Police legitimacy is an important topic of criminological research, yet it has received only sporadic attention in societies where there is widespread police corruption, the position of the police is less secure, and social order is more tenuous Jackson et al. 2014. The present article – divided into four parts – is linked to a greater project entitled: “Building Democracy Daily: human rights, violence and institutional trust”. The first part addresses some critical issues regarding confidence, trust, and (subsequent) obedience, looking at the relationship between those notions and youth from a theoretical perspective. The second and third parts are dedicated to the empirical part of the study, namely the methodological framework and the data analysis, respectively; while at the last part includes the limitations of the empirical research, as well as the conclusions. Nevertheless, before proceeding to the first part of this study, some important background information on institutional violence in Brazil is provided.

Brazil’s police are notoriously violent. Brazil is also a country with a long history of social inequalities, racial biases, and economic disparities. Without doubt, poverty and social injustice are important factors that help explain the context in which police violence arises. This is definitely a contributing factor to Brazil’s soaring rates of homicide, being amongst the highest in the world. According to official figures from the Brazilian Forum of Public Security (Fórum Brasileiro de Segurança Pública), every year Brazil’s police are responsible for at least 2,000 deaths. In other words, at least six people are killed by police officers in Brazil every single day, four of which are young people aged 15 to 29 (Roque, 2014). As shocking as it seems, this figure is probably an underestimate, as most states refrain from disclosing such information. Looking at homicide figures from a comparative viewpoint, in the period 1995–2015, São Paulo’s military police killed 11,358 people over the last 20 years.1 Simply put, the Brazilian state of São Paulo2 saw more police killings than there were in 50 US states. This fact alone makes the São Paulo state police 53 times more violent than those in the entire US.

Therefore, not coincidentally the present research study also took place in Sao Paolo – a mega-city marked by repressive police interventions that are decimating a significant part of a generation of young, (usually black), and poor men living in communities so deprived and marginalized they are hardly visible (Ciconello, 2015). At the same time, the victims’ families wait from a failing criminal justice system to deliver justice and reparations for human rights violations; however, such killings are rarely investigated and brought to justice. Some of those deaths, along with other acts of violence, are caused in the course of massive raids into favelas (shanty towns). These raids are often designed to carry out legitimate police actions, such as the apprehension of criminal suspects. But in the course of these actions, police have repeatedly engaged in unjustified fatal shootings of criminal suspects and inappropriate use of excessive force. The killings by on-duty police officers are often registered as “resistance followed by death”,3 which prevents independent investigations and shields the perpetrators from the civilian courts. But even when such investigations do take place, their effectiveness is not guaranteed. For example, when reviewing the status of 220 investigations of police killings opened in 2011, it was found that after four years, only one case led to a police officer being charged. As of April 2015, 183 investigations were still open (idem).

One important factor that could explain police violence in Brazil is, perhaps, the authorization for police officers in certain states to carry a second weapon while on duty, not licensed to the military police and owned by the particular officer. This authorization dates back in 1995 and is based on the ambiguous fact that “military police officers must, necessarily, be technically, physically and emotionally prepared for the full exercise of their mission” (Human Rights Watch, 1997). This justification, however, opens the door to potential abuse (eg. extrajudicial executions, fabrication of evidence), providing that corruption is, generally, a persistent hindrance in Brazil.

Although it is obvious how the aforementioned policy may stimulate violence, the most crucial factor behind the legacy of police violence against criminal suspects and innocent civilians is definitely the persistent impunity that “protects” those officers who commit grave human rights abuses against this class of victims. Impunity results from the general inefficiency of the Brazilian judicial system, and deteriorates when the victims are poor favela residents with possible involvement in crime and the suspects are police. The second most crucial factor is that violence does not impact Brazilian society equally. Even though they are affected in massive numbers, young people residing in poor areas have become invisible in Brazil, either because residents of their neighborhoods have become used to witnessing violence or because the problem is out of sight (and out of mind) of those who can take action to alter this situation. Prejudice and negative stereotypes associated with the favelas have a key role in perpetuating this violence. The war against the youth, the black, and the poorest areas in Brazil; and moreover, against those individuals who combine all the above.

Unfortunately, cases of brutal police violence and killings in Brazil appearing on international human rights reports4 are innumerable. The main target of the present study – conducted in a country plagued with deep socio-economic inequalities, like Brazil – is to understand the role youth’s opinions have in formation of police’s authority through the investigation of the mutual relationship between legitimacy, obedience and trust. This work aims to collect youth’s perceptions on legitimacy and authority as a way to understand the dynamics among these notions, and their impact on obedience in laws, norms, and regulations. An additional aim is to measure the level of youth’s trust in public institutions – namely the police – and whether they would refer to them according to their motivations for (not) doing so. The results provided here could become a stepping-stone in our efforts to discover the extent of (il)legitimacy of the Brazilian police among those who have been most-affected by their violence, and subsequently underline the urgency to change the paradigm of how the police operate in Brazil.

Crucial Issues among Youth

The issue of confidence & trust

Confidence and trust are considered to be a decisive element of a democratic political system (Almond & Vebra, 1963; Hetherington, 2005). As suggested by scholars, low levels of political trust might have a broader social impact and a negative effect on their willingness to comply with the law or exhibit a law-abiding behavior (Chanley et al. 2000). Within this generic argument it is crucial to determine how institutional trust is developed in the first place. Although previous studies have highlighted various determinants of the development of institutional trust, they remain relatively confined to country-level factors without considering the possible influence of individual characteristics, such as age. Nevertheless, attitudes of political trust – as the perceptions of legality – are already developed early in the course of life (Abramson, 1972; Hooghe & Wilkenfeld, 2008). More specifically, assuming that a discrepancy between law on the books and law in action is manifest at an early age, it could lead children to become cynical about the law’s capacity to resolve conflict and deliver justice, and make them less likely to trust legal authorities and use the law as a resource in the future (Stevens, 2013).

