The Importance of Political Support

Seymour Martin Lipset maintained that a regime has a greater likelihood of survival when the most significant groups in society view the existing political system as the most appropriate.Footnote ¹⁰ This conviction is not created overnight. In democratic regimes, public support is in part the product of government performance, especially in the economic area. David Easton characterised those short-term attitudes towards the government as specific support, namely, public backing for the perceived actions and decisions of the incumbent authorities.Footnote ¹¹ However, such support can also extend to the principles, processes, and formal institutions that transcend particular governments.Footnote ¹² In this case, we are dealing with diffuse support for the regime; this is, a ‘reservoir of favourable attitudes that helps members to accept or tolerate inputs to which they are opposed’.Footnote ¹³ This article focuses on both levels of support: specific and diffuse.

Traditionally, the economy and the performance of incumbents have dominated the literature on political support.Footnote ¹⁴ However, research on the impact of crime and insecurity has made significant inroads in recent years.Footnote ¹⁵ Moreover, citizens' expectations, direct experiences with government organisations and demographic variables, such as education and social status, have been proven to shape the legitimacy of regimes.Footnote ¹⁶ As Diamond has pointed out, legitimacy ‘derives partly from the performance over time of the democratic regime, but it is also influenced (especially in the early life of a regime) by how specific democratic institutions articulate with traditionally legitimate forms of authority’.Footnote ¹⁷

The Political Importance of the Police

Diamond's argument has important implications for the study of police impact on legitimacy in Central America. In societies with a long authoritarian tradition, police are critical to assessments of regime performance, because they are the ones called to guarantee order and public security. At the same time, they have the potential to abuse their claim to the legitimate use of force and resort to the authoritarian practices of the past. Police performance is vital in consolidated democracies, but even more so in post-transition countries where public support for the regime is not yet firmly established.Footnote ¹⁸ In many Latin American countries, as Mark Ungar and others have pointed out, law enforcement institutions had to be reformed not only to improve the capacity of the emerging democratic states but, more importantly, to prevent a return to the oppressive practices of the authoritarian past.Footnote ¹⁹

Police institutions have a distinct two-fold quality within the state apparatus. In one sense, by contributing to public safety, they provide a service and they join the ranks of organisations whose performance has immediate consequences on the everyday well-being of the population: they perform an administrative function of the state. In another sense, police institutions also control the essential tools of state force, and they determine how that power is exerted. Societal stability and day-to-day citizen cooperation with street authorities depend on the widespread belief that law enforcement institutions and their operatives are legitimate.Footnote ²⁰ The police's behaviour is crucial, because it instils a sense among the population that their decisions are proper and must be obeyed, reducing the amount of force used by officers to achieve people's compliance.Footnote ²¹ Such behaviour is also critical because the less the police employ public use of force, the less likely it is that the government will find itself justifying some abuses. Hence, gross police misconduct may have a direct influence on political stability not only because it affects public safety and order, but because it showcases the regime's willingness to abuse its power. As David Bayley says: ‘what the police do affects the character of government’.Footnote ²²

However, scholarship on the impact of police on legitimacy has only focused on the immediate criminal-justice institutions and has overlooked the general political sphere because the prevailing scholarship emphasises the administrative role of maintaining order, but neglects the function of regime's power agents. Since most studies have been conducted in Western consolidated democracies, authors seem to assume that police actions are important to the extent that they lead to citizens' acceptance of direct police authority, when in fact police actions may also contribute to the maintenance of political order by creating a sense of loyalty to the fundamental institutions of the regime.

Nonetheless, the literature on attitudes towards the police has contributed to an understanding of the importance of procedural factors in the creation of legitimacy for the police. In contrast to studies of political legitimacy, which tend to underscore the importance of outcomes (i.e. how effective the regime is in riding out economic crises, generating well-being and fighting crime), the predominant research on police legitimacy highlights the significance of procedures in the creation of people's support for the police. Tyler and Sunshine have insisted that the most decisive factor in producing police legitimacy is whether citizens perceive the exercise of authority as being fair.Footnote ²³ This occurs regardless of the effectiveness of police performance. More important than the outcome of police work, is the way in which police forces interact with the population.

