Doyle, Caroline --- "Perceptions and Realities of Violence in Medellin, Colombia" [2019] IntJlCrimJustSocDem 22; (2019) 8(2) International Journal for Crime, Justice and Social Democracy 149

	Home \| Databases \| WorldLII \| Search \| Feedback International Journal for Crime, Justice and Social Democracy

Abstract

Latin America is one of the most violent regions in the world, and this is particularly evident in its many cities. While urban violence scholars and policymakers generally rely on homicide rates to measure levels of violence in urban environments, these objective indicators often do not capture its realities. By drawing from over six months of fieldwork in the Latin American city of Medellín, Colombia, this paper shows how Medellín has experienced a significant reduction in homicides, but both real and perceived violence continues to have a significant effect on residents’ lives. The article contributes to the urban violence debate by highlighting its complexity in Latin America and how it is not fully quantifiable.

Keywords

Colombia; Homicides; Medellín; qualitative data; urban Violence.

Doyle C (2019) Perceptions and realities of violence in Medellín, Colombia. International Journal for Crime, Justice and Social Democracy 8(2): 149-166. DOI: 10.5204/ijcjsd.v8i2.1010.

Introduction

According to the United Nations (UN) Office of Drugs and Crime (2013), in 2013 Latin America was the most violent region in the world. In 2015, it constituted 9% of the world’s population but 33% of global homicides, with a murder occurring approximately every 15 minutes (Marinho and Tinoco 2017). As most of the violence occurs in the cities, it is unsurprising that in 2016 43 of the 50 most dangerous cities in the world were located in Latin America (Igarapé Institute 2016). Scholars and policymakers from a wide range of fields have questioned how best to conceptualise, measure, prevent and reduce incidents of violence in urban environments. There is a general consensus that homicide rates do provide the most convenient and useful indicators of levels of violence; however, relying exclusively on homicide rates alone to measure such incidences means scholars and policymakers fail to capture the realities of urban violence. These can include the effect violence has on the daily lives of residents, how residents respond to violence and community perceptions.

The purpose of this paper is not to try and solve the problem of quantifying urban violence, but rather to contribute to the debate by highlighting the complexities of its measuring in Latin American cities. This paper achieves this by drawing from over six months of fieldwork in the Latin American city of Medellín, Colombia. Over the past 30 years, this city has achieved a significant reduction in homicides, from 381 homicides per 100,000 residents in 1991 to 23 per 100,000 in 2017. However, as discovered during fieldwork, despite this decrease, violence—both real and perceived—continues to have a significant effect on the daily lives of many residents.

The structure of this article is as follows. First, a review of the methodological literature on measuring urban violence is presented, focusing on homicide rates and highlighting their strengths and limitations. The review also shows how complementary qualitative methodologies have been shown to address some of these limitations, such as gaining access to residents’ perceptions of violence. The article then presents qualitative data collected in Medellín between 2014 and 2015 to show how despite the significant reduction in homicides, violence (real and perceived) continues to characterise everyday life for many residents. Finally, the article synthesises the data presented to show the complexities of urban violence and the difficulties in quantifying this phenomenon. The article argues how urban violence reduction and prevention policymakers should draw from the combined views of objective indicators (such as homicide rates and insights from residents) through qualitative research methods to fully comprehend the complexities of this violence and, ultimately, to design and implement policies for its subsequent reduction.

A specific problem encountered in the urban violence debate concerns the difficulty of measuring levels of violence. There is general consensus among academics and officials that homicide, or the ‘intentional killing of a person’, provides the most reliable indicator of such incidence. This is largely because the severity of homicide makes it more likely to be reported than other violent crimes, such as assaults or robberies (Krause 2009; Muggah 2012; World Bank 2011). For example, in 2016 Brazilian Think Tank, the Igarapé Institute (2017) released a comprehensive measurement of lethal violence around the world and relied on homicide rates for its analysis, arguing that these are the ‘most comparable rates’. Increases and/or decreases in homicides can also indicate changes in levels of urban violence for policymakers and scholars when evaluating the success of urban violence reduction and prevention policies. Notably, a police pacification program launched in 2008 in Rio de Janeiro, Brazil has generally been viewed as a successful urban violence reduction policy, given the 65% reduction in homicides in ‘pacified areas’ some four years later (Muggah and Szabó De Carvalho 2014). Analysing levels of violence within a city using homicide rates can likewise assist policymakers in targeting violence reduction policies and resources in areas with high incidence. However, this micro-level analysis generally reveals ‘pockets of insecurity’ among the poorer and most marginalised neighbourhoods rather than in more middle-class communities (Briceño-León and Zubillaga 2002; Winton 2004). For example, in 2002 in Rio de Janeiro the homicide rate in the lower-income areas was 177 per 100,000 compared to 38 per 100,000 in its higher-income counterparts (Briceño-León and Zubillaga 2002).