Unfortunately, several past studies have considered the notion that young people are disengaged from the democratic institutions (Henn & Foard, 2012). Yet, youth, and particularly adolescents, remains an ideal category to investigate the trustfulness and trustworthiness5 of institutions, assuming that they lack personal experiences with the political-legal system. Considering that not all youth have first-hand experience with all the institutions (moreover with the police), the impact of their performance might be indirect, and the experiences acquired might be mediated by a learning mechanism (Claes et al. 2012). A basic assumption is that “much of what we know and think we know comes not from personal experiences but from the stories we hear” (Morgan & Shanahan, 1997, p. 5). This argument appears to be valid especially for adolescents who rarely have any first-hand experience about the way the public institutions function (Lauglo, 2013). Their knowledge is rather gathered by the sporadic information they receive either from family, school, peers, and mass media (Rothstein & Uslaner, 2005), or from brief contacts with legal authorities for administrative arrangements (Hohl et al. 2010).

It is also worth noticing that despite the indirect, limited and impersonal nature of young people’s experiences with the police, they are often generalized towards the political system (Easton, 1975). This procedure of trust development among youth significantly contributes to the formation of institutional trust at country level, which is regarded as stable (Newton, 2007; Zmerli & Newton, 2008). Findings on a country level, however, do not inform us about the individual causal mechanisms that might be involved in the development of trust among youth. They fail to provide an answer to how exactly citizens (especially the youth) learn to trust or distrust public institutions.

According to cultural psychologists, the morality and trust that children develop is a function of the culture they are growing up in (Woods, 2013). School is regarded as the main hub where young people learn about democracy and the way public institutions function (Dewey, 1926). It is in the school setting that children have their first direct and extended experience with institutional authority (Sanches et al. 2012, p. 607). At the same time, research evidence demonstrates a positive relation between experiences in classrooms and levels of trust (Torney-Purta et al. 2004), as well as between attitudes towards school authorities and those towards other institutional authorities (Reicher & Emler, 1985; Rigby & Rump, 1981). Assuming that knowledgeable youth are better able than others to judge basic conditions of justice in their country (Rothstein & Stolle, 2003; Pattie et al. 2004, pp. 138–140), the correlation between knowledge and trust in public institutions can still be negative (Torney-Putra et al. 2004, p. 396). On an aggregate level, it has been argued that corruption regimes offer few incentives to their citizens to develop trust in the system (Mishler & Rose, 2001). In reality, it is hard to see how excessive corruption is compatible with legitimate authority6 (Tankebe, 2010). Following this view, Kahne and Westheimer (2006) rightly argue that schools should not be merely concerned about the transmission of knowledge, but they should also embody fairness and procedural justice as a way to prepare citizens for a democratic society.

The issue of compliance to authority

Across the social sciences, exists a widespread recognition of the importance to understand how to motivate cooperation. Compliance may be ingrained in everyday life (Robinson & McNeill, 2008, p. 436), which means that it may be related to habitual behaviors or to the absence of criminogenic etiologies (eg. substance abuse, emotional problems, academic failure, inadequate prevention and intervention) (Allen et al. 1994; Ary et al. 1999; Jessor et al. 2003). Keeping in mind that trust and the perceived obligation to obey do not formulate a single factor (Reisig et al. 2007; Gau, 2011, 2013), there exist various reasons why individuals choose to disobey or comply with the law (Bottoms, 2001). From the perspective of Emler and Reicher (1995, 2005), delinquency is the behavioral manifestation of a broken relationship between youth and institutional authority.

Generally, youth-authorities relationships are characterized as problematic. One does not necessarily need to look to scientific studies to support the view that young people are the demographic group most likely to have a strained relationship with authority figures (predominantly with police). A substantial literature on the socialization process from childhood to adulthood records negative attitudes and an often-conflictual relationship with legal agencies and social institutions (Hinds, 2009). Sometimes it even depicts an “anti-authority syndrome” against all conventional forms of authority (Clark & Wenninger, 1964, p. 488; Schuck, 2013, p. 583), which is strongly linked to the involvement in delinquent acts (Hirschi, 1969; Emler & Reicher, 1987; Loeber, 1996; Levy, 2001; Tarry & Emler, 2007). Young people’s extensive use of public space often attracts the attention and intervention of the police. From a subcultural perspective (Cohen, 1955; Cloward & Ohlin, 1960; Leiber et al. 1998), adolescents exhibit negativity toward the police not necessarily because of negative experiences with it, but because they belong to a peer group that tends to reject authority altogether. Further, not abiding to the peer group’s belief system places them at risk of rejection by peer group members.