However, in an interesting study in Ghana, Tankebe concluded that Tyler's argument had little validity in the African context. There, he argues, people cooperate with the police not because they perceive fair-mindedness in police treatment but because they believe in the police's capability to deliver results in the fight against crime, even in the face of blatant abuse and corruption. For Tankebe, effectiveness is more important than procedures in generating police legitimacy in Ghana.Footnote ²⁴ But back in Latin America, in an empirical study of police forces in Mexico, Daniel Sabet claimed some validation of the Tylerian thesis. He found that police corruption is the strongest single predictor of lack of police legitimacy over crime victimisation or temporal evaluations of security. In other words, for many Mexicans, police procedures regarding transparency are more important than police outcomes.Footnote ²⁵

This discussion is useful in advancing my argument because it suggests that police misconduct might be more decisive in influencing political legitimacy than levels of crime and insecurity. In other words, I am hypothesising here that since direct police abuse shatters any sense of fairness, outrage at police misconduct – whether perceived or experienced – may be more significant in reducing citizens' loyalty towards the overall regime than failures in tackling crime.

Police Misconduct

Maurice Punch identifies three broad categories of police deviance: namely, corruption, misconduct and crime.Footnote ²⁶ While the term police misconduct usually refers to violations of institutional regulations, police corruption is frequently employed to distinguish police actions that illegally seek economic gains. The category of police crime is less used and refers to events in which police officers not only infringe rules, but do so for criminal purposes.Footnote ²⁷ Depending upon analytical objectives, these categories can be expanded to differentiate the many forms in which police officers violate laws. Here, to simplify my analysis on police deviance in Central America, I will adopt police misconduct as the term that includes all types of illegal police behaviour: it suggests the violation of rules that ultimately are abuses of police authority.

However, there is another conceptual matter to consider before examining the issue of police misconduct in Central America. That is, the extent of irregular behaviour within police institutions. Punch argues that, in contrast to the predominant view in the literature of corruption, which focuses on ‘low-level’ deviance, police corruption is in many cases the result of ‘either gross failure in the system to perform adequately or systemic rule-bending to achieve formal or informal institutional ends'.Footnote ²⁸ In several Latin American countries with authoritarian pasts, such as in northern Central America, cases of police misconduct do not appear to be isolated cases of individual deviance, but systematic forms of abuse perpetrated by entire units and organisations acting under the justification of tackling violent crime. Moreover, recurrent police misconduct is not only related to organisational variables, as is frequently argued, but mainly to structural factors. As Costa explains when analysing the case of the authoritarian Brazilian regime, ‘the ability of the police to act freely and abuse their authority was a necessary condition of the prevailing structure of domination, which explains the historical weakness of the institutional mechanisms to control police activity’.Footnote ²⁹ In some countries, the onset of democratic rule did little to transform these relations of domination in the streets, and the pervasiveness of police corruption under the era of authoritarian regimes continued during post-transition regimes. The question is, then: does that continuity in police practices have some effect on the legitimacy of the relatively new political regimes?

Police Misconduct in Central America

Police misconduct is a critical problem in Central America. In Honduras, the situation is so dire that all the officers from the national investigative police unit were suspended in 2013 over charges of corruption and criminal involvement, as repeated attempts of police reform failed.Footnote ³⁰ Reports by monitoring organisations have denounced the participation of police forces in cleansing squads and extra-judicial killings of youth in Guatemala, and Honduras.Footnote ³¹ The problem of police brutality became apparent in Honduras during the events surrounding the 2009 coup d’état, and human rights organisations have been raising the alarm about an increase of police abuse since the coup.Footnote ³² In Guatemala and Nicaragua, top chiefs have been charged with links to organised crime groups.Footnote ³³

Yet as part of the political transitions, all police institutions in Central America were either reformed or created from scratch in the last 20 years. In El Salvador and Guatemala, entire public security apparatuses, which had been in the hands of the military, were dismantled in order to create new civilian-ruled institutions.Footnote ³⁴ In Honduras, the 1990s transition was completed only after the military were removed from the police and the internal security agencies.Footnote ³⁵ In Nicaragua, the transition to electoral democracy came about with the 1990 elections, which separated the Sandinistas from power and pushed them to boost reforms in the police and the intelligence apparatuses.Footnote ³⁶ Only in Costa Rica, police reform emerged as an autonomous process from political transformation, and it focused on the integration of a human-rights-based doctrine. There, reforms came as a result of mounting demands from the population for improvement in the conditions of public security.Footnote ³⁷

The seemingly wide variation in the levels of police misconduct in the region and their impact on citizens' attitudes towards the political regime (whether a democracy or not) can help us to unravel the importance of police in post-transitional societies, and shed light on the difficulties that some of these countries have in building their democratic regimes.

