distinct cultures and histories (Schaap & Scheepers, 2014). Data for one of the key independent variables, tough-on-crime messaging, are only available for 12 of 18 Latin American countries (Mexico, Guatemala, Colombia, Venezuela, El Salvador, Chile, Argentina, Honduras, Panama, Peru, Bolivia, and Brazil). Fortunately, these countries represent significant variation on relevant characteristics and are extremely similar to the full sample on key variables. In both samples, the self-reported crime victimization rate is 18%. In the 12-country sample, 27% of respondents indicate that crime is the worst problem facing their country, contrasted with 23% in the full regional sample. The mean values of the dependent variables are indicated in Table 1.

Table 1.

Means of Key Independent Variables and Dependent Variables.

Variables	Full Sample (18 Countries) Mean	Small Sample (12 Countries) Mean
Dependent Variables: Scale = 1–7
Trust in the CJS	3.63 (out of 7)	3.66
Trust the police	3.70 (out of 7)	3.71
CJS effectiveness	2.21 (out of 4)	2.20
Police effectiveness	2.17 (out of 4)	2.13
Independent variables
Tough-on-crime messaging	Not available	12.25
Crime victim	.183	.183
Fear of crime	2.30 (out of 4)	2.34 (out of 4)
Crime anxiety	.102	.106

Note. Full and Small Samples. CJS = Criminal Justice System. Adapted from Americas Barometer (2008).

Table 2.

Summary Statistics.

Variables	Mean	SD	Minimum	Maximum
Tough-on-crime messaging	12.235	1.879	7.98	15.37
Crime prioritization	15.179	1.859	12.29	18.06
Elite polarization on crime	3.251	1.943	0.637	6.53
Homicide	3.127	19.755	3.2	64.6
Polity	6.946	1.319	4	10
Trust CJS	3.664	1.694	1	7
Trust police	3.707	1.813	1	7
Faith in judicial system	2.203	0.961	1	4
Faith in police	2.134	0.939	1	4
Crime victim	0.183	0.386	0	1
Fear crime	2.344	0.924	1	4
Crime anxiety	0.106	0.307	0	1
Police bribery	0.111	0.314	0	1
Male	0.484	0.500	0	1
Age	38.705	15.773	18	94
Catholic	0.737	0.440	0	1
Protestant	0.044	0.205	0	1
No religion	0.082	0.275	0	1
Education	9.075	4.653	0	18
Wealth	2.918	1.397	1	5
Media exposure	9.827	2.499	4	16
Political knowledge	2.685	1.425	0	5
Approve of president	3.179	0.920	1	5
Trust parties	3.087	1.678	1	7
Life satisfaction	3.158	0.808	1	4
Trust	2.703	0.888	1	4
City size	2.924	1.508	1	5
N	1,9721

Dependent Variables

I test the hypotheses related to elite messaging on four different indicators of trust in the justice system—two related to criminal justice institutions and two regarding the police. For each category I include a dependent variable gauging broad faith in the system and another indicating trust in a specific institution. Trust in the CJS is based on the following survey question: “To what extent do you trust the justice system?” Values range from 1 (not at all) to 7 (a lot). The interclass correlation coefficient (ICC) is 6.93, which means that approximately 7% of the variance is caused by cross-national differences. Although somewhat modest, the ICC is of a greater magnitude than that of much published research (Morgan & Buice, 2013). A scatterplot of the mean levels of trust in the CJS and crime victim in Latin American countries can be seen in Figure 2. As expected, there is a negative trend in which higher levels of fear correlate with less trust in the system. Trust in the police comes from a survey question asking “To what extent do you trust the national police?” Values range from 1 (not at all) to 7 (a lot). The ICC is 9.0, meaning that 9% of the variation in the dependent variable can be explained by national level factors. This is a good indication that a multilevel model is appropriate and that previous individual-level models may suffer from omitted variable bias.

Figure 1.

Scatterplot of average elite tough-on-crime messaging and elite polarization. Adapted from Wiesenhomeier & Benoit (2009).

Figure 2.

Crime victimization and trust in the criminal justice system. Adapted from Americas Barometer (2008).

Because the variables described above may be excessively general in that trust in the CJS can be a multifaceted concept, I address Hypothesis 1b using more specific measures of citizen attitudes about institutional effectiveness. CJS effectiveness is based on the question, “If you were a victim of a robbery or assault how much faith do you have that the judicial system would punish the guilty?” It has values ranging from 1 (none) to 4 (a lot). Police effectiveness derives from a similar question with the response options: “If you were a victim of a robbery or assault how much faith do you have that the police would apprehend the guilty?” Approximately, 5% of the variation in these dependent variables occurs at the national level.

Independent Variables

Tough-on-crime messaging measures the policy preferences of political elites on a continuum from those favoring the protection of civil liberties to those who prioritize safety even when this requires curtailing liberties. Although lacking a direct measure of elite rhetoric, I use a reasonable proxy from a 2006–2007 expert survey of national political party platforms. Nina Wiesehomeier and Kenneth Benoit’s survey of hundreds of Latin American scholars and practitioners’ measures the positions of political parties on numerous issues including criminal justice policy (Wiesehomeier & Benoit 2009). The experts placed the parties on a scale from 1 to 20, denoting priorities ranging from the protection of individual liberties to security. At the left end of the continuum, the party “promotes protection of civil liberties, even when this hampers efforts to fight crime/delinquency, violence, and organized crime” (=1). The right end of the scale indicates that the party “supports tough measures to fight crime/delinquency, violence, and organized crime, even when this means curtailing civil liberties” (=20). An average party indicator is created from the various expert responses and then weighted by the percentage of votes garnered by the party in the most recent national election before the survey. The mean of tough-on-crime messaging for all 12 countries is 12.23 (on a scale of 20). Argentina gives the greatest priority to protection of civil liberties (7.98) while Honduras is the most reflective of an “iron-fist” approach (15.37).⁶ In bivariate analyses, tough-on-crime messaging has the expected positive and statistically significant correlation with all four dependent variables.