While homicide rates can provide important indicators for policymakers and scholars in evaluating the success of urban violence reduction and prevention policies, there are limitations deriving from how these rates are collected and analysed. For example, in 2016 the Citizens’ Council for Public Security and Criminal Justice (Seguridad, Justica y Paz 2017), which tallies an annual index of the world’s most violent cities, classified Caracas in Venezuela as ‘the most violent city in the world’, with a homicide rate of 130 per 100,000, while the Igarapé Institute (2017) reserved that label for San Salvador (the capital of El Salvador), with a rate of 136 per 100,000. The reason for this difference was that the latter excluded Venezuela from its classification due to data discrepancies on homicide rates encountered in the country (Aguirre and Muggah 2017). Similar issues involved with the collection and analysis of homicide data have also been acknowledged by InSight Crime, an organisation dedicated to the study of organised crime in Latin America and the Caribbean. In its 2017 report ‘Homicides in Guatemala: The Challenge and Lessons of Disaggregating Gang-Related and Drug Trafficking-related Murders’ the way in which authorities gathered and analysed homicide data in Guatemala was particularly scrutinised. InSight Crime concluded that their overall analysis on gang-related and drug-trafficking related murders was limited because of the difficulties encountered when collecting data on homicides in Guatemala, including poor training of police and public officials, and technological challenges (Dudley 2017).

To circumvent the unreliability of homicide and crime statistics, researchers from a range of fields have used different qualitative methodologies to provide insights into how residents understand violence and its effect on their daily lives (Abello-Colak and Pearce 2015; Auyero, Burbano and Fernanda Berti 2013; Baird 2012; Berents and ten Have 2017; McCord 2003; Moser 2004; Moser and McIlwaine 2004; Orjuela 2010; Winton 2004). For example, victimisation surveys have been used in criminology since at least the 1990s (see Coleman and Moynihan 1996; Kesteren, Dijk and Mayhew 2013). Some international organisations also carry out such measures, through which respondents are asked if they or their family members have been victims of crime over the past 12 months. Surveys of this kind in Latin America have revealed how average victimisation rates fluctuate between 30% and 40%, which is considerably higher than Europe or North America, whose rates average 15% (Latinobarómetro 2011).

Other approaches undertaken by researchers to understand the realities of urban violence have included participatory urban appraisals. Notably, anthropologists Moser and McIlwaine (1999) undertook group discussions, semi-structured interviews, direct observations and local stories to examine the dynamics of violence in Colombian and Guatemalan communities. More recently, a report on the complexities of urban violence published by the World Bank in 2011 triangulated quantitative and qualitative data from cities in Kenya, South Africa, Haiti, Brazil and Timor-Leste. The findings revealed how in the everyday experiences of urban residents coping mechanisms to deal with violence occur mostly on an individual level, which may actually undermine long-term violence prevention generating perverse social capital (World Bank 2011). Not only is capturing residents’ experiences of daily violence important, but also policymakers need to understand how residents perceive levels of violence. Gaining access to these perceptions is important, as there is often marked difference between residents’ perception of crime and its actual incidence (Restrepo and Moreno 2007). Cultural criminologists have also criticised the exclusive reliance on quantitative studies to understand violence (Ferrell 2013), and in Latin America these discrepancies are especially noted. For example, a 2013 Development Bank of Latin America (Corporacion Andina de Fomento; 2014) study found that respondents estimated the percentage of households that were victims of crime to be 47%, whereas the recorded rate was considerably lower at 27%. Previous research has also shown how the discrepancy between perceptions of crime and recorded levels can be related to an individual’s exposure to crime by friendship groups, crime covered in the media, one’s trust in the police, the availability of statistical information, and the gender, age, class and ethnicity of a person (Hume 2007; Maris and Ortega 2014; Orjuela 2010; Reiner, Livingstone and Allen 2003; Russo, Roccato and Vieno 2013). For example, from qualitative interviews in Colombo, Sri Lanka, Orjuela (2010) found how perceptions of security can vary among different minority groups.

Further, the perception of security can have various effects on residents’ lives, ranging from access to employment to trust in institutions (Development Bank of Latin America 2014). According to the Development Bank of Latin America (2014), only 43% of Latin American urban residents reported crime to authorities, with many participants citing their distrust in the police to fulfil their duties as the main reason for their hesitance. Similarly, Corbacho, Philipp and Ruiz-Vega (2012) found that individuals in Latin America who were victims of crime have 10% less trust in the police than those who had not encountered such incidents.

Indeed, some Latin American countries do use other indicators to measure levels of violence. For example, in Mexico the National Survey of Victimization and Perception of Public Security (Encuesta Nacional de Victimización y Percepción sobre Seguridad Pública; ENVIPE) has become international best practice for measuring victimisation and crime (Asmann 2017)—the results of which provide important information for policymakers designing and implementing public policies in areas of victimisation and public safety (National Institute of Statistics and Geography 2017). The survey collects data from residents on areas such as their perception of safety and their trust in public institutions. For example, in 2016 61% of residents reported that insecurity and crime were the number one problems affecting them at a state level (National Institute of Statistics and Geography 2017). In the same year, the ENVIPE survey also revealed how 93% of crimes were not reported to authorities. When residents were questioned about why they did not alert authorities, 33% responded that it was a ‘waste of time’, while 16% stated that they did not have ‘faith’ in local authorities (National Institute of Statistics and Geography 2017). The insights from this survey are important for local and national policymakers, as they provide reliable evidence for comparing whether decreases or increases in actual crime levels affect perceptions of safety. Further, Braakmann (2012) and Avila et al. (2016) argue that any Mexican crime prevention policies, such as those that improve access to public spaces and combat organised crime activities, will be limited in their effectiveness if these high perceptions of security and distrust in institutions are not addressed.