Data, Hypotheses and Variables

In this article, I use a selection of variables from the 2008 AmericasBarometer Democracy Surveys that tap perceptions and events regarding police behaviour, and test their impact on attitudes related to the political regime. I empirically test if diverse forms of police transgressions affect what Easton called specific and diffuse support for the regime.Footnote ³⁸ The AmericasBarometer Democracy Survey is unique in its scope, nature and commitment to the scientific method, and operates with the assumption that people's attitudes and experiences are valuable inputs for political analysis.Footnote ³⁹ The surveys from which these data are drawn were carried out during the first half of 2008. In each country, stratified probabilistic national samples were designed to precisely represent the population. Approximately 1,530 adult citizens over the age of 16 were interviewed per country, in a total of 7,649 interviews.

The main argument of this article is that police behaviour affects political support at two levels: the government's legitimacy and the overall regime's legitimacy. Since police behaviour is a broad concept, I have limited the scope of the study and focused on police misconduct for two reasons: one conceptual and one methodological. Conceptually, police abuse represents the opposite of the fair treatment that police forces are supposed to bestow under democratic regimes. This facilitates the measurement of police misconduct's impact on political legitimacy as even in countries with a long history of authoritarianism, the population would recognise abuse as a repudiation of fairness.Footnote ⁴⁰

In operational terms, I hypothesise that police misconduct reduces people's support for the specific and diffuse institutions of a regime. To tap specific support, I constructed a single indicator averaging the results of five separate questions that collect opinions about government performance. Two questions picked up opinions about the work of the government in fighting poverty and unemployment;Footnote ⁴¹ one question referred to the role of the administration in granting public security,Footnote ⁴² another to the role of the government in fighting corruption, while the last one addressed the protection of democratic procedures.Footnote ⁴³ Each of those questions offered a scale of responses ranging from one to seven that was later converted into a 0 to 100 continuous scale. Scores close to 100 represent positive attitudes towards the government, while scores approaching 0 mean negative opinions. The reliability indicator for these questions turned out to be high, with a Cronbach's alpha of 0.908 for the whole sample.

To tap diffuse support for the regime, I employed an indicator drawn from the scholarship on political culture in Central America. This indicator was developed by Mitchell Seligson when he analysed sources of stability in the isthmus during the 1990s.Footnote ⁴⁴ It incorporates an index of five questions that have been extensively used in the AmericasBarometer. In the questionnaire, these questions offered a scale of seven response choices that reflect the opinion of the interviewee on the basic institutions in the political system. In order to facilitate the analysis, the scale was also transformed into scores averaging from 0 to 100, in which an average close to 100 represents a high level of support for the system. The items collected opinions about trust in the justice system, respect for political institutions, pride in living under the country's political system, the belief that the political system protects the basic rights of citizens, and the belief that the political system should be supported.Footnote ⁴⁵ A Cronbach's alpha test showed a satisfactory level of consistency among these items (0.7830).

Police misconduct can take different forms. In this article, I operationalise three types of police misbehaviour: police corruption, police abuse and police participation in criminal activities. Often, these forms of police misconduct are related. An officer involved in criminal activities may also be a corrupt official prone to abuse and to extorting the public. Corruption is the most common form of police misconduct in many countries and it was measured by asking citizens about bribe victimisation: ‘Has a police officer asked you for a bribe during the past year?’ Police abuse was measured using the following item included in the 2008 survey: ‘In the past 12 months, has any police officer abused you verbally or physically or assaulted you?’ Finally, people's perceptions of police involvement in criminal activities are a good way to measure police misconduct, as this not only picks up perceptions of offences perpetrated by police officers, but can also indirectly reflect the degree of corruption and abuse that people face from their local police.Footnote ⁴⁶ Hence, the survey included the following question, which directly tapped into perceptions of police involvement in crime: ‘Some people say that the police in this neighbourhood protect people from criminals, while others say that it is the police that are involved in crime. What do you think?’ ‘The police protect people or the police are involved in crime?’ The three indicators of police misconduct were combined into a single cumulative index of police misconduct which represented the number of types of misconduct reported by the respondent, hence the index ranged from zero to three. A score of zero represented people who did not suffer or witness any form of police misconduct, whereas three reflected those who faced all types of events.

In addition to the composite variable for police misconduct, I included other explanatory variables in the analyses. Following the literature about the impact that perceptions of the economy has on support for the basic institutions of the regime, I added a variable constructed from four items that collect opinions about the economic situation.Footnote ⁴⁷ Two of the items tap individual economic situations whereas the other two refer to the country's economy.Footnote ⁴⁸ Rotated factor loadings for these four items formed a single factor with a Cronbach's alpha of 0.728. Positive responses to all questions would yield higher scores, whereas negative responses would produce low marks in the variable, coded on a scale from 0 to 100. Based on the theory, I expect that people who have favourable opinions about the economy will have more supportive attitudes towards the political regime.