Tough-on-crime messaging is hypothesized to condition the effect that being a crime victim and fear of crime have on institutional support. Crime victim is a dichotomous variable identifying whether the respondent has been a victim of any type of crime in the past 12 months (yes = 1, no = 0).⁷ Experience with crime is typically a traumatic or at least unexpected event that likely triggers the surveillance system. Fear of crime measures the extent to which respondents feel “very safe, somewhat safe, somewhat unsafe, or very unsafe” in their respective neighborhoods. Consistent with affective intelligence theory, both crime victim and fear of crime may be indicators of anxiety resulting from physical insecurity. While victims of crime should pay more attention to their environment, the fear and anxiety that this victimization produces may be short lived. Many victims of crime do not express personal anxiety about insecurity in their specific neighborhoods. At the same time, nonvictims may experience vicariously the kind of fear and anxiety that activates the surveillance system, which makes fear of crime a similarly valid measure for evaluating the hypotheses. Both variables negatively correlate with trust in the police in much of the scholarly literature (Boateng, 2016; Ivkovic, 2008; Malone, 2010; Sabet, 2012).

Control Variables

To account for spuriousness, the empirical models include several variables common to this area of research. One of these is police bribery which indicates whether the respondent has been approached by a police officer for a bribe or not. This could contribute to an increase in fear and can also be a strong predictor of negative attitudes toward the police and other criminal justice and political institutions (Chu, Song, & Dombrink, 2005; Cruz, 2015; Ivkovic, 2008; Sabet, 2012). Other controls include individual factors such as gender, age, and years of education which are thought to correlate with both fear of crime and institutional support (Ivkovic, 2008; Kwak, San Miguel, & Carreon, 2012) and commonly used in research on political attitudes and behavior in Latin America (Booth & Seligson, 2009). Media exposure indicates the extent to which citizens seek new information from TV, newspapers, and the Internet. Dummy variables for Catholic, Protestant, and no religion account for religious beliefs and associations, and political knowledge measures a respondent’s political knowledge with an index of questions identifying important foreign and domestic politicians. Quintiles of wealth is a relative indicator of household ownership of goods, using principal components analysis (Cordova, 2009). I also control for the size of the respondents’ city, their level of trust, and their life satisfaction (Kwak et al., 2012).

Another important phenomenon to control for is the partisan environment. Institutional trust is often strongly influenced by partisan affiliation and polarization, and these may also be correlated with anxiety and victimization. The models include three variables to account for this: approve of president, trust political parties, and crime prioritization. The first two are individual factors measuring the extent to which the respondent supports the executive and the performance of political parties. The third is a contextual-level variable measuring the extent to which national political elites prioritize the issue of crime⁸. It is based on expert opinion about Latin American parties collected by Wiesenhomeier and Benoit (2009) and described above. Finally, I include two additional contextual variables, polity and homicide rate to account for the effects that larger environments may have on citizen attitudes toward their political system (Booth & Seligson, 2009; Carreras, 2013; Ceobanu et al., 2011; Ivkovic, 2008) and more specifically to concerns about the justice system (Jang et al., 2010; Nivette, 2016). Because of concerns about including too many level-2 variables in the regressions, I also run all models without these final two controls, and the results are not substantively different.

Results

The hypotheses are tested using multilevel regressions with cross-level interactions. The hypotheses require an interaction term as tough-on-crime messaging conditions the effect of crime victim and fear of crime on the dependent variables. The models can be represented mathematically by the following equation

P r (y_{i j} = 1 | x_{i j}, ζ_{j}) = β_{0} + β_{1} x_{1 i j} + β_{2} x_{2 i j} + β_{3} x_{3 j} + β_{4} x_{4 j} + β_{5} x_{3 j} x_{4 j} \dots β_{n} x_{n i j} + ζ_{j} .

with i indexing the individual and j indexing the country.

The main argument of this article predicts a varying effect of crime-related anxiety (crime victimization and fear of crime) on trust in the justice system across different levels of tough-on-crime messaging. Specifically, the negative effects of crime victimization and fear of crime on trust should decrease due to higher levels of mano dura or “iron-fist” policy promotion by elites. Before estimating this modifying effect, I test the direct effect that these variables have on trust in the police and the larger CJS (see Table 3). Models 1 and 2 assess the effects of crime-related variables on trust in CJS. Both crime victim and fear of crime are statistically significant (p < .001), but the substantive effect is relatively small. In Model 1 which only includes individual-level factors, victims rate their trust in the CJS 0.09 units lower on a scale of 7 than nonvictims. The effect of a one-unit change in fear of crime (i.e., the difference between feeling “somewhat unsafe” and “very unsafe”) is slightly lower at 0.08. Police bribery appears to be a stronger predictor of trust, as the effect of being a victim of attempted extortion by the police is approximately 2½ times as large as the effect of being a crime victim. In the second model (Table 3), I include contextual factors, and as predicted, tough-on-crime messaging has a positive but insignificant effect on trust in CJS. The effect of crime prioritization, however, is both positive (0.08) and significant (p <.01). As in Model 1, police bribery appears to be the crime-related variable that has the strongest effect on trust in the CJS (β = −.25; p < .001). The other contextual variables, polity and homicide rate, have coefficients that are indistinguishable from zero.