While qualitative data collection methodologies have been shown to provide a more robust understanding of violence, researchers have acknowledged the challenges of conducting qualitative research in environments with high levels of violence. For example, resident groups may be sceptical of a researcher’s motives (Belousov et al. 2007; Nordstorm and Robben 1995). Locals may be conditioned to dissemble when outsiders start asking questions, as anyone exhibiting such behaviours are often tax collectors, police authorities or housing inspectors (Goldstein 2014). Structured or rigid semi-structured questionnaires and interviews may also appear threatening and inappropriate (Barakat and Ellis 1996). Thus, qualitative researchers in these environments need to be flexible and adopt a wide range of data collection methods to ensure the data they collect are both valid and reliable (Doyle and McCarthy-Jones 2017; Gill 2004; McGee and Pearce 2009; Shenaz Hossein 2016). Further, as Berents and ten Haven (2017) acknowledged, certain considerations must be taken into account when researchers from the Global North are conducting research in the Global South. Particularly, researchers should be aware of the potential cultural and uneven power relations that exist between participants and researchers (Escobar 1995; Santos 2014).

The key themes that emerged from the literature review point to the complexities of measuring violence in urban environments. While policymakers and scholars have relied on homicide rates to measure levels of urban violence, this objective indicator does not acknowledge other forms of violence and its effects on residents. This study contributes to this debate by providing a case study approach to highlight some of the recent complexities of violence in a Latin American urban environment.

Medellín is a compelling case study to demonstrate the complexities of measuring urban violence. Given the recent mass reduction in homicides the city has recently experienced, Medellín has become known as a benchmark for policymakers seeking to understand and implement successful urban violence reduction and prevention policies (Inter-American Development Bank 2009; Muggah 2014, 2015; UN Human Settlements Programme 2007; World Bank 2011, 2016). However, the policymakers, journalists and academics ‘singing the praises’ of the violence reduction in the city all used homicide rates as their primary indicator to state their claims. For example, the Inter-American Development Bank (2009) said:

Given the limitations of relying on homicide rates alone to measure levels of urban violence, fieldwork was conducted in 2014 and 2015 to gain further understanding of residents’ perceptions and experiences of insecurity and violence. A significant element of qualitative research concerns the accuracy of representing participants’ realities of a social phenomena and its credibility to them (Schwandt 1997). Hence, Creswell (2014) argues the importance of using certain strategies, such as prolonged engagement and persistent observation in the field, accessing multiple sources of data and conveying findings through ‘thick’ descriptions to enhance credibility.

The fieldwork conducted for this research was authorised by the Human Research Ethics Advisory Committee at the University of New South Wales, Canberra. During fieldwork in Medellín, semi-structured interviews were conducted with a range of local stakeholders. After obtaining the participants’ consent, interviews were recorded using a recording device, and the data obtained were made identifiable to the researcher only. These were later stored in a locked cabinet and on a password-protected computer, which was only accessible by the researcher. The interviews were next transcribed and translated by the researcher, who is from the Global North (Australia) but has over seven years’ experience living in Latin America, and is fluent in both Spanish and English. Given the nature of the research, all participants requested that their identities remained anonymous.

The inclusion criteria for participants was that they should be over the age of 18 and have knowledge about the recent, and historical, levels of violence in Medellín through their policy or research work experience. Participants were recruited using two non-probability techniques: purposive sampling and snowball sampling. For the purposive sampling, samples were drawn strategically from individuals who were, on paper, to have knowledge about violence in Medellín, while snowball sampling was used to complement the purposive sample by interviewing actors recommended by participants from the latter group (Valdez and Kaplan 2008). However, not all actors recommended for the study were interviewed, but all participants were assessed for suitability.

The participants interviewed included four former government officials, three government officials, three community leaders, two members from non-government organisations (NGOs), one academic, one official from a research centre and one business leader. The interview questions were open-ended and designed to ensure the interviews felt like a conversation and allowed a number of topically relevant questions about the patterns of the violence to be discussed (Liamputtong and Ezzy 2005). Analysing data from the interviews followed the ‘thematic analysis model’, adhering first to the researcher’s familiarisation with the data, generating the initial codes, searching for the codes, reviewing the themes, defining and naming the themes, and finalising the analysis (Braun and Clarke 2006). In this article, the insights from two NGO workers, a community leader, an official from a research centre, a journalist and a government worker are provided. These individuals have all spent numerous years either researching violence in Medellín or designing policies to address the violence in the city.