Until today, research findings have shown a robust association between legitimacy and compliance with the law (Tyler, 2006b; Jackson et al. 2012a). More specifically, legitimacy has been associated with increased willingness to conform to the decisions of criminal justice agents (e.g. police officers, judges) (Tyler & Huo, 2002), to cooperate with legal authorities (Jackson et al. 2012b), and to comply with laws generally (Mastrofski et al. 1996). It also appears to be a strong relationship between attitudes toward the police and perceptions of police legitimacy (Skogan & Frydl, 2004). Results of a recent study suggest that there is a dramatic decline in favorable attitudes toward the police during adolescence that begins at about 12 years old and subsequently stabilizes around 17 years old (Schuck, 2013). This pattern seems to be general and represents the best average description for all adolescents in the study, regardless of gender, race, or socioeconomic status. In other words, although the socio-cultural context, psychological factors, and peer-networks could be significantly influential in shaping youth’s attitudes (Giordano, 1976; Leiber et al. 1998), the issue of quality of treatment by authorities remains central to the reactions of juveniles (Hinds, 2007, 2009; Crawford, 2009). Indeed, broader community factors may play a key role in the emergence of proactive aggression7 (Fite et al. 2009). As Nelsen et al. (1982) argued, young people’s behaviors are largely shaped by direct experiences, such as treatment received or encounters, rather than generic beliefs about authority in abstract. This argument underscores the importance of positive interactions with authorities early in life, including parents and teachers. Given the strong inter-relationship between attitudes toward the police and law-abiding behavior, prolonged negative attitudes into adulthood may also have a significant impact on the manifestation of negative outcomes later in life (Fagan & Tyler, 2005; Schuck, 2013), which could be even transmitted from one generation to the next (Brunson & Weitzer, 2011).

The Methodological Framework

Whilst the empirical work was absolutely central to the study, the research itself is not simply a report of an empirical project. Rather, it contains a mixture of theoretical reflection on the mutually inter-related notions of authority-legitimacy-trust and empirical analysis. Empirical findings are used to verify or reject past theoretical arguments and assumptions and hopefully generate new discussions. What makes this empirical research different from current evaluation studies on the above notions is the fact that it is not designed primarily to answer questions imposed by governments pursuing particular interests. Instead, it aims to let young people living in a socio-economically disadvantaged area speak for themselves and explain their position about the police, their authority, and trust. The objective is to express their views and opinions, to raise concerns and criticisms, and to bring their unique insights to the ongoing debate.

The hypotheses

The principal hypotheses here are:

  1. The (il)legitimacy of the police influences people’s willingness to obey rules and regulations, as well as their actual conformity to them.
  2. Equality plays an important role in the development of youth’s perceptions of the police performance.
  3. Young people trust the police based on their perceptions about how they are treated, and whether they operate in an effective way.

The sample

The size of the sample depended on the overall number of (early) teenagers between 13 and 16 years old. According to the Brazilian educational system, this age-group corresponds to the last two grades of elementary school II (age 13–14), and to the first two grades of high school (age 15–16). No random computerized selection was needed as all eligible teens where given the opportunity to take part in the study. Their participation was absolutely voluntary and their responses strictly confidential.8 The sample was collected from youth frequenting the “Oliveira Viana (OV)” secondary school, the “José Raul Poletto (JRP)” high school, and the “Eudoro Villela (EV)” secondary/high school, which are the only ones in the neighborhood of Vila Santa Lúcia located in the “heart” of Jardim Ângela. The sample derived by public schools, either State or municipal,9 for comparative reasons. As of October 2014, the OV secondary school had 1.447 matriculated students, of whom 346 were 13–14 years old; while at the JRP high school were 380 students aged 15–16 years with a total number of 1.390. As for the EV secondary/high school, in the period of research it was attended only by the 103 students who needed recuperation classes in order to pass the school year (DRE,10, 2014). All three schools were delivering morning, afternoon and evening classes from 7:00am to almost 11:00pm.

The particular location was not chosen arbitrarily. Jardim Ângela is a district in the south zone of Sao Paolo that is often considered as a separate city due to its numerous divisions into small neighborhoods. According to the latest demographic census of IBGE11 in 2010, the level of illiteracy among children of 5–8 old was among the highest in Brazil with a percentage of 58,6%.12 Analogically high (approximately 2,5%) was also the representation of adolescents from Jardim Ângela in institutions for juvenile offenders, known as “foundation houses (fundação casas)”, in Sao Paolo (SEADE & USP/NEV,13, 1995). Although information on juvenile delinquency trends in Brazil is scarce, in 1996, Jardim Ângela (one of the city’s most disadvantaged districts) was unofficially14 given the title of “the most violent place in the world” (Ponciano, 2001: 113) with an average of 25 homicides per month (Cardia, 2000) committed mainly by male youths between 15 and 24 years old (CEDEC,15, 1996). A year before, in 1995, the “Risk of Violence Map” had showed a homicide rate of 222.2 per 100,000 inhabitants in the 15-to-24-age bracket (Adorno et al. 1998). Today, the criminal phenomenon in Jardim Ângela has changed. Violence has been considerably lowered; though, use and trafficking of drugs has taken its place.16

Distribution and collection

Initially, the one-page questionnaire, printed on A3 paper, was pre-tested in order to identify any inconsistencies and ambiguities. This preliminary check was conducted at the Centre for children and adolescents of a local non-governmental organization named “Social Institute of Santa Lúcia”. The particular Centre provides consultation and assistance to youth in social risk or vulnerable situations.17 Once it was made sure that, language and context-wise, all questions were understandable by the juveniles who volunteered to read them and all consent forms where signed by their parents or guardians, the questionnaire sheets were personally distributed during November and December 2014 along with a brief explanation about the nature and the objectives of the study. Students still facing difficulties with the context, but willing to take part in the research were grouped by age and they were provided the necessary assistance by simplifying the questions. The questionnaires were filled in immediately after being distributed to the students and in continue they were collected in a similar way, hand-by-hand, giving them the opportunity to ask for any further clarifications.