Table 3.

Fixed Effects and Multilevel Models.

Independent Variables	(1) Trust CJS	(2) Trust CJS	(3) Trust Police	(4) Trust Police
Individual factors
Crime victim	−.086*** (.026)	−.128*** (.027)	−.181*** (.027)	−.123*** (.027)
Fear of crime	−.083*** (.012)	−.158*** (.027)	−.112*** (.012)	−.156*** (.022)
Police bribery	−.210*** (.034)	−.254*** (.031)	−.369*** (.035)	−.469*** (.111)
Male	−.002 (.021)	.050 (.032)	−.003 (.021)	−.048 (.031)
Age	.001 (.001)	.004 (.003)	.004*** (.001)	.007* (.003)
Catholic	−.064* (.030)	.061 (.096)	.039 (.031)	.267** (.111)
Protestant	.055 (.054)	.121*** (.029)	−.006 (.056)	.297** (.103)
Education	−.009** (.003)	.000 (.005)	−.009** (.003)	.000 (.007)
Wealth	−.006 (.008)	−.001 (.007)	−.008 (.008)	.036** (.015)
Political knowledge	−.026** (.009)	−.029 (.021)	−.020* (.009)	−.046** (.016)
Approve of president	.258*** (.011)	.271*** (.053)	.265*** (.011)	.212*** (.028)
Trust political parties	.367*** (.006)	.370*** (.009)	.354*** (.006)	.372*** (.029)
Life satisfaction	.068*** (.013)	.093*** (.023)	.085*** (.014)	.047 (.029)
Trust	.089*** (.012)	.084*** (.026)	.145*** (.012)	.124*** (.016)
City size	−.029*** (.007)	−.047*** (.010)	−.040*** (.007)	−.100** (.035)
Contextual factors
Crime prioritization		.077** (.030)		.115*** (.037)
Tough-on-crime messaging		.027 (.032)		.160** (.056)
Polity		−.004 (.041)		.255*** (.066)
Homicide		−.003 (.002)		−.020*** (.006)
N	23,833	15,174	24,054	15,310
R ²	.199	.226	.203	.250

Note. Constant, no religion, and media exposure suppressed. CJS = criminal justice system.

^† p < .10. * p < .05. **p < .01. ***p < .001.

Models 3 and 4 look at the effects of crime-related variables on trust in the police. As in Models 1 and 2, vrime victim and fear of crime are negative and statistically significant, with victimization having approximately the same effect on trust as a one-unit change in the level of fear. The effect of police bribery, however, is twice as strong as crime victimization and 3 times greater than a one-unit increase in fear. In Model 4, which accounts for both individual and contextual level factors, tough-on-crime messaging has a positive and significant effect as expected and seems to have a much stronger influence on trust in the police than in the CJS at large. The coefficient is 0.16 (p < .01), indicating that on average, citizens in countries in the 90th percentile of tough-on-crime messaging have a full unit (or approximately 14%) more trust in the police than those in the 10th percentile. Crime prioritization is also positive and significant (p < .001). Unlike in Models 1 and 2, the other contextual variables appear to exert a statistically significant effect on trust in the police. A change in polity from the 10th to 90th percentile increases trust police by well over half a unit on a 7-unit scale. This suggests that in environments of greater democratic procedure, trust in institutions like the police force is higher, ceteris paribus. The expected negative coefficient for homicide is small but statistically significant (p < .05), which may indicate that the level of violence in the environment can affect attitudes toward the police, independent of one’s own crime victimization and vulnerability.

Table 4 reports the results of interactive models. As expected, the interaction between crime victim and tough-on-crime messaging has a positive and significant effect on both Trust in CJS (β = .025, p < .05) and trust police (β = .023, p < .10). Substantively, these are modest effects, indicating that the difference in the marginal effect (ME) between the least and most “iron-fist” informational environments is about 0.2 on a scale of 7. The interaction effect between fear of crime and tough-on-crime messaging is also positive and significant for both trust CJS (β = .033, p < .01) and trust police (β = .017, p < .10).

Table 4.

Multilevel Models With Interactions: Independent Variable = Tough-on-Crime Messaging.