Previous studies conducted in Latin America have found that perceptions of insecurity and direct crime victimisation affect political support.Footnote ⁴⁹ Crime and insecurity were measured using a pair of items. One item picks up events of crime victimisation that the respondent experienced in the preceding 12 months. The second question addresses perceptions of insecurity and whether a person feels safe or unsafe in his or her own neighbourhood.Footnote ⁵⁰ Since crime and common violence represent a direct threat to the well-being of citizens, I anticipate that people who have been victims of crime, or who have strong feelings of insecurity, will tend to express less support for the regime in both specific and diffuse dimensions.Footnote ⁵¹

People's political ideology can play an important role in enhancing or reducing support for public institutions.Footnote ⁵² I measure ideology based on a ten-point self-placement scale. People who placed themselves at one to three identify with the left, those who scored from four to seven see themselves at the centre of the ideological spectrum, and people who identified as leaning to the right positioned themselves between eight and ten.Footnote ⁵³ For the analysis, I created dummy variables for each ideological orientation – left, centre and right – and employed the dummy for the centre as the reference category. Additionally, studies have shown that members and electoral supporters of the incumbent party tend to display higher political legitimacy than the rest of the population.Footnote ⁵⁴ Hence, I created a variable that denotes whether people have voted or not for the sitting president. For instance, in the case of Guatemala, I identified respondents who said they voted for Álvaro Colom's coalition and I expect that they will consistently score higher for measurements of political legitimacy than people who did not vote for him in the 2007 elections.Footnote ⁵⁵ I anticipate the same type of association for every country in the sample.

To test my proposition that police misconduct erodes specific and diffuse political support, I conducted a series of multivariate analyses using ordinary least square (OLS) procedures, taking into account the survey sampling design, with all the variables and indicators depicted above.Footnote ⁵⁶ In addition, I included demographic variables, such as gender, age, education, level of wealth, and exposure to news outlets, as controls. I contend that police behaviour is so important to the provision of public endorsement of the regime that its effects can be detected even after controlling for relevant demographic and intervening variables (see Appendix Table for descriptive statistics of the variables used in the analyses).

Police Misconduct and Regime Support in Central America

The initial results paint an interesting picture of police misconduct in Central America. In regard to perceptions about police involvement in crime, the data show a considerably high percentage in three out of the five countries of the isthmus (see Table 1). In Guatemala, El Salvador and Honduras, more than 45 per cent of the interviewed populations thought that their local police were involved in criminal activities. The data are especially disturbing for Guatemala, where two-thirds of the surveyed population viewed their local police as corrupted by crime in 2008. Only Nicaragua and Costa Rica have negative views of the police below 31 per cent. These results concur with reports that show more troubling police institutions in northern Central America. Police abuse – that is, the excessive use of physical or verbal force – presented a different image. Data do not seem to convey the image of a pervasively abusive police. In the regional average, only 4 per cent of Central Americans reported having been victims of police abuse in the last 12 months before the survey; however, the results show a relatively high level of police abuse in El Salvador (8.3 per cent) in comparison with the rest, especially Honduras (2.8 per cent). Finally, police corruption, measured as street-bribe victimisation, affected nearly 10 per cent of Central America's citizens. Little variation occurs between Honduras, Nicaragua and Costa Rica; in Guatemala, however, corruption victimisation by police officers reached more than 11 per cent.Footnote ⁵⁷Table 1 also displays the Latin American average for each indicator of police misconduct, drawing from LAPOP data. The comparison between the Central American and Latin American averages suggests that levels of police misconduct in this region fall near the continental averages.Footnote ⁵⁸

A quick look at the results for legitimacy also provides a sense of where Central American countries stood in terms of political support in 2008. Not surprisingly, given its history of democratic stability, Costa Rica showed the highest levels of both kinds of legitimacy: support for the actions and policies of the current government, as well as support for the regime in general. In the other countries, levels of diffuse political support are quite similar but remain below 52 points (on a scale of 0 to 100), signalling a rather low public endorsement of the Central American systems. In terms of specific support, country averages show more variation among nations. In Honduras and Nicaragua, opinions about the government are the least positive. The case of Honduras is particularly interesting, because when the AmericasBarometer survey was conducted, the crisis that would culminate in the coup d’état had already begun, and symptoms of instability in the political culture could already be detected.Footnote ⁵⁹ The relatively favourable level of specific support in Guatemala, in comparison with other countries, is probably the result of the then-new government administration, which brought a centre-left coalition to power for the first time after the end of the armed conflict.

Source: AmericasBarometer by LAPOP.