The Data Analysis

All participants were assigned a serial number to ensure confidentiality. In continue, the basic analysis was carried out to produce frequencies, percentages, and appropriate measures of association for all the questions included in the questionnaire. Individual responses to the open-ended questions were grouped together to produce categories of like answers that could then be all scored. Multinomial logistic regression models were created to assess the associated factors with our dependent variables. The dependent variables (Y) were the following: “confidence in the police”, “evaluation of its functions”, “(frequency of) obedience”, and “equality in treatment”. For this analysis, odds ratio with 95% confidence intervals were yielded. Finally, association analyses were performed using multiple correspondence analyses in order to describe the pattern of relationship between categorical variables. The bi-plots were presented and the profiles were described. The statistical software SPSS (v.20) was used to perform all analysis and the level of significance for our tests was 5%.

Demographic questions

The exact number of eligible participants (total number of students aged between 13 and 16 years old) and subsequently, the return rate were impossible to be measured with accuracy. Each school was only able to provide the number of students being officially matriculated, but without necessarily frequenting. In reality, the exact number of students was difficult to calculate even for the school authorities due to the fact that students were constantly skipping classes for no, or unknown, reason. Bearing this into consideration, 644 young adolescents agreed to participate in the research and filled in the relevant questionnaire. In total, 22 students18 refused, were not authorized by their parents/guardians to take part or were facing difficulties to read and write in Portuguese. The majority of respondents were females (n19 = 337, 52.3%20), 16 years old (n = 195, 30.3%) of Brazilian nationality (n = 627, 97.4%) (Figure 1). All juveniles were attending a public school (due to a lack of private schools in the vicinity and scarce financial resources). Almost half of the questionnaires were received from the “José Raul Poletto” school (n = 321, 49.8%), followed by the educational establishments of “Oliveira Viana” (n = 227, 35.2%) and “Eudoro Villela” (n = 96, 14.9%) (Figure 2).

Figure 1 

Age and gender.

Figure 2 


Concerning the family structure, 35.4% (n = 228) of the adolescents were living only with their mother (n = 201, 31.2%) or, less commonly, their father (n = 27, 4.2%). In a small percentage (n = 68, 10.6%), another person was living together with the mother or father in the same house and that being their grandmother/father or stepmother/father. The juveniles often described their relationship with their parent(s) or guardians as “good” (n = 507, 78.7%), and only 7 participants (1.1%) had a bad relationship due to occasional fighting (n = 66, 10.2%) and lack of interaction (n = 26, 4.0%). Although the vast majority (n = 584, 90.7%) of those had siblings – usually up to three (n = 446, 69.2%) – there were few cases with 16, 18 or even 26 brothers and sisters who were not necessarily living under the same roof. In fact, there was often highlighted by the participants that the number mentioned on the questionnaire was corresponding to the siblings they were aware of.

Thematic questions

Hypothesis 1: The (il)legitimacy of the police influences peoples’ willingness to obey rules and regulations, as well as their actual conformity to them.

Trust is widely accepted as being central to institutional effectiveness and legitimacy (Sunshine & Tyler, 2003). Trust in the police is part of the concept of legitimacy, and police legitimacy is a prerequisite of democratic governance (Beetham, 1991). Democratic systems might need even more of it than non-democratic ones, because they are (usually) limited in using violence or coercion and they are bound to build cooperative relations with citizens. Earlier we saw that legitimacy shapes citizens’ reactions to laws, polices and regulations (Weatherford, 1992). It was argued that people do not obey the law because they fear the sanctions, but because they put trust in it. More specifically, if people believe that laws serve the common good, and that the judicial system is both effective and fair, then they are more willing to obey the law (Tyler, 2001, 2006b, 2010). In the present research, the legitimacy of the police was measured according to two criteria: the effectiveness of their function (distributive justice) and their equality in treatment (procedural justice). Cross-tabulating the variables of obedience in laws with effectiveness and equality, it was found that, in both cases, the participants gave an average score. No matter what the juveniles thought on whether laws and orders should be always obeyed or not, the majority of them believed that the police “sometimes” to “almost never” treat people in an equal form (Figure 3) evaluating its function as “almost efficient” to “neither inefficient nor efficient” (Figure 4).

Figure 3 

Obedience & equality.

Figure 4 

Obedience & efficiency.

In contrast to Tyler’s claims, our sample did not produce any significant correlation between willingness to obey the laws and police’s legitimacy (adding the frequency of obedience as a covariate, and controlled for age and gender). Despite of the fact that youth’s trust was not influenced by the levels of effectiveness and equality of the police, their perceptions on the legitimacy of laws and regulations was associated with their (lack of) confidence in the police (p = .010). It was found that the adolescents who do not show any trust in the police were 4.3 times more likely to believe that laws and orders could be disobeyed than those who believed that they should always be followed.