Independent Variables	(1) Trust CJS	(2) Trust CJS	(3) Trust Police	(4) Trust Police
Interactions and constituent terms
Tough-on-crime messaging	−.062 (.039)	.022 (.032)	.106* (.056)	.156** (.056)
Crime victim	−.128*** (.029)	−.399** (.160)	−.122*** (.026)	−.366*** (.090)
Fear of crime	−.512*** (.126)	−.158*** (.027)	−.344** (.132)	−.156*** (.022)
Tough on Crime × Crime Victim		.025* (.015)		.023** (.009)
Tough on Crime × Fear of Crime	.033** (.012)		.017⁁ (.011)
Contextual factors
Polity	−.034 (.044)	−.004 (.041)	.226*** (.064)	.255*** (.066)
Crime prioritization	.075* (.035)	.076** (.030)	.104*** (.030)	.115*** (.037)
Homicide	−.002 (.003)	−.003 (.002)	−.019*** (.006)	−.020*** (.006)
Individual factors
Police bribery	−.253*** (.030)	−.256*** (.031)	−.467*** (.112)	−.471*** (.111)
Catholic	.057 (.096)	.062 (.096)	.266** (.112)	.269** (.111)
Protestant	.116*** (.029)	.120*** (.029)	.295** (.105)	.296** (.104)
Media exposure	.003 (.014)	.003 (.014)	.013 (.008)	.013 (.008)
Political knowledge	−.029 (.022)	−.029 (.021)	−.047** (.016)	−.046** (.016)
Approve of president	.271*** (.052)	.271*** (.053)	.212*** (.028)	.212*** (.028)
Trust political parties	.371*** (.009)	.370*** (.009)	.372*** (.029)	.372*** (.029)
Life satisfaction	.092*** (.023)	.093*** (.023)	.047 (.029)	.047 (.029)
Trust	.088*** (.026)	.084*** (.026)	.126*** (.017)	.124*** (.016)
City size	−.046*** (.010)	−.047*** (.010)	−.099** (.035)	−.100** (.034)
N	15,174	15,174	15,310	15,310
R ²	.228	.227	.253	.251

Note. Const., male, age, education and wealth suppressed. CJS = criminal justice system.

^† p < .10. *p < .05. **p < .01. ***p < .001, one-tailed test.

Analyzing the interactive effects of the independent variables conditioned by tough-on-crime messaging is best accomplished via ME plots (Brambor, Clark, & Golder, 2006). The ME plots in Figures 2 and 3 demonstrate the conditional influence of crime victim and fear of crime on the dependent variables at varying levels of tough-on-crime messaging. At lower levels of tough-on-crime messaging, the MEs of both crime victim and fear of crime is strongly negative, but in contexts where elites are promoting more “iron-fist” approaches to crime, these effects are weakly negative or null. This suggests that in places like Honduras, Guatemala, and El Salvador, tougher elite messaging mitigates the negative effect of crime-related fears, while in countries like Argentina, Brazil, and Peru, insecure citizens are less trusting of their institutions, ceteris paribus. The MEs plot for the interaction between fear of crime and tough-on-crime messaging also shows that the relationship is consistent with the hypothesis that in more rights-respecting environments of elite messaging, the effect of crime-induced fear on trust is strongly negative, but at high levels of tough-on-crime messaging, the effect is null or approaching null. In other words, at higher levels of tough-on-crime messaging (mean = 12.25), the conditionally negative effect of fear of crime on trust in the justice system becomes less and less intense.

Figure 3.

Marginal effects plots: conditional effect of crime victim (top row) and fear of crime (bottom row) at different levels of tough-on-crime messaging.

Being a victim of crime predictably decreases one’s faith in the system and its institutions, but the negative effect of being exposed to a solicitation for a bribe by a police officer is even stronger, which is consistent with previous research (Sabet, 2012). Other variables that are important predictors of a respondent’s trust in the justice system are approve of the president, trust political parties, life satisfaction, and interpersonal trust, all of which have a strongly positive correlation with the dependent variables, ceteris paribus. Political variables are particularly influential at both the individual and contextual level. One’s trust in political parties, for example, increases trust in the justice system by 1.5 points between the first and third quartiles. At the contextual level, crime prioritization is positive and significant in all four models. Catholic affiliation is also associated with greater trust in the system and in the police. Wealth, age, education, and male do not appear to be good predictors of attitudes toward the justice system in relation to other variables.

Alternate Dependent Variables

Table 5 reports the results of models testing the effects of crime and elite messaging on more specific measures of citizen faith in the institutional effectiveness of the police and the justice system. In models 1 and 3, we see that for both dependent variables, the interaction between fear of crime and Tough-on-Crime Messaging has a positive and significant (p < .001) effect, similar in effect size (0.02) to the models on general trust in the institutions. The interactive effects for crime victim and Tough-on-Crime, however, are null. The direct effect of “iron-fist” messaging among nonvictims is positive and significant.

Table 5.

Multilevel Models.

Independent Variables	(1) CJS Effective	(2) CJS Effective	(3) Police Effective	(4) Police Effective
Interactions/constituent terms
Tough on crime	.008 (.025)	.052** (.020)	.006 (.022)	.051*** (.015)
Fear of crime	−.317*** (.050)	−.106*** (.016)	−.332*** (.063)	−.115*** (.015)
Crime victim	−.138*** (.023)	−.157 (.113)	−.131*** (.015)	−.203*** (.065)
Tough on Crime × Fear of Crime	.019*** (.004)		.020*** (.005)
Tough on Crime × Crime Victim		.002 (.010)		.007 (.007)
Contextual factors
Polity	.052* (.025)	.051* (.023)	.079*** (.019)	.078*** (.018)
Crime prioritization	.009 (.017)	.009 (.016)	.018 (.013)	.018 (.013)
Homicide	−.004* (.002)	−.004* (.002)	−.006*** (.002)	−.005*** (.002)
Individual factors
Police bribery	−.181*** (.026)	−.180*** (.026)	−.160*** (.025)	−.161*** (.025)
Male	.047** (.018)	.046** (.018)	.017 (.019)	.016 (.019)
Protestant	−.095** (.036)	−.092** (.034)	−.054** (.019)	−.050** (.017)
Education	−.013*** (.002)	−.013*** (.002)	−.010*** (.002)	−.010*** (.002)
Political knowledge	−.012** (.004)	−.012** (.004)	−.015* (.008)	−.014* (.008)
Approve of president	.071*** (.017)	.071*** (.017)	.054*** (.013)	.053*** (.013)
Trust political parties	.083*** (.021)	.083*** (.021)	.082*** (.022)	.081*** (.022)
Life satisfaction	.054*** (.015)	.055*** (.014)	.044*** (.014)	.044*** (.013)
Trust	.055*** (.008)	.052*** (.007)	.062*** (.007)	.059*** (.006)
City size	−.022*** (.006)	−.023*** (.006)	−.033*** (.008)	−.034*** (.008)
N	15,201	15,201	15,262	15,262
R ²	.092	.090	.089	.089

Note. Dependent variable = CJS effective and police effective; Catholic, media, age, wealth and const. suppressed. CJS = criminal justice system.