Figure 1. Support for Political Regimes in Central America, 2008 (Averages 0–100)

Do these views and events relating the police to abuse, corruption, and crime have an impact on legitimacy? I first examine the results about specific support for the government and then move to the more general implications regarding the overall regime.

Table 1. Police Misconduct in Central America, 2008 (in %)

Source: AmericasBarometer by LAPOP.

Specific Political Support

Table 2 shows the OLS regression coefficients and linearised standard errors using the three-item based indicator of police misconduct (police corruption, police abuse, and perception of police involvement in crime) as a predictor, and the five-item scale of government assessment as the dependent variable. I have fitted models for every country in the sample with the variables depicted in the previous section.Footnote ⁶⁰ I introduced controls for standard demographic and socio-economic variables because support for the political regime can be a function of those variables. Furthermore, as explained above, I included variables that can affect public opinion about the government while interacting with measures of police behaviour. Variables such as exposure to news outlets, crime victimisation and political ideology not only can impact on support for the regime, but can also influence the way in which police actions are perceived. For instance, given the history of repression by police forces against leftist movements in Central America, people self-identified with left-leaning ideology may exhibit less support for government institutions than people identified with the right.

Table 2. The Impact of Police Misconduct on Support for Government in Central America, 2008

* p ≤ 0.05

** p ≤ 0.001

In general, the results of the first model support my hypothesis that police transgressions erode support for governments in Central America. As expected, police misconduct is related to lowered levels of specific support in every country. However the only country in which the relationship is not statistically significant is Honduras. In the rest of the region, corrupted behaviour by the police is statistically associated with significant decreases in the index of support for the government, especially in Guatemala and Nicaragua. For example, perceiving that police are involved in crime and having been a victim of the two forms of police misconduct in Guatemala is associated with a 19-point decline on the scale for support of the government (Coefficient = −6.39). In El Salvador, the model accounts for the highest explained variance (R² = 0.343) and shows that despite the strong association with economic perceptions and political allegiance, police wrongdoing has a negative effect on public support for the government. The results also indicate that even in Costa Rica, with the lower substantive effect among the statistically significant coefficients, the decline of specific support associated with police misconduct is nearly 11 points, ceteris paribus, suggesting that the image of the government is appreciably affected by police misbehaviour. These results are in line with other studies showing that police abuse corrodes public confidence in immediate criminal-justice institutions and the government.Footnote ⁶¹

However, police misconduct is not the only variable associated with changes in support for the government. According to the results, perceptions of economic performance have a positive and strong association in every country. These findings hardly come as a surprise, as the economy has repeatedly been singled out as a crucial factor for the survival of any regime. Political ideology is also strongly associated with specific support for the institutions of all countries in the sample. In Guatemala, El Salvador and Costa Rica, people who identified themselves with the right consistently showed more support for their respective government administrations in contrast to people identified with the centre. In fact, right-leaning respondents exhibited higher levels of specific support than individuals identified as centrist in the five countries – although the relationship was not statistically significant in Honduras and Nicaragua. In contrast, left-leaning people displayed less support for the administration than centrists in El Salvador, Honduras and Costa Rica. In Nicaragua and Guatemala, respondents self-identified with the left showed substantive and significantly more support for the incumbent government. Both countries were governed by left-leaning or centre-left forces in 2008, and that may explain the favourable attitudes from that side of the political spectrum. In Guatemala, it is noteworthy that a left-leaning government coalition attracted specific support from both ideological poles. In sum, these findings attest to the particular relevance of ideology in people's attitudes towards the political leaderships in the Central American countries. They hint at the legacies of the ideological conflicts of the past. In the same way, having voted for the incumbent in the last election is a strong predictor of support for the government across the region.

Exposure to news outlets is associated with increased support for the government in all countries, although the relation is statistically significant in Guatemala and Nicaragua only. In these countries, frequent exposure to news outlets (newspapers, news radio and television) translates into public endorsements for the government. The effect of news outlets on government legitimacy is even stronger than perceptions about performance of the economy in Guatemala. Perceptions of insecurity appear related to government legitimacy only in Guatemala, whereas crime victimisation has no significant effect in any country at all. It is puzzling that in the two countries with the highest homicide rates in the world (Honduras and El Salvador) the effects of crime victimisation are the most negligible. These findings suggest that in Central America, gross police misbehaviour might be more important than insecurity in reducing government's legitimacy. Demographic variables, such as gender, age, years of education, and level of wealth, exhibited no consistent impact on specific political legitimacy. Young people, for instance, are more supportive of government institutions than the rest of the population in Nicaragua, but in Costa Rica youths are seemingly less supportive. Education was substantively and significantly related to our dependent variable only in El Salvador.