During fieldwork the researcher adopted different ethnographic methods to gain insight into the everyday realities of local residents, and different opportunities of participant observation were subsequently engaged. For example, this meant attending local conferences, courses and public events including a five-day course called ‘Social Urbanism in Medellín: Urbanism, Environment and Society’ (‘Urbanismo Social en Medellín: Urbanismo, Medio Ambiente y Sociedad’) at a local university, as well as a two-day conference organised by the Colombian National Police, which focused on exchanging experiences and approaches on addressing levels on crime throughout Latin America. The researcher also shared an apartment with a local from Medellín, which was important for not only gaining knowledge about day-to-day life in the city but also for understanding its local security measures. A field diary was simultaneously kept with detailed observations from these opportunities.

Primary, secondary and tertiary documents were analysed during fieldwork to build and deepen understandings of the levels of violence, including how residents perceive such incidents, in turn (Burnham et al. 2004). For example, this saw the researcher obtain access to local newspaper archives and media sources reporting on current levels of violence in the city, such as ‘How is Medellín Going’ (Medellín Como Vamos) and ‘Citizens Report’ (Denuncias Ciudadanas).

Challenging and dangerous situations were encountered during fieldwork in Medellín. Notably, in 2015 a local journalist was threatened by local criminal groups, and three human rights workers were assassinated in less than a month (including a potential participant), after being shot some 25 times in a poor neighbourhood (Garcia 2015; Laverde Palma 2015). These events meant the researcher needed to be creative, flexible and adopt a wide range of data collection methods to achieve the objectives of the research and ensure that the data presented are valid and reliable (Doyle and McCarthy-Jones 2017).

The following data show how despite a marked reduction in homicides in Medellín, violence (both real and perceived) continues to affect the daily lives of its residents. These findings show the complexities of urban violence and how it is not fully quantifiable.

In Medellín, the perpetrators of violence (homicides) have generally been classified by government officials and the media into three groups: criminal gangs (bandas), left-wing militias and right-wing paramilitaries. Over the past 30 years, these groups have all used violence (homicides) to maintain and/or obtain territorial control, and often to engage in crime, such as narco-trafficking (Doyle 2018). While it is beyond the scope of this article to provide a detailed analysis of these perpetrators, a brief explanation has been included to highlight their complexities and their organised nature.

The first group that used homicidal violence to maintain and/or obtain territorial control was the criminal bandas. These organised perpetrators of violence emerged around the 1980s and worked for the Medellín cartel in trafficking narcotics from cultivation to retail sale. By 1990, it was estimated that up to 150 different bandas were operating in the city (Bedoya Marin and Jaramillo Martinez 1991; Salazar and Jaramillo 1992). There were numerous conflicts between the different bandas, including between them and the state, which resulted in high levels of violence. For example, in June 1990 over 150 young males were assassinated in 20 different massacres, and in 1992 over 620 police officers were killed by banda members working for the Medellín cartel (Human Rights Watch 1994). Following the death of Pablo Escobar, who led the Medellín cartel, most banda members began working in the ‘next generation’ of the Medellín cartel, the Envigado Office (Oficina de Envigado). Led by an individual known as Don Berna, this criminal organisation connected customers to the services offered by the Oficina de Envigado, such as assassinations or kidnappings (Gutiérrez Sanín and Jaramillo 2004; Hylton 2007). From the late 1990s, there were numerous conflicts between this organisation and the left-wing militia groups in the city that had emerged around the early 1980s following national negotiations with the left-wing guerrilla group, the 19th of April Movement (Movimiento Abril de 19 or M-19), who brought ‘peace camps’ to Medellín (Gutiérrez Sanín and Jaramillo 2004; Lamb 2010). These camps were eventually made illegal; however, ‘outlaw’ militia groups remained and they used violence against individuals who threatened the security of their host communities, such as those working in the bandas (Gutiérrez Sanín and Jaramillo 2004). Another organised group that engaged in violence was the United Self-Defense Forces of Colombia (Autodefensas Unidad de Colombia; AUC), which had the national objective of countering the left-wing guerrillas. The AUC entered Medellín in the late 1990s and engaged in violence against the left-wing militia groups and any opposing bandas to obtain control of their territories (Doyle 2018). There were two AUC ‘groups’ (or ‘blocks’) in Medellín: the Metro Block (Bloque Metro; BM) and the Block Cacique Nutibara (Bloque Cacique Nutibara; BCN), with the former led by Don Berna, who also controlled the Oficina de Envigado (Doyle 2018).