^† p < .10. * p < .05. **p < .01. ***p < .001, one-tailed test.

The results for the other contextual variables are more consistent across the models for institutional effectiveness than for generalized institutional trust. Polity has a positive and significant effect in all four models, while the effect of the homicide rate is negative in each. While crime prioritization has a positive effect on trust in the justice system in the police, its effect on the effectiveness of these institutions is null.

Since the results modeling fear of crime and crime victim are somewhat inconsistent, I create a new independent variable, crime anxiety, which treats victimization as the same concept in terms of inducing the surveillance system and making citizens more receptive to elite messaging on crime solutions. Although most research in the area of crime and public opinion include both victimization and neighborhood fear variables as separate concepts, it may be odd to interpret the effects of these variables independent of each other. In particular, it is curious to speak of the effect of crime victimization independent of the levels of fear of crime one has. Hence, I rerun the models using a variable measuring the concept of crime-induced anxiety, as one which captures both the experience of victimization and fear together. Crime anxiety is therefore a dichotomous variable indicating that one is both (a) a victim of crime and (b) experiencing moderate to high levels of crime-induced fear. Table 6 reports the results of this robustness check. The effects of crime anxiety are negative and significant, with a larger coefficient size than either crime victim or fear of crime which is to be expected given that these variables are strongly correlated. The interactive effect between Crime Anxiety and Tough-on-Crime Messaging is statistically significant in Model 1 and has a marginally significant effect (p < .10) on trust police in Model 2. The interaction between Crime Anxiety and Tough-on-Crime Messaging has a positive and significant effect on both CJS Effectiveness and Police Effectiveness. The MEs plots shown in Figure 4 indicate that across all four dependent variables, the negative effect of crime-induced anxiety weakens as the messaging of political elites is more resolute and aggressive, and less concerned about civil liberties.

Table 6.

Multilevel Models.

Independent Variables	(1) Trust CJS	(2) Trust Police	(3) CJS Effective	(4) Police Effective
Interactions and constituent terms
Tough on crime	.021 (.033)	.154** (.055)	.046* (.020)	.048*** (.015)
Crime anxiety	−.609*** (.162)	−.484* (.240)	−.716** (.256)	−.522*** (.084)
Tough on Crime × Crime Anxiety	.034* (.018)	.027⁁ (.020)	.044* (.021)	.029*** (.007)
Contextual factors
Polity	−.010 (.043)	.249*** (.066)	.047* (.023)	.075*** (.018)
Crime prioritization	.083** (.031)	.121*** (.038)	.014 (.015)	.022* (.012)
Homicide	−.002 (.002)	−.020*** (.006)	−.004* (.002)	−.005*** (.002)
Individual factors
Police bribery	−.266*** (.032)	−.480*** (.112)	−.194*** (.022)	−.174*** (.022)
Male	.053* (.028)	−.032 (.031)	.051** (.018)	.019 (.019)
Age	.005 (.003)	.007* (.003)	.001* (.001)	.002* (.001)
Catholic	.064 (.098)	.261** (.109)	.040* (.022)	.011 (.021)
Protestant	.103*** (.027)	.278** (.098)	−.096** (.035)	−.056** (.021)
Education	−.000 (.005)	.000 (.007)	−.013*** (.002)	−.010*** (.002)
Political knowledge	−.027 (.021)	−.044** (.016)	−.011** (.005)	-.014 (.009)
Approve of president	.285*** (.053)	.219*** (.031)	.081*** (.016)	.064*** (.013)
Trust political parties	.370*** (.009)	.372*** (.031)	.083*** (.022)	.081*** (.023)
Life satisfaction	.098*** (.027)	.059* (.031)	.055*** (.011)	.044*** (.010)
Trust	.114*** (.027)	.155*** (.015)	.069*** (.009)	.079*** (.007)
City size	−.067*** (.014)	−.119*** (.036)	−.033*** (.006)	−.047*** (.009)
N	15,328	15,469	15,349	15,413
R²	.197	.248	.085	.086

Note. Independent variable = crime anxiety. Constant, wealth and media exposure suppressed. CJS = criminal justice system.

^† p < .10. *p < .05. **p < .01. ***p < .001 (one-tailed test).

Figure 4.

Marginal effects plots: conditional effect of crime anxiety at different levels of tough-on-crime messaging. Adapted from Americas Barometer 2008.