Overall, the perceptions of juveniles on law conformity and their willingness to obey the laws were significantly correlated. The likelihood to evaluate the police as “neither inefficient nor efficient” was 2.8 times higher for those who reckoned that laws should never be followed (p = .044) than the respondents who, at least theoretically, always respect the laws. Age also appeared to be an influential factor in terms of laws’ legitimacy. The multiple correspondence analyses showed that the younger participants were more likely to have a negative reaction on whether laws and orders should always be obeyed than the older ones (Table 1). The significance of the correlation was becoming weaker as the age of the sample was rising (from p = .004 to p = .007 and p = .035). The MCA (Figure 5) illustrated the spatial relationship between the significantly related variables. It is worth noticing that the opinions of our participants on conformity in laws was changing from positive to negative as their age was increasing from 13 to 16 years old. The actual conformity in laws was following almost the same path where their tendency to obey fluctuated from “always” (13 and 14 years old) to “never” (15 years old) and subsequently to “sometimes” at the age of 16. It was also observed a ‘gap’ between the obedience in theory and the obedience in practice as the adolescents who did not believe that laws and orders should always be followed were actually obeying on an irregular basis rather than never.

Table 1

Logistic regressions explaining the influence of selected variables on perceptions of law, willingness to obey and actual obedience*.

95% Confidence Interval for Exp(B)
Obedience to ordersa b B Sig. Exp(B) Lower Bound Upper Bound

No Intercept –2.754 .000
Equality_police ( is ‘Always’)
Never –.057 .879 .945 .455 1.962
Almost never .213 .531 1.237 .635 2.410
Sometimes .166 .620 1.180 .613 2.274
Almost always .057 .891 1.058 .468 2.392
Evaluation_police (ref. cat. is ‘Efficient’)
Inefficient .823 .153 2.276 .737 7.034
Almost inefficient .568 .291 1.764 .615 5.059
Neither efficient nor inefficient 1.050 .044 2.856 1.028 7.939
Almost efficient .674 .195 1.963 .708 5.443
Trust_police (ref. cat. is ‘Yes’)
No 1.467 .010 4.338 1.422 13.229
More or less .979 .075 2.662 .906 7.820
Age ( is ‘16’)
13 –.911 .004 .402 .217 .745
14 –.685 .007 .504 .307 .828
15 –.523 .035 .593 .364 .965
Gender ( is ‘Female’)
Male .298 .129 1.347 .917 1.979
95% Confidence Interval for Exp(B)
Frequency of obedience to lawsc d B Sig. Exp(B) Lower Bound Upper Bound

Never Intercept –3.304 .002
Equality_police ( is ‘Always’)
Never .811 .213 2.249 .628 8.059
Almost never .307 .632 1.359 .388 4.761
Sometimes .260 .690 1.297 .361 4.656
Almost always .081 .922 1.085 .211 5.565
Evaluation_police (ref. cat. is ‘Efficient’)
Inefficient 1.288 .112 3.625 .741 17.741
Almost inefficient .086 .915 1.090 .223 5.317
Neither efficient nor inefficient .022 .977 1.023 .216 4.838
Almost efficient –.201 .799 .818 .174 3.837
Trust_police (ref. cat. is ‘Yes’)
No .659 .419 1.933 .390 9.574
More or less .211 .791 1.234 .261 5.838
Age ( is ‘16’)
13 .288 .569 1.334 .495 3.596
14 .113 .811 1.120 .444 2.822
15 .281 .535 1.324 .545 3.213
Gender ( is ‘Female’)
Male .106 .760 1.111 .565 2.186
Sometimes Intercept –1.331 .010
Equality_police ( is ‘Always’)
Never .393 .236 1.481 .774 2.836
Almost never –.011 .970 .989 .545 1.794
Sometimes –.073 .807 .930 .519 1.665
Almost always .126 .726 1.135 .561 2.296
Evaluation_police (ref. cat. is ‘Efficient’)
Inefficient .645 .179 1.906 .744 4.878
Almost inefficient .773 .064 2.167 .956 4.912
Neither efficient nor inefficient .619 .129 1.856 .836 4.122
Almost efficient .343 .391 1.409 .643 3.084
Trust_police (ref. cat. is ‘Yes’)
No .191 .637 1.211 .547 2.679
More or less .204 .587 1.226 .588 2.557
Age ( is ‘16’)
13 –.303 .271 .739 .431 1.266
14 –.273 .251 .761 .477 1.213
15 .103 .658 1.109 .702 1.753
Gender ( is ‘Female’)
Male .189 .294 1.208 .849 1.717

aThe reference category is: Yes

bNagelkerke R2 = .171

cThe reference category is: Always

dNagelkerke R2 = .032

*Results appear in odd-ratios.

Figure 5 

Joint-plot of category points.

Hypothesis 2: Equality plays an important role in the development of youth’s perceptions of the police performance performance

The police are the most visible agent of social control that is empowered to define right and wrong behaviour. If the police abuse their powers and wield their authority in unfair ways, this can not only damage people’s sense of obligation to obey their directives, but it can also harm public perceptions of morality that dictate lawful behaviour (Sampson & Bartusch, 1998). In other words, rules are no longer considered binding in the public life because “if they can do it (break the law), I can do it”. Generally, all available evidence suggests that fostering trust and legitimacy via fair and efficient treatment would be beneficial across the population. But, what happens in the opposite case? The reply comes through the very low satisfaction rates given by the youth for the function of the police. More specifically, only 7.3% (n=47) of the respondents provided a purely positive evaluation. Nonetheless, the remaining percentage was almost equally divided. Nearly 1/3 of the answers (n=220, 34.1%) described the police as inefficient or almost inefficient; 27.5% (n=177) marked the average point of “neither efficient nor inefficient”, and the last 191 adolescents (29.7%) defined the police function as almost efficient (Figure 6).

Figure 6 

Evaluation of police’s functions.