By the early 2000s, there were numerous conflicts between these different organised perpetrators of violence. There was conflict between the militia, the AUC and state authorities, and the BCN emerged as the victor by late 2002 (Doyle 2018; Rozema 2008). Another conflict arose between the two AUC blocks, the BCN and the BM, due to their differences regarding how they should obtain profits for the organisation. For example, the leader of the BM Doble Cero preferred to make money from renting out private security services and collecting extortion payments, while the BCN largely relied on profits from protecting cocaine production facilities and narco-trafficking (Amnesty International 2005; Doyle 2018; Lamb 2010). Eventually, the BM was defeated around mid-2003 and Don Berna obtained a monopoly control over the criminal underworld of Medellín; this continued until 2008 when he was extradited to the United States on drug trafficking charges. Given that there was no successor to assume control over the Oficina de Envigado, conflict soon followed for this prospective leadership role. Another group (loosely linked to the AUC) known as the Urabeños (Los Urabeños) also used the extradition of Don Berna as an opportunity to expand its drug trafficking network and attempted to take over the Oficina de Envigado (McDermott 2014). Indeed, violence soon followed, including the massacre of nine Oficina de Envigado members in December 2012, with rival factions being the primary suspects (McDermott 2013). From 2008 to 2009, homicide rates consequently increased over 60% (from 49 homicides per 100,000 residents to 99 per 100,000) largely because of the tension within the Oficina de Envigado and between the Oficina de Envigado and Los Urabeños (Alcaldía de Medellín 2017; Doyle 2018).

While the purpose of this article is not to explain the recent reduction in homicides—from 99 per 100,000 in 2009 to 23 per 100,000 in 2017—some international organisations and researchers have cited local government policies such as the social urbanism schema (which addressed the city’s socio-economic inequalities) to explain this marked reduction. For example, Muggah (2015), Patiñ o et al. (2014) and Cerdá et al. (2012) argue that because such policies appeared to borrow from the most influential urban violence reduction and prevention approaches, social disorganisation and broken windows, their subsequent implementation can largely explain the recent drop in homicides. However, others have claimed that a more plausible explanation for this reduction stems from a pact, known as the gun pact (pacto de fusil), between Los Urabeños and the Oficina de Envigado in 2013, in which both criminal actors agreed to lower homicides to avoid the detection of authorities (Doyle 2018). During fieldwork, participants did acknowledge the importance of this pact in explaining the recent reduction in homicides. Notably, an NGO worker who has detailed knowledge of the criminal actors in the city spoke of the agreement and how it provides a more plausible explanation in this situation:

A community leader who also held knowledge of the city’s criminal actors too provided a detailed description of the pact:

Despite the reduction in homicides, recent data has shown how up to 60% of homicides continue to be related to territorial disputes between bandas that work for the Oficina de Envigado and those serving Los Urabeños (Duncan and Peciado 2018). In 2018, the pact appears to still be in place, with recent acts of violence having been attributed to these same criminal actors. For example, in February 2016 over a period of 48 hours three dismembered bodies were found in different parts of the city (Analisis Urbano 2016). In January 2018, a suitcase containing human remains was found in a public park in an elite Medellín neighbourhood; in the same month authorities found the bodies of five tortured males wrapped in sheets and dumped in different parts of the city, with authorities indicating that criminal actors were behind the assassinations (Agudelo 2018; Zambrano Benavides 2018).

Such acts of real and perceived violence certainly affect the lives of residents. When the researcher collected data in the city in 2014 and 2015, it was experiencing ‘the lowest homicide rates in 35 years’ (Alcaldía de Medellín 2017). Despite this, daily acts of violence were noted, as evidenced in the following field diary notes:

Further, while homicide rates have been reducing since 2009, residents do not feel safer. As reported by the Medellín-based organisation Medellín Como Vamos (2015), the perception of citizens who feel safe in the city actually decreased from 55% in 2012 to 44% in 2015. The daily reality for many residents was best summarised by one NGO official:

One of the control mechanisms used by the bandas that continue to operate in the city—and one of the reasons why residents continue to feel unsafe—is because they demand payments (through extortion, or vacuna) from many Medellín residents. There are estimates that one criminal group can earn up to $3.5 million each year from these payments (Semana 2016). Local media have also cited examples of criminal groups demanding payments for using the innovative forms of transport for which Medellín has become so famous, notably its outdoor escalators in one such neighbourhood (Gualdron 2012). Meanwhile, in 2017 Medellín Como Vamos (2017) reported 437 cases of extortion, a figure that was soon disputed in a victimisation survey by the National Administrative Department of Statistics, which found that around 2.9% of the Medellín population (or 59,000 individuals) were actually victims of extortion. The difference between these sources demonstrates just one example of the importance of victimisation surveys when measuring levels of crime.

Any resistance of payment from residents can lead to threats of violence or even assassination. For example, in October 2015 a local media outlet reported on the assassination of a bus driver just hours after he had participated in a protest to raise awareness of local bandas demanding extortion payments from fellow drivers (El Colombiano 2015). Previously, a manager of a local bus company had reported such drivers were forced to pay up to USD$40 a week to criminal actors. The manager explained how he had been sent threatening photos from a local banda, including one of a bus driver who had been shot multiple times, with the message ‘pay up or look what will happen to you’ (Hidalgo 2013). Similarly, during fieldwork in May 2015 a community leader known for making complaints against the criminal bandas for extortion payments was brutally assassinated (Garcia 2015). This murder, along with two other community leaders within a month, and the assassination of 16 bus drivers from 2009 to 2013 who refused to pay extortion payments, demonstrates the control that the perpetrators of violence have against anyone who speaks out against them (Garcia 2015; Hidalgo 2013).