Discussion

This article builds on previous work to improve our knowledge about crime’s effect on support for police and criminal justice institutions in Latin America. While it was previously demonstrated that crime undermines specific support for democracy and institutions, the role of political elites in distinct national contexts was lacking. Scholars have referenced the importance of country-level factors (e.g., Malone, 2010) but stopped short of testing it empirically and/or lacked a theoretical framework consistent with our knowledge of political psychology. Conceiving of crime as an initiator of the surveillance system and including the influence of political elites over the informational environment, I explain how elite rhetoric lessens the damage to institutional trust caused by victimhood and fear of crime. This is typically accomplished through the advocacy of mano dura or “iron fist” crime control policies. The influence of elites is not trivial but still does not completely counteract the negative influence of victimhood and fear. Though not the first to examine contextual variables (previous work includes measures such as homicide rates and institutional performance), this study is original in its empirical investigation of the specific political messaging of politicians to influence individual attitudes. It therefore is an important extension in that it explicitly addresses what we know from political psychology about opinion formation.

A typical weakness of cross-sectional observational studies like this is its failure to deal with potential endogeneity. For the context of this article, it is reasonable to expect that politicians choose their rhetoric based to some degree on their expectations of citizen preferences. It is still, however, unclear why the interactive effect described above would exist if in fact citizen attitudes were driving the elite rhetoric, particularly given the weakness of the potential claim that trust (or lack thereof) in institutions is causing one’s victimhood status. Future research could nevertheless address endogeneity issues by including longitudinal variation and more subnational structural variables to allow for practical use of instrumental variable estimation to improve confidence in causal inferences.

The important evidence offered here about the role that political elites play in fomenting institutional trust in the face of significant violent crime will hopefully motivate more investigation in this area. As such it is a useful contribution to our knowledge of the influential variables promoting trust in relatively weak democratic institutions of Latin America. The findings suggest that the frequent use of iron-fist messaging related to crime is rational because it is effective. It conveys to victims and fearful individuals that their politicians are serious about addressing their main concerns about crime, which in turn promotes greater confidence in criminal justice institutions. This nevertheless points to a potentially alarming paradox. While shoring up trust in otherwise weak institutions may appear auspicious for democracy, the fact that politicians may accomplish this via attacks on civil liberties and arguably ineffective policies could well be counterproductive over the long term.

Footnotes

Appendix

Table A1.

Multilevel Models.

Independent Variables	(1) Trust CJS	(2) Trust CJS	(3) Trust Police	(4) Trust Police
Interactions/constituent terms
Tough on crime	−.115* (.052)	.001 (.041)	.068 (.046)	.133* (.058)
Crime victim	−.127*** (.029)	−.409** (.155)	−.121*** (.026)	−.367*** (.090)
Fear of crime	−.528*** (.134)	−.159*** (.027)	−.355** (.138)	−.156*** (.023)
Tough on crime × crime victim		.026* (.015)		.023** (.009)
Tough on crime × fear of crime	.034** (.012)		.018⁁ (.012)
Contextual factors
Polity	−.072 (.049)	−.008 (.043)	.209*** (.053)	.252*** (.058)
Elite polarization	.093* (.042)	.069* (.034)	.097** (.037)	.086* (.046)
Homicide	.003 (.005)	.001 (.004)	−.014** (.005)	−.016** (.006)
Individual factors
Police bribery	−.252*** (.031)	−.259*** (.031)	−.467*** (.112)	−.471*** (.112)
Catholic	.056 (.096)	.060 (.096)	.266** (.111)	.268** (.111)
Protestant	.118*** (.029)	.124*** (.029)	.296** (.105)	.297** (.104)
Wealth	−.001 (.006)	−.000 (.006)	.036** (.015)	.037** (.015)
Political knowledge	−.030 (.022)	−.030 (.021)	−.047** (.016)	−.047** (.016)
Approve of president	.271*** (.052)	.272*** (.052)	.212*** (.028)	.213*** (.028)
Trust political parties	.370*** (.009)	.370*** (.009)	.372*** (.029)	.372*** (.029)
Life satisfaction	.093*** (.023)	.094*** (.024)	.047 (.029)	.048 (.029)
Trust	.087*** (.026)	.083*** (.026)	.126*** (.017)	.124*** (.016)
City size	−.046*** (.010)	−.048*** (.010)	−.099** (.035)	−.100** (.034)
N	15,230	15,174	15,310	15,310
R ²	.223	.230	.266	.261

Note. Independent variable = Tough-on-crime messaging; Male, age, education, media and constant suppressed.

^† p < .1. *p < .05. **p < .01. ***p < .001 (One-tailed test).

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Ahmad

Hubickey

McNamara

IV . (2011). Trust in the national police. AmericasBarometer Insights: 201, 59, 1–1.

Bateson

(2010, 10 8). Crime, impunity, and the popular deligitimization of human rights in Guatemala. Paper presented at LASA2010: XXIX International Congress of the Latin American Studies Association, Toronto, ON.

Bateson

(2012). Crime victimization and political participation. American Political Science Review, 106, 570–587.

Bergman

M. S.

(2006). Crime and citizen security in Latin America. Latin American Research Review, 41, 213–227.

Bergman

Whitehead

(2009). Criminality, public security, and the challenge to democracy in Latin America. Notre Dame, IN: University of Notre Dame Press.

Boateng

F. D.

(2016). Neighborhood-level effects on trust in the police: A multilevel analysis. International Criminal Justice Review, 26, 217–236.

Booth

J. A.

Seligson

M. A.

(2009). The legitimacy puzzle in Latin America. New York, NY: Cambridge University Press.

Brambor

Clark

W. R.

Golder

(2006). Understanding interaction models: Improving empirical analyses. Political Analysis, 14, 63–82.