All things considered, despite the empirical influence of variables such as perceptions of the economy, ideology, and political allegiance, police misconduct still has substantive effects on specific support for most Central American regimes, to an even larger degree than perceptions of insecurity and experiences of crime victimisation. In general, where police are perceived to be involved in crime, accept bribes and abuse the population repeatedly, prospects for a population supportive of the government are lower than in places where the police do not regularly abuse the citizenry. This association is especially relevant in Guatemala, El Salvador and Nicaragua, countries that underwent transitions from civil conflicts in the last two decades. However, we should not underestimate its importance for Costa Rica either. The historical low levels of abuse in this country should be understood as part of a virtuous cycle, in which governments have enjoyed higher levels of support precisely because of their institutional strength. However, the results show that, even in Costa Rica, the police can further the erosion of public allegiance to government institutions if people perceive officers blatantly breaking the rules.

Diffuse Political Support

But if police wrongdoing affects views on government performance, what about the more diffuse and enduring political institutions of the regime? Do police actions have any relationship to the degree to which people embrace the overall political regime? Table 3 offers a view of the association between police misconduct and legitimacy of political regimes in Central America. The findings validate my hypothesis that police deviance is negatively and statistically associated with system-level legitimacy in all Central American countries. In this case, the association is stronger in Costa Rica (Coefficient = −5.71) than in any country in the region. There, a victim of police abuse and corruption, and who also believes that local police have colluded with crime, has a score of political support 17 points lower than somebody who has not faced any police wrongdoing. Even in El Salvador, where the effect is rather modest but still statistically significant (Coefficient = −3.75), victims of all forms of police misconduct can be expected to score, on average, 11 points less on the diffuse legitimacy scale than people who have not faced any police offences, ceteris paribus.

Table 3. The Impact of Police Misconduct on Diffuse Support for Political Regimes in Central America

* p ≤ 0.05

** p ≤ 0.001

The effects of police misconduct on the legitimacy of the overall regime might be deemed negligible if we compare them with the effects of the variables of political ideology;Footnote ⁶² however, even as I included those factors and kept them constant into the model, police misbehaviour remained a significant variable in the equation for diffuse support. In other words, police matter for the political system, and they count not only because they can erode immediate support for the government, but also because they reduce long-term attitudinal reserves of legitimacy for the whole system.

We can find further evidence that illegal police actions can be an issue for the stability of the regime by looking at the results of the model in Honduras. In this country, police misconduct had a significant effect on diffuse support for the regime, while assessments of the economy did not.Footnote ⁶³ This is noteworthy considering that police deviance did not have a statistically significant relation with specific government's legitimacy. It suggests that police wrongdoings may not affect the support for the sitting president, but they reduce the loyalty towards the whole system. Honduras's findings are also striking because by the time the survey data were collected, Honduras was moving towards political collapse. The coup resulted in the reinstatement of a transitory regime that would conspicuously tolerate police abuse.Footnote ⁶⁴

Despite the already soaring crime rates in Honduras in 2008, the variables that tap into crime victimisation and insecurity failed to show any statistically significant association with broad legitimacy. This finding questions the assumption that high levels of violence would always erode support for the regime. Instead, it seems that widespread police deviance was more important in plunging Hondurans into alienation from the political regime than any of the crime-related factors. Nevertheless, in Guatemala, El Salvador, and Costa Rica, perceptions of insecurity were significant factors undermining people's support for the overall regimes. In Nicaragua, direct crime victimisation has a greater effect on diffuse legitimacy of the regime than the public's sense of insecurity, but police deviance still remains an important and significant factor.

In sum, it is necessary to add police transgressions to the list of variables that debilitate people's support for the political regime. All these findings confirm that police violence is a problem for the legitimacy of immediate government institutions. But more importantly, they also indicate that police wrongdoing can be an obstacle to prospects for regime stability. They go directly to the core of the Weberian state function: namely, the monopoly on the use of legitimate force. Abuse of police force questions the legitimate dimension of police activity and impairs the development of a political culture capable of sustaining the fundamental institutions of the regime.

Potential Limitations of the Data

As with any survey that collects public opinion, there may be some potential limitations of the data affecting the findings. First, respondents may not understand police abuse in the same fashion across the region. The way people view police misconduct might be different in democratically stable Costa Rica than in Guatemala, where police have been notoriously corrupt and repressive. Second, it can be argued that my measure of police misconduct might be problematic for two reasons. First, because it may not be reflecting the levels of abuse by Central American law enforcement institutions, but only picking up the public image of criminal-justice and government organisations. Thus, when people report that police officers are involved in criminal activities, they might be iterating a common discourse about government authorities in general, and not only about actual police behaviour. Second and most importantly, because my indicator of police misconduct is conflating perceptions of police criminal involvement and actual experiences of police abuse, it is jumbling up two different dimensions in one single measure.