Another form of control used by the criminal bandas is forced urban displacement. This practice involves threatening residents to leave their territory or face severe consequences, such as death. Bandas use this type of violence, as it provides access to the profits (such as from extortion payments) in particular territories (Medellín Como Vamos 2015). Forced urban displacement has increased considerably since 2007, with 5,376 victims reported in 2015 (Medellín Como Vamos 2015). Victims of this type of crime did decrease to 1,206 in 2016, but this figure soon increased to over 3,000 in 2017 (Medellín Como Vamos 2017). Given the distrust residents have in public institutions and the fear of possible retaliation from criminal actors, it is assumed that this number is, in fact, higher and that there are many more victims than previously thought (Abello-Colak 2015). This type of violence has been acknowledged by the Medellín Government as a more effective mechanism of obtaining or maintaining control of territories by the bandas working for Los Urabeños or the Oficina de Envigado, as it does not leave the same physical evidence as homicides (Alcaldía de Medellín 2011). For example, one researcher with knowledge of criminal actors in the city interviewed in 2014 commented on how increases in victims of forced urban displacement can generally be attributed to an increase in gang members within a particular neighbourhood:

It is not just extortion and forced urban displacement that affect the daily lives of Medellín residents; so too does their perception of the city’s levels of violence. For example, a 2015 study by the Medellín-based Centre of Political Analysis (Centro de Analisis Politico; CAP) found that residents feel safer in their own residential neighbourhood compared to other parts of the city (Casas-Casas and Giraldo Ramírez 2015). In 2015, the comuna (similar to a shire or commune) with the highest homicide rate (105 homicides per 100,000 residents) in Medellín was the economic centre of the city La Candelaria. When residents from this comuna were questioned about their perception of safety, 45% reported feeling safe in their neighbourhood, 37% did so in their comuna and 18% in the urban boundaries of Medellín. The comuna with the lowest homicide rate (five per 100,000), El Poblado, which is also considered the highest socio-economic area in Medellín, saw 54% of its residents claim to feel safe in their neighbourhood, in addition to 55% in their comuna and 19% in the urban boundaries of Medellín. When asked to respond to why they feel unsafe, the most common response from residents across all comunas in Medellín was their perception or knowledge of the presence of bandas, the low numbers of police, news reports on levels of crime and violence, and comments made to them from their social groups (Casas-Casas and Giraldo Ramírez 2015).

The findings of the 2015 CAP study were evident during fieldwork. Part of this fieldwork involved visiting the most iconic improvements in the poorer neighbourhoods, and these have largely been cited as being an important factor in the reduction of Medellín’s homicide rates (Cerdá et al. 2012; Jaitman and Guerrero Compeán 2015; Patiño et al. 2014). However, residents reportedly had no interest to visit these sites because they did not ‘want to look for problems’, given their perception of high crime rates, which they obtained from media reports in neighbourhoods different from their residential ones. Locals would also offer cautionary words on ‘being careful’ when visiting other neighbourhoods, but often they themselves had not visited such neighbourhoods in recent years, if ever (2015, field notes). In some cases, taxi drivers would refuse to travel to particular neighbourhoods, as they did not think it was safe to do so, given that work colleagues and friends had warned them about past incidents of crime. During an interview with a research official who had knowledge of the crime rates in the city, she commented on her fear of travelling from her residential area, El Poblado, to the economic centre of the city, La Candelaria:

The perception of violence also has an effect on the trust residents have in institutions, such as the police. For example, the CAP study found that residents in Medellín perceived police as ‘inefficient and corrupt’ (Casas-Casas and Giraldo Ramírez 2015: 71). During personal observations and interviews, residents also spoke at large of their distrust in government institutions, especially the police. Research participants would comment on their knowledge or perception of collusion between local public security forces and local bandas (2015, field notes), and this possible collaboration was best explained by one journalist:

One government official also explained how he had knowledge or an understanding that private security cameras, such as those used by the government and private businesses, were infiltrated by ‘illegal businesses’: ‘you can invest all you want ... You can put 1,500 cameras, but their structures are all infiltrated by illegal businesses. Has anyone ever pointed this out to you?’.

In addition, participants commented on how this knowledge or perception reduced the likelihood of reporting criminal actors to the authorities, as they were aware of the possible repercussions. One NGO official interviewed did provide a particular example of the consequences that residents could expect from crossing criminal actors:

The perception or knowledge of the relationships that residents have between local public security forces and local bandas has recently been confirmed with the arrest of the Medellín security secretary who was charged with having ties to local criminal groups and protecting ringleaders of criminal organisations for his own personal and political gain (Clavel 2017; Semana 2017). This is the not the first time a Medellín public official has been arrested with connections to organised crime in the city. In March 2017, four police officials were arrested and 11 were detained the month before for having similar connections (RCN Radio 2017).