To call an institutional arrangement legitimate implies that there are people who sincerely believe that are treated equally by these arrangements (Matravers & Meyer, 2011). In our sample, the number of negative replies again prevailed over the positive ones. Only 86 (13.4%) of the adolescents considered that the police treat everyone equally. The highest number of replies (189, 29.3%) was assigned to the neutral option (Sometimes) followed by those who believe that people never (n = 122, 18.9%) or almost never (n = 168, 26.1%) receive an equal treatment by the police (Figure 7).

Figure 7 

Frequency of equal treatment by the police.

As foreseen, the correlation between the variables of equality and efficiency was overall found significant (p = .000). The strongest correlations, though, appeared among the negative responses. More specifically, the participants who saw the police as inefficient or almost inefficient were on average 25.1 times more likely to also perceive it as completely unfair or unequal (p = .000 for both), and 7 times more likely to describe it as almost unfair or unequal (p = .005 and p = .000) than those who perceived it as efficient. Likewise, the group of participants who replied that the police almost never treat people equally was on average 7 times less likely to consider it as efficient or almost efficient than those who had a positive impression for the police. Furthermore, the adolescents who had an average to good opinion about police’s treatment in term of equality were found to be 5.9 times more likely to have an average to good impression of it in terms of efficiency (p = .001 for both). Interesting enough, any past personal experience the adolescents had with the police was not significantly correlated with the evaluation they gave to it (the highest significance was p = .053). Yet, the correlation between the two dependent variables was found significant (p = .004). The adolescents’ views about the police in terms of equal treatment were generally influenced by their personal experience they had with it. Once again, although the gender of the respondents did not seem to influence their perceptions on the police, it appeared significantly correlated with their experience with the police (p = .018). In terms of age, the regression model showed a significant relation with (in)efficiency but mainly limited to the juveniles of 13 years old who had an average to bad opinion about the police (Table 2). The one-way analysis of variance (ANOVA) extended the above argument (p = .044 for the ages of 13 and 16), pointing that the opinion of our sample gradually changes from positive to (more) negative over the years. The same result was illustrated in the multiple correspondence analysis (Figure 8) where the average responses were mainly given by the participants of 14 and 15 years old.

Table 2

Logistic regressions explaining the influence of selected variables on the evaluation of police*.

95% Confidence Interval for Exp(B)
Evaluation of police functiona b B Sig. Exp(B) Lower Bound Upper Bound

Inefficient Intercept 1.295 .131
Equality_police ( is ‘Always’)
Never 3.108 .000 22.385 4.772 105.009
Almost never 1.589 .011 4.900 1.432 16.758
Sometimes .549 .466 1.732 .395 7.589
Almost always –1.028 .264 .358 .059 2.176
Experience_police ( is ‘Yes’)
No –1.178 .053 .308 .093 1.017
Age ( is ‘16’)
13 –1.713 .005 .180 .055 .594
14 –1.242 .048 .289 .084 .989
15 –1.199 .031 .302 .102 .895
Gender ( is ‘Female’)
Male –.210 .605 .810 .365 1.798
Almost inefficient Intercept .840 .292
Equality_police ( is ‘Always’)
Never 3.037 .000 20.834 4.929 88.070
Almost never 1.994 .000 7.346 2.522 21.394
Sometimes 1.859 .002 6.418 2.007 20.519
Almost always –.218 .735 .804 .228 2.838
Experience_police ( is ‘Yes’)
No –.784 .177 .457 .146 1.424
Age ( is ‘16’)
13 –.1.222 .025 .295 .101 .860
14 –.198 .715 .820 .283 2.374
15 –.844 .099 .430 .158 1.173
Gender ( is ‘Female’)
Male .141 .695 1.152 .569 2.330
Neither inefficient nor efficient Intercept 1.783 .017
Equality_police ( is ‘Always’)
Never 1.563 .028 4.773 1.187 19.194
Almost never .907 .066 2.477 .940 6.522
Sometimes 1.761 .001 5.821 2.076 16.324
Almost always –.341 .519 .711 .252 2.004
Experience_police ( is ‘Yes’)
No –.876 .127 .416 .135 1.281
Age ( is ‘16’)
13 –1.263 .019 .283 .098 .811
14 .086 .871 1.090 .388 3.058
15 –.725 .146 .484 .182 1.287
Gender ( is ‘Female’)
Male –.041 .907 .960 .487 1.894
Almost efficient Intercept 1.951 .008
Equality_police ( is ‘Always’)
Never 1.072 .132 2.921 .723 11.793
Almost never .467 .343 1.596 .607 4.196
Sometimes 1.762 .001 5.826 2.120 16.015
Almost always .392 .418 1.480 .574 3.816
Experience_police ( is ‘Yes’)
No –1.061 .061 .346 .114 1.049
Age ( is ‘16’)
13 –.635 .219 .530 .193 1.458
14 –.015 .977 .985 .353 2.751
15 –.714 .149 .490 .186 1.291
Gender ( is ‘Female’)
Male .046 .893 1.047 .537 2.042

aThe reference category is: Efficient

bNagelkerke R2 = .276

*Results appear in odd-ratios.

Figure 8 

Bi-plot of multiple category points.