In spite of these potential issues with the data, there is evidence that they do not greatly affect the empirical validity of the findings. For instance, despite the possible differences between country and country, the questions used to gauge police deviance were very specific. They collected actual events of victimisation during a defined time period. Questionnaire items gathered people's experiences in their dealings with local law enforcement officers, and those items formed an important part of the police misconduct variable. The only question that actually collected perceptions referred to the involvement of neighbourhood police in criminal enterprises. But even in this case, the question pointed to police officers working in the community where the respondent lived, not to the police in general. Thus, my indicator of police misconduct ultimately picks up activities that affect the relationships citizens are having with their law enforcement agents. Moreover, the results for the overall misconduct indicator correspond to views of independent organisations about corruption among Central American police forces. Reports by the UNDP, the Washington Office for Latin America and others indicate that by 2008, police in Guatemala and Honduras were significantly more affected by problems than in El Salvador and in this country more than Nicaragua and Costa Rica.Footnote ⁶⁵

Still, one concern about collecting perceptions about police abuse is that they may belong to a broader system of attitudes regarding political support. Therefore, when analysing the relationship between police misconduct and political support, we might be correlating two manifestations of political support and not the impact of actual police misconduct on legitimacy of the regime. Most seriously, by combining perceptions and direct experiences in a single indicator, my measurement of police misconduct might be overstating the importance of perceptions of police deviance on the political sphere.

To address those issues, I reran the OLS regressions breaking down the variable of police misconduct into its three constitutive items: direct abuse, corruption victimisation, and perception of police involvement in crime. First, I regressed specific political support on the separated indicators of abuse, corruption and police crime.Footnote ⁶⁶ Results indicate that in all countries, the three expressions of police misconduct (experienced and perceived) are negatively associated with government legitimacy. However, statistical significance in the associations varied per country and type of misconduct. Perceptions of police crime were statistically significant in Guatemala, El Salvador, Nicaragua, and Costa Rica; whereas victimisation of police corruption appeared significant in Nicaragua only. Police physical abuse turned out significantly associated with specific legitimacy in Honduras and Nicaragua.

Although at least one type of police misconduct was statistically related with reduced legitimacy for the government, these results underline differences between perceptions and experiences, and also variations from country to country when it comes to the development of political support. As studies have pointed out, in some cases, perceptions of police behaviours are more relevant in shaping attitudes towards the police than direct experiences with police brutality and corruption.Footnote ⁶⁷ In this case, results show that in Guatemala, El Salvador, and Costa Rica, perceptions are more important in sculpting government legitimacy than direct negative contacts with police officers.

However, evidence of the effects of direct police victimisation appears more clearly regarding diffuse support for the regimes (see Table 4). In Guatemala and Costa Rica, experiences of police abuse significantly reduce diffuse legitimacy, whereas in Nicaragua and Honduras, it is direct police corruption that emerges as the significant variable driving down public support for the regime. Furthermore, actual experiences of victimisation by the police are more important than perceptions when reducing support for the overall regime in Guatemala and Costa Rica. In the latter, individuals who were abused by law enforcement scored on average nearly nine points lower on support for the regime in contrast to those who were not mistreated. These results accompany data showing that perceptions of police involvement in crime reduce, as expected, political legitimacy in all countries of the region. The findings, indicating that perceptions of police misconduct have a wider effect than direct experiences, echo the conclusions of Rosenbaum et al., who argue that in several communities vicarious experiences with policemen are more important than direct experiences. Still, as some country models show, under some contexts direct police victimisation can inflict more damage on political attitudes than perceptions.

Table 4. The Impact of Different Forms of Police Misconduct on Support for Political Regimes in Central America, 2008

* p ≤ 0.05

** p ≤ 0.001

Expectations, Outcomes and Police Behaviour

The examination of different forms of misconduct allows us to understand the consequences of police actions on support for the regime. Even with many controls and robust intervening variables, at least two indicators of police misconduct remained as significant factors in relation to diffuse legitimacy in most countries. In this study, I am providing evidence that, with the right indicators, it is possible to measure the political implications of police actions, whether perceived or directly experienced. The consequences extend beyond public trust in police institutions themselves: they spill over onto people's allegiance to the political order.