Latin American urban environments have become known as the ‘most dangerous and violent in the world’ due to their high levels of homicide. Indeed, such rates do provide a useful and accessible indicator of community violence and have generally been accepted by scholars and urban violence reduction and prevention policymakers as the most reliable sources (Igarapé Institute 2017; Muggah 2012; World Bank 2011). They also allow policymakers and scholars to measure the success of urban violence reduction and prevention policy approaches (Briceño-León and Zubillaga 2002; Muggah and Szabó De Carvalho 2014). However, as has been acknowledged, these rates do not provide a complete picture of urban violence. This is not only because of the limitations concerning accuracy and consistency in the collection and analysis of homicide data, but also because such rates do not provide the nature and effect violence has on the daily lives of residents (Aguirre and Muggah 2017; Dudley 2017). Qualitative research on violence has revealed important insights as to the causes, dynamics and possible solutions related to violence perceived by residents (Moser and McIlwaine 1999; World Bank 2011), and these can assist in the effective design, implementation and evaluation of urban violence reduction and prevention policies.

This article has shown how despite the significant reduction of homicides in Medellín, violence (both real and perceived) continues to have an effect on the lives of residents. As the researcher discovered during interviews and through participant observation, residents continue to feel threatened by the presence of perpetrators of violence who have the capacity to engage in such acts, should they wish. Often, these criminals use multiple forms of force to send messages to residents that they are, in fact, in control of their daily lives.

Data were presented to show how the homicide rates in Medellín are just the ‘tip of the iceberg’. When analysing its ‘submerged’ parts, a complex situation is revealed in which residents are often afraid of reporting to local authorities the presence of criminal gangs, known as bandas or combos, in their neighbourhoods for fear of a potential retaliation. Therefore, this lack of reportage has an effect on official crime statistics, as they may only represent a small percentage of actual crime in the city. In Medellín, residents are also afraid of travelling to different parts of the city and lack a perception of safety in their own neighbourhoods. Many continue to feel unsafe because of their awareness or knowledge of gangs, low numbers of police, news reports on levels of violence and through crime and social network groups. Indeed, the insights provided in this article do complement previous studies on why residents may have different perceptions of the city’s levels of violence and insecurity (Development Bank of Latin America 2014; Maris and Ortega 2014; Moser and McIlwaine 1999; Orjuela 2010).

This article contributes to the urban violence debate by demonstrating how a city can experience a significant reduction in homicide rates, but violence (both real and perceived) can continue to affect the lives of residents. Thus, this article offers crucial insights for policymakers in the design and implementation of urban violence reduction and prevention policies, as it has provided a case study example of its complexities and the difficulties in quantifying this phenomenon. As such, Latin American urban violence reduction and prevention policymakers should ensure they are using a combination of reliable data, which is both quantitative and qualitative in nature, in not only attempting to measure urban violence but also in evaluating future policy approaches.

Latin America is the most violent region in the world. Currently, there is no single approach to reduce such incidents or prevent future outbreaks. However, the first step for scholars and policymakers is to ensure they are using evidence that provides a holistic view of the realities of urban violence. This article has showed how despite a massive reduction in homicides in Medellín, Colombia, the voices and experiences of its urban residents revealed that violence and the threat thereof continues to have a significant effect on their daily lives. Indeed, the Medellín experience also shows how homicide rates do provide a useful indicator of urban violence and remain an important tool for measuring the success of its subsequent reduction and prevention policies. Conversely, it is the voices and experiences of urban residents that can likewise reveal deep and rich insights into the realities of such incidence. By capturing these perspectives, some of the limitations encountered with the collection and analysis of homicide rates can be addressed, and urban violence scholars and policymakers are, in turn, provided multifaceted insights into the phenomenon’s very complexities.

Nonetheless, these insights can be better equipped to help design and implement policies that address the growing rates of violence in Latin America. For example, future research could analyse other cities that have experienced reductions in homicides (such as São Paulo, Brazil), and if (real and perceived) violence continues to affect their residents’ lives, in turn. It is only through these case study approaches that Latin American policymakers can be provided with some of the tools needed to fully comprehend the realities and complexities of the high levels of urban violence in the region.

Correspondence: Dr Caroline Doyle, Lecturer and Postdoctoral Fellow in the School of Business, The University of New South Wales, Canberra, Northcott Drive, Campbell, ACT 2600, Australia. Email: c.doyle@adfa.edu.au

Abello-Colak A (2015) Security Provision and Governing Process in Fragile Cities of the Global South the Case of Medellín 2002–2012. PhD Thesis, University of Bradford, United Kingdom.

Abello-Colak A and Pearce J (2015) Securing the global city?: An analysis of the ‘Medellín Model’ through participatory research. Conflict, Security & Development 15(3): 197–228. DOI: 10.1080/14678802.2015.1055136.

Agudelo MB (2018) Encontraron quinto cuerpo envuelo en sabanas en menos de una diaen Medellín. El Colombiano, 1 February. Available at http://www.elcolombiano.com/antioquia/seguridad/homicidio-en-robledo-aparece-quinto-cuerpo-AF8117244 (accessed 21 March 2019).