Hypothesis 3: Young people trust the police based on their perceptions about how they are treated, and whether they operate in an effective way

Normative evaluations play a very important role in shaping people’s trust (or distrust) in institutions, and to a great extent they pertain to the fairness of procedures applied by the institutions (Tyler, 2001, 2006a, 2006b). People trust the police based on perceptions about how it treats them and whether it makes decisions in a fair way. In some cases, this might be even more important than institutional outcomes. People are often more inclined to accept decisions with negative consequences to them when they are convinced that this is the right or the fair thing to do (Frey & Osterloch, 2005). Contemporary studies – considering whether individuals’ perceptions of efficacy serve as a social-psychological cognitive orientation that influences levels of trust in the police – found that procedural justice evaluations are a primary source of trust in the police (Nix et al. 2014). Yet, is this argument verified in our research? Is the procedure more important than the outcome when it comes about motivating people? In the present study, yes.

Seeking to understand the rationale behind the predominantly negative image the young adolescents had about police and its functions, an open-ended question became the ideal tool to gather their personal motivations without influencing them with a predetermined list of possible answers. Once their replies were collected and separated into groups, it was deemed that any lack of trust towards the police is largely caused by high levels of corruption and disobedience to the law (n = 188, 29.2%). A common expression among the given answers was that “they are worse than thieves/criminals” or that “they are involved in drug dealing”. The second and third most popular responses were related to abuse of power (n = 172, 26.7%) and excessive violence (n = 88, 13.7%) followed by inefficiency and unavailability (n = 80, 12.4%). In these groups of answers, the youth often used improper language to express their feelings against the acclaimed “excessive shootings and killings of innocents” by police officers. These feelings were reflected in numbers when for the past year (2014) one out of five homicides in Sao Paolo were committed by the police (SSP, 2015).21 Only 5.1% (n = 33) of respondents believed that police is efficient, or able to provide security and help; surprisingly, a percentage lower than that given for the measurement of confidence. This could be partially explained by the few (n = 6) adolescents who claimed that they show trust to the police because they have no other choice, or out of necessity (Figure 9).

Figure 9 

Explaining the level of confidence in the police.

In terms of procedural justice, the participants who considered the police as unfair that always treat people unequally, were almost 6 times more likely not to show any confidence in the police than those who were trusting them (p = .006). Marginally insignificant was also the relationship between confidence and equality (p = .050), where their belief that the police almost never treat people equally was not associated with the level of trust in them; but they would still refer to them (p = .024). Similarly, the number of responds who described the police as “occasionally” or “almost always” fair had no influence on the adolescents’ ability to trust or not the police (p = .306 and p = .420, respectively). In terms of distributive justice, the correlation between confidence and efficiency was found to be stronger than that of procedural justice, and for almost all reply-options. More specifically, the respondents who had an average opinion on the police’s performance were 10.8 times less likely to trust the police than the ones who were showing confidence to it (p = .000). This likelihood was increasing as the ranking of the police’s performance was turning from “almost inefficient” (14.9 times) into “inefficient” (16.3 times). The evaluation of the police’s function became more positive once the participants responded that they somehow trust them (Table 3). The juveniles who perceived the police as “almost efficient” were 4 times more likely to partially trust them – but not necessarily refer to them (p = .809) – than those who fully do so (p = .002). The strength of all the above correlations is shown in the joint-plot below (Figure 10). As expected, the closest relationships were the “Yes-Efficient” and “No-Never”. The respondents who evaluated the police’s function as efficient were those who were fully trusting them; and also, the ones who had no confidence were the ones who deemed that the police never treats people equally. The juveniles who were “more or less” showing trust in the police had also given an average mark for its function and treatment of others describing them as neither “inefficient nor efficient” (to “almost efficient”) who are “sometimes” (to “almost never”) just and fair. Finally, for both independent variables (efficiency and equality), age and gender appeared to have no effect on their relationship with the dependent variable (confidence).

Table 3

Logistic regressions explaining the influence of selected variables on confidence in the police*.

95% Confidence Interval for Exp(B)
Confidence in policea b B Sig. Exp(B) Lower Bound Upper Bound

No Intercept –.745 .331
Referral_police –.604 .024 .547 .324 .924
Equality_police ( is ‘Always’)
Never 1.786 .006 5.968 1.683 21.158
Almost never 1.073 .050 2.923 1.002 8.527
Sometimes .532 .306 1.703 .615 4.718
Almost always .502 .420 1.652 .487 5.606
Evaluation_police (ref. cat. is ‘Efficient’)
Inefficient 2.795 .000 16.362 3.608 74.200
Almost inefficient 2.705 .000 14.957 3.752 59.615
Neither efficient nor inefficient 2.380 .000 10.808 2.987 39.105
Almost efficient 1.457 .014 4.295 1.341 13.760
Age ( is ‘16’)
13 –.450 .380 .638 .234 1.742
14 .337 .472 1.401 .559 3.511
15 .588 .238 1.801 .677 4.788
Gender ( is ‘Female’)
Male .167 .634 1.182 .594 2.354
More or less Intercept –.271 .681
Referral_police –.062 .809 .940 .569 1.553
Equality_police ( is ‘Always’)
Never .781 .201 2.184 .660 7.222
Almost never .853 .082 2.346 .896 6.139
Sometimes .683 .127 1.980 .823 4.761
Almost always .630 .231 1.878 .669 5.266
Evaluation_police (ref. cat. is ‘Efficient’)
Inefficient .896 .205 2.450 .613 9.787
Almost inefficient 2.017 .001 7.517 2.307 24.489
Neither efficient nor inefficient 2.096 .000 8.131 2.844 23.247
Almost efficient 1.401 .002 4.058 1.691 9.736
Age ( is ‘16’)
13 –.025 .956 .975 .402 2.369
14 .183 .675 1.201 .511 2.821
15 .662 .157 1.938 .775 4.848
Gender ( is ‘Female’)
Male .455 .160 1.575 .836 2.969

aThe reference category is: Yes

bNagelkerke R2 = .000

*Results appear in odd-ratios.