Yet differences in the findings from country to country yield new questions and open interesting discussions. Why is police abuse significant only in Guatemala and Costa Rica, but not in the rest of Central America? Why is police corruption especially important in Nicaragua and Honduras? More generally, why do violence and insecurity variables not make a dent in political support in crime-ridden Honduras, but do affect legitimacy in relatively peaceful Costa Rica and Nicaragua?

These findings show important nuances in the way political legitimacy – whether specific or diffuse – is adjudicated from country to country. Although perceptions of police misconduct are critical in almost every Central American context, the effects of direct police brutality and bribery seem to be more dependent on the context in which they occur. Therefore, we need to revisit the arguments that underline the importance of context in the formation and preservation of political legitimacy. From the literature of legitimacy, we know that contexts shape expectations towards institutional behaviour.Footnote ⁶⁸ Legitimacy, after all, emerges from the interaction between expectations and perceived and actual experiences so when people perceive and endure police behaviour, they are also assessing the political system against the backdrop of their existing views on the police.

Furthermore, according to Steven Van de Walle and Geert Bouckaert, the traditional paradigm that better-performing public services will always lead to more legitimacy needs to be qualified.Footnote ⁶⁹ Yet again, we need to take into consideration expectations towards the police and government agencies because not all expectations about institutional behaviour revolve around outcomes but also around procedures and treatment.Footnote ⁷⁰ In this study, I have controlled several intervening factors (demographics, perceptions of the economy, ideology, etc.) when testing the effects of police misconduct, but I have not measured citizens' expectations when it comes to police performance. Varying expectations about institutional performance and interaction with the community may be at the root of the differential impact of the types of police misconduct and security issues. For instance, relentlessly high levels of crime and increasing political turmoil in Honduras may have rendered the public more cynical and more accustomed to tolerating insecurity and criminality as features of their daily social life than in Costa Rica, where its low levels of violence and its political stability may have made citizens more sensitive to institutional disruption.Footnote ⁷¹

The recent literature on crime and legitimacy in Latin America has led us to believe that extremely high levels of crime and insecurity will automatically shatter political support for regime institutions, as if all that mattered was how effective the regime is at keeping crime and insecurity at low levels. This assumption is problematic for it taps into remnants of the authoritarian era and ignores the importance of new democratic rules for the creation of citizen's expectations. Authoritarian regimes built their legitimacy around public security on the notion that they were better at controlling crime because they were extremely severe with criminals. Some of that rhetoric survives nowadays in criticism of universal regard for human rights. However, democracy brought new forms of understanding, generating, and maintaining legitimacy. People with memories of prior abusive institutions may as well appreciate a fairly-behaved police. In this post-transition era, legitimacy may be preserved insofar as police agents do not blatantly abuse the people they are supposed to protect.

Collective political history transforms the context under which people create their expectations about the role of police. In some Central American countries, the arrival of electoral democracy created hopes that authorities would be able to keep social order without resorting to brutality and illegality. These expectations determined which behaviours would become more salient than others in shaping political legitimacy. The way police behave matters, even if they are not always successful in reducing crime and insecurity. Here, I have focused on the dark side of police conduct because it underlines the importance of observing the rule of law, but elsewhere I have shown that in El Salvador and Panama, police engagement with residents in non-confrontational ways can boost political legitimacy to a larger degree than residents' experiences of victimisation and their perceptions of insecurity.Footnote ⁷²

All in all, whether democracy is consolidated (as in Costa Rica) or not (as in the rest of Central America), police misbehaviour and persistent perceptions of police collusion with organised crime are having devastating yet inconspicuous effects on all levels of political legitimacy. Police wrongdoings can effectively subvert the prospects of democratic governance and affect the cultural reserves of political stability. To be sure, a law-abiding police force is not a sufficient condition for political stability or for democratic governance, but a corrupt police force may be a significant deterrent to democratic consolidation. Unfortunately, considering that the problem of police misconduct is widely systemic in several countries, we can expect that support for the political regimes will continue to falter in some Central American societies.

In this chapter, we have dealt with a phenomenon long underestimated by the academic literature. Political scientists have been concerned about the political effects of the economy, and just recently they have turned to crime and insecurity. But by focusing on the economy or on levels of crime as proxies of institutional performance, they have overlooked another fundamental issue: whether by searching legitimacy exclusively through results, government officials have turned the state into another perpetrator. The findings of this chapter suggest that legally-bound police behaviour is important for the preservation of a set of attitudes conducive to democratic governance. By tolerating systematic abuse and corruption, police authorities contribute to demolishing the foundations of democratic order; they open the door to informal security responses and delegitimise the role of government and the state. Hence, the issue of widespread police abuse and corruption is ultimately a political problem as it touches on the foundations of governance and stability.