Cao

Zhao

J. S.

(2005). Confidence in the police in Latin America. Journal of Criminal Justice, 33, 403–412.

10.

Carreras

(2013). The impact of criminal violence on regime legitimacy in Latin America. Latin American Research Review, 48, 85–107.

11.

Ceobanu

A. M.

Wood

C. H.

Ribeiro

(2011). Crime victimization and public support for democracy: Evidence from Latin America. International Journal of Public Opinion Research, 23, 56–78.

12.

Chevigny

(2003). The populism of fear. Punishment and Society, 5, 77–96.

13.

Chong

D. M.

Druckman

J. N.

(2007). Framing public opinion in competitive democracies. American Political Science Review, 101, 637–655.

14.

Chu

Song

J. H. J.

Dombrink

(2005). Chinese immigrants’ perceptions of the police in New York City. International Criminal Justice Review, 15, 101–114.

15.

Cordova

(2009). Methodological note: Measuring relative wealth using household asset indicators. Americas Barometer Insights, 6, 1–9.

16.

Cruz

J. M.

(2011). Criminal violence and democratization in Central America: Survival of the violent state. Latin American Politics & Society, 53, 1–33.

17.

Cruz

J. M.

(2015). Police misconduct and political legitimacy in Central America. Journal of Latin American Studies, 47, 251–283.

18.

Delli-Karpini

M. X.

Keeter

(1996). What Americans know about politics and why it matters. New Haven, CT: Yale University Press.

19.

Diamond

(1999). Developing democracy: Toward consolidation. Baltimore, MD: Johns Hopkins University Press.

20.

Druckman

J. N.

(2004). Priming the vote. Political Psychology, 25, 577–594.

21.

Easton

(1965). A framework for political analysis. New York, NY: Prentice Hall.

22.

Easton

(1975). A reassessment of the concept of political support. British Journal of Political Science, 5, 43–57.

23.

El Peruano . (2011, 4 4). El estado debe responder. p. 5

24.

Feldman

Stenner

(1997). Perceived threat & authoritarianism. Political Psychology, 18, 741–77.

25.

Fernandez

K. E.

Kuenzi

(2010). Crime and support for democracy in Africa and Latin America. Political Studies, 58, 450–471.

26.

Fruhling

(2012). A realistic look at community policing in Latin America. Policing and Society, 22, 76–88.

27.

Gabbidon

S. L.

Higgins

G. E.

(2009). The Role of race/ethnicity and race relations on public opinion related to the treatment of blacks by police. Police Quarterly, 12, 102–115.

28.

Garzon-Vergara

J. C.

(2015). Tough on the weak, weak on the tough: Drug laws and policing. Washington, D.C: Latin American Program.

29.

Holland

A. C.

(2013). Right on crime? Conservative party politics and mano dura policies in El Salvador. Latin American Research Review, 48, 44–67.

30.

Huddy

Feldman

Taber

Lahav

(2005). Threat, anxiety, and support for antiterrorism policies. American Journal of Political Science, 49, 593–608.

31.

Ivkovic

S. K.

(2008). A comparative study of public support for the police. International Criminal Justice Review, 18, 406–434.

32.

Jacobs

L. R.

Shapiro

R. Y.

(1994). Issues, candidate image, and priming. American Political Science Review, 88, 527–54.

33.

Jang

Joo

Zhao

(2010). Determinants of public confidence in police. Journal of Criminal Justice, 38, 57–68.

34.

Johnson

Forman

J. M.

Bliss

(2012). Police reform in Latin America: Implications for U.S. policy. Washington, DC: Center for Strategic and International Studies.

35.

Krause

(2013). Supporting the iron fist: News media and public attitudes toward crime in Latin America (Doctoral dissertation, Notre Dame, IN, University of Notre Dame).

36.

Krause

(2014). Supporting the iron fist: Crime news, public opinion, and authoritarian crime control in Guatemala. Latin American Politics & Society, 56, 98–119.

37.

Kwak

San Miguel

C. E.

Carreon

D. L.

(2012). Political legitimacy and public confidence in police: An analysis of attitudes toward Mexican police. Policing: An International Journal of Police Strategies and Management, 35, 124–146.

38.

Lagos

Dammet

(2012). La seguridad ciudadana: El problema principal de América Latina. Retrieved from http://www.latinobarometro.org/documentos/LATBD_La_seguridad_ciudadana.pdf

39.

Lodge

Taber

C. S.

(2013). The rationalizing voter. New York, NY: Cambridge University Press.

40.

Maestas

Atkeson

L. R.

(2012). Catastrophic politics. New York, NY: Cambridge University Press.

41.

Malone

M. T.

(2010). The verdict is in: The impact of crime on public trust in Central American justice systems. Journal of Politics in Latin America, 3, 99–128.

42.

Malone

M. T.

(2012). The rule of law in Central America: Citizens reactions to crime and punishment. New York, NY: Bloomsbury.

43.

Malone

M. T.

(2013). Does crime undermine public support for democracy? Findings from the case of Mexico. The Latin Americanist, 57, 17–43.

44.

Marcus

G. E.

MacKuen

Neuman

W. R.

(2011). Parsimony and complexity: Developing and testing theories of affective intelligence. Political Psychology, 32, 323–336.

45.

Marcus

G. E.

Neuman

W. R.

MacKuen

(2000). Affective intelligence and political judgment. Chicago, IL: The University of Chicago Press.

46.