Figure 10 

Joint-plot of category points.

Challenges and Limitations

No empirical research is free of limitations. Following this dictum, such limitations should be acknowledged before trying to conceptualize our hypotheses. Walzer (1983) has argued that moral concepts cannot be applied universally because first, different social spheres and contexts imply different norms and values inside even of a given society and second, moral cultures are different across societies. Maybe, this is true. It still may be that beyond the differences, some commonalities also exist among different regions in Sao Paolo or even among other Brazilian cities in the way they interpret the moral background of a legitimacy claim. However, these elements are not sufficient to ignore the issue of representativeness. Bearing in mind the exceptional high levels of fatal violence in the particular district, as well as the cultural and ethnical diversity of a population within a vast geographical area like Brazil, it would have been scientifically imprecise to claim that the results are generalizable and that the present sample represents an area larger than the Jardim Ângela itself.

Ultimately, the percentage of students who were willing and able (because they provided the authorization form) to participate in the research does not depict the actual response rate due to a high number of absentees. Numerous students were absent for no reason during a series of visits at the schools. The actual reasons were frequently unknown by the school authorities as the parents were not aware of the truancy, or, unfortunately, they did not have such a close relationship with them as to be aware of it. Hence, the particular response rate refers to the number of matriculated students and not to the actual number of those who were attending the classes on a daily basis; the number of whom was practically impossible to calculate with accuracy.

Overarching Conclusions

This article has focused on a topic, which is scarcely addressed in the criminological literature and more specifically, in the study of police legitimacy reflected on obedience and trust. To a certain extent, it was based on the clear tendency of Brazil (following the example of many other developing and industrialized countries) to progressively toughen up its criminal justice system as a way to reduce the alarming rates of juvenile delinquency and ensure normative compliance (Lappi-Seppälä, 2011). Such policies may have an impact on public confidence, as the probability of a negative relation between legitimacy and trust becomes high (Murray et al. 2013).

In the first part of the present empirical research, the hypothesised relationships between empirical legitimacy, compliance and confidence emerged as strongly significant. In line with recent studies conducted by Hough et al. (2013) using the European Social Survey, and Zsolt (2012) working on the European “Fiducia Project”, the findings presented here show clear relationships between dimensions of trust and obedience in the police, and dimensions of perceived police legitimacy. Indeed, the juveniles who had no (or partial) trust in the police appeared considerably more likely to refrain from referring to it than the ones who had responded positively. A relatively high probability was given also for the juveniles who regarded the police as occasionally untrustworthy due to corrupted, unfair and racist behaviour, or because of their inefficiency. The participants who were seeing the police as inefficient or almost inefficient were more likely to also perceive it as almost to completely unfair or unequal than those who described it as efficient. Likewise, the participants who replied that the police almost never treat people equally were less likely to consider it as efficient or almost efficient than those who had a positive impression for the police. In addition, the adolescents who had an average to good opinion about police’s treatment in terms of equality were found to be more likely to have an average to good impression of it in terms of efficiency.

At the same time, our research (like that of Gouveia-Pereira, 2008) revealed a substantial link between legitimacy and rule-breaking behavior. Potentially this means that we should consider under what conditions the members of society are obliged to obey the state’s laws. The adolescents who did not show any trust in the police were more prone to believe that laws and orders could be disobeyed than those who believed that they should always be followed. The opinion on conformity in laws – as well as the frequency of their actual obedience – was changing from positive to negative (or from “always” to “never”) as the age was increasing from 13 to 16 years old. The same tendency was observed for the evaluation of the police in terms of efficiency where the answers were gradually changing from “efficient” to “inefficient”. Of particular importance was the strong relationship between trust, fairness and dimensions of perceived legitimacy. Researching trust may have practical relevance because confidence in the police is part of its legitimacy; and legitimacy is not necessary for its own sake but because it is a condition of effective policing. Therefore, the participants who considered the police as unfair that always treat people unequally were the least likely not to show any confidence in the police than those who trusted them. As far as distributive justice is concerned, the respondents who had an average opinion on the police’s performance were more likely to distrust the police than the ones who were showing confidence in them. The evaluation of the police’s function became more positive once the participants responded that they somehow trust them. Finally, the juveniles who perceived the police as “almost efficient” were more likely to partially trust them – but not necessarily refer to them– than those who were fully doing so.

The above result brought us to the conclusions that citizens who are willing to obey (and they feel an obligation to do so) can be motivated in ways besides the fear and use of punishment or violence. The assumption that threads of punishment backed up with coercion typically create compliance is fundamentally inconsistent with trust, creating a gap – rather than a bond – between the state and its citizens (Meares, 2009, p. 660; McCluskey, 2003, p. 108). On this, we found that the higher level of coercive power displayed, the less likely targets are to comply, as felt duty to obey seems to be rooted in positive connections with authorities. Hence, in order to mark social connections as positive, citizens should fully agree with legal restrictions; otherwise, further interventions will eventually be required. Not surprisingly, willing conformity is able to lead to long-term acceptance, rather than short-term compliance (Bottoms & Tankebe, 2013). This means that coercion serves little to enhance compliance although it may be a central component of the role of the police.