Marcus

G. E.

Sullivan

J. L.

Theiss-Morse

Stevens

(2005). The emotional foundation of political cognition. Political Psychology, 26, 949–963.

47.

Merolla

J. L.

Mezini

Zechmeister

E. J.

(2013). Crime, economic crisis, and support for democracy in Mexico [Special Issue]. Political Challenges in Mexico. 221–251

48.

Merolla

J. L.

Zechmeister

E. J.

(2009). Democracy at risk. Chicago, IL: The University of Chicago Press.

49.

Morgan

Buice

(2013). Latin American attitudes toward women in politics: The influence of elite cues, female advancement, and individual characteristics. American Political Science Review, 107, 644–662.

50.

Muggah

Szabo de Carvalho

Alvarado

Marmolejo

Wang

(2016). Making cities safer: Citizen security innovations from Latin America. Igarapé Institute, Inter-American Development Bank. Retrieved from https://publications.iadb.org/handle/11319/7757

51.

Nivette

A.E.

(2016). Institutional ineffectiveness, illegitimacy, and public support for vigilantism in Latin America. Criminology, 54, 142–175.

52.

Perez

O. J.

(2004). Democratic legitimacy and public insecurity: Crime and democracy in El Salvador and Guatemala. Political Science Quarterly, 118, 627–644.

53.

Reisig

M. D.

Parks

R. B.

(2000). Experience, quality of life, and neighborhood context: A hierarchical analysis of satisfaction with police. Justice Quarterly, 17, 607–629.

54.

Romero

Magaloni

Diaz-Cayeros

(2016). Presidential approval and public security in Mexico’s war on crime. Latin American Politics & Society, 58, 100–123.

55.

Sabet

(2012). Corruption or insecurity? Understanding dissatisfaction with Mexico’s police. Latin American Politics and Society, 55, 22–45.

56.

Sales

S. M.

(1972). Economic threat as a determinant of conversion rates in authoritarian and non-authoritarian churches. Journal of Personality & Social Psychology, 23, 420–428.

57.

Sampson

R. J.

Bartusch

D. J.

(1998). Legal cynicism and subcultural tolerance of deviance: The neighborhood context of racial differences. Law and Society Review, 32, 777–804.

58.

Sampson

R. J.

Raudenbush

S. W.

Earls

(1997). Neighborhoods and violent crime: A multilevel study of collective efficacy. Science, 277, 918–924.

59.

Sarsfield

Echegaray

(2005). Opening the black box: How satisfaction with democracy and its perceived efficacy affect regime preference in Latin America. International Journal of Public Opinion Research, 18, 153–173.

60.

Schaap

Scheepers

(2014). Comparing citizens’ trust in the police across European countries: An assessment of cross-country measurement equivalence. International Criminal Justice Review, 24, 82–98.

61.

Schuberth

(2016). Beyond gang truces and mano dura policies: Toward substitutive security governance in Latin America. Stability: International Journal of Security & Development, 5, 1–2.

62.

Schuck

A. M.

Rosenbaum

D. P.

Hawkins

D. F.

(2008). The influence of race/ethnicity, social class, and neighborhood context on residents’ attitudes toward the police. Police Quarterly, 11, 496–519.

63.

Sotirovic

(2003). How individuals explain social problems: The influences of media use. Political Communication, 53, 122–137

64.

Stack

S. J.

Cao

(1998). Political conservatism and confidence in the police: A comparative analysis. Journal of Crime and Justice, 21, 71–76.

65.

Swanson

(2013). Zero tolerance in Latin America: Punitive paradox in urban policy mobilities. Urban Geography, 34, 972–988.

66.

Uang

R. S.

(2012). Careful crackdowns: Human rights and campaigning on public security in Latin America. (Doctoral dissertation, Austin, TX, University Texas-Austin).

67.

Uang

R. S.

(2013). Campaigning on public security in Latin America: Obstacles to success. Latin American Politics & Society, 55, 26–51.

68.

Ungar

(2011). Policing democracy: Overcoming obstacles to citizen security in Latin America. Baltimore, MD: Johns Hopkins University Press.

69.

United Nations Office on Drugs and Crime. (2014). Global study on homicide. Retrieved from https://www.unodc.org/documents/gsh/pdfs/2014_GLOBAL_HOMICIDE_BOOK_web.pdf

70.

Wazed

Akhtar

(2015). The divergence of democratic policing: Reflection on Brazil and Argentina. Sociology and Anthropology, 3, 361–369.

71.

Wiesehomeier

Benoit

(2009). Presidents, parties, and policy competition. Journal of Politics, 71, 1435–1447.

72.

Weitzer

(2000). Racialized policing: Residents’ perceptions in three neighborhoods. Law and Society Review, 34, 129–155.

73.

Weitzer

Tuch

S. A.

(2002). Perceptions of racial profiling: Race, class, and personal experience. Criminiology, 40, 435–457.

74.

Weitzer

Tuch

S. A.

(2005). Determinants of public satisfaction with the police. Police Quarterly, 8, 279–297.

75.

Weitzer

Tuch

S. A.

(2006). Race and police in America. New York, NY: Cambridge University Press.

76.

Wolf

(2011). Mano dura: Gang suppression in El Salvador. Sustainable Security. Retrieved August 5, 2017, from https://sustainablesecurity.org/2011/03/01/mano-dura-gang-suppression-in-el-salvador/

77.

Zaller

J. R.

(1992). The nature and origins of mass opinion. Cambridge, MA: Cambridge University Press.