Aguirre K and Muggah R (2017) The trouble in naming Latin America’s most violent city. Americas Quarterly, 15 June. Available at http://www.americasquarterly.org/content/trouble-naming-latin-americas-most-violent-city (accessed 21 March 2019).

Alcaldía de Medellín (2011) Laboratorio Medellín: Catalogo de Diez Practicas Vivias. Medellín: Mesa Editores.

Alcaldía de Medellín (2017) Sistema de Información para la Seguridad y Convivencia. Available at https://www.medellin.gov.co/irj/go/km/docs/wpccontent/Sites/Subportal%20del%20Ciudadano/Convivencia%20y%20seguridad/Secciones/Plantillas%20Gen%C3%A9ricas/Documentos/2010/Sistema%20de%20Informaci%C3%B3n%20para%20la%20Seguridad%20y%20la%20Convivencia%20%E2%80%93SISC-.pdf (accessed 21 March 2019).

Amnesty International (2005) Colombia: The Paramilitaries in Medellín: Demobilization or Legislization? Available at https://www.amnesty.org/en/documents/amr23/019/2005/en/ (accessed 21 March 2019).

Analisis Urbano (2016) Desmembrado y enmaletado hallados en Medellín. Analisis Urbano, 7 February. Available at https://analisisurbano.org/desmembrado-y-enmaletado-hallados-en-Medellín (accessed 21 March 2019).

Asmann P (2017) Mexico victims’ survey belies growing homicides. InSight Crime, 24 October. Available at https://www.insightcrime.org/news/analysis/mexico-victims-survey-belies-growing-homicides/ (accessed 21 March 2019).

Auyero J, Burbano AL and Fernanda Berti M (2013) Violence and the state at the urban margins. Journal of Contemporary Ethnography 43(1): 94–116. DOI: 10.1177/0891241613494809.

Avila M, Martinez-Ferrer B, Vera A, Bahena A and Musitu G (2016) Victimization, perception of insecurity, and changes in daily routines in Mexico. Revista de Saude Publica 50: 50–60. DOI: 10.1590/S1518-8787.2016050006098.

Baird A (2012) The violent gang and the construction of masculinity amongst socially excluded young men. Safer Communities 11(4): 179–190. DOI: 10.1108/17578041211271445.

Barakat S and Ellis S (1996) Researching under fire: Issues for consideration when collecting data and information in war circumstances, with specific reference to relief and reconstruction projects. Disasters 20(2): 149–156. DOI: 10.1111/j.1467-7717.1996.tb00525.

Bedoya Marín DA and Jaramillo Martínez J (1991) De la Barra a la Banda. Medellín: Lealón.

Belousov K, Horlick-Jones T, Bloor M, Gilinskiy Y, Golbert V, Kostikovsky Y, Levi M and Pentsov D (2007) Any port in a storm: Fieldwork difficulties in dangerous and crisis-ridden settings. Qualitative Research 7(2): 155–175. DOI: 10.1177/1468794107076019.

Berents H and ten Have C (2017) Navigating violence: Fear and everyday life in Colombia and Mexico. International Journal for Crime, Justice and Social Democracy 6(1): 103–117. DOI: 10.5204/ijcjsd.v6i1.373.

Braakmann N (2012) How do individuals deal with victimization and victimization risk? Longitudinal evidence from Mexico. Journal of Economic Behavior & Organization 84(1): 335–344. DOI: 10.1016/j.jebo.2012.04.001.

Braun V and Clarke V (2006) Using thematic analysis in psychology. Qualitative Research in Psychology 3(2): 77–101. DOI: 10.1191/1478088706qp063oa.

Briceño-León R and Zubillaga V (2002) Violence and globalization in Latin America. Current Sociology 50(1): 19–37. DOI: 10.1177/0011392102050001003.

Burnham P, Gilland Lutz K, Grant W and Layton-Henry Z (2004) Research Methods in Politics. New York: Palgrave Macmillan.

Casas-Casas A and Giraldo Ramírez J (2015) Seguridad y Convivencia en Medellín: Aproximaciones Empiricas a sus Atributos y Desafios. Available at http://www.eafit.edu.co/centros/analisis-politico/publicaciones/proyectos-investigacion/Documents/Seguridad_y_convivencia_en_Medellin._Apr.pdf (accessed 21 March 2019).

Cerdá M, Morenoff JD, Hansen BB, Tessari Hicks KJ, Duque LF, Restrepo A and Diez-Roux AV (2012) Reducing violence by transforming neighborhoods: A natural experiment in Medellín, Colombia. American Journal of Epidemiology 175(10): 1045–1053. DOI: 10.1093/aje/kwr428.

Clavel T (2017) Medellín, Colombia security minister’s arrest shows crime–politics link. InSight Crime, 7 July. Available at https://www.insightcrime.org/news/brief/arrest-of-Medellín-colombia-security-minister-shows-crime-s-lasting-influence-on-politics/ (accessed 21 March 2019).

Coleman C and Moynihan J (1996) Understanding Crime Data: Haunted by the Dark Figure. Buckingham: Open University Press.

Corbacho A, Philipp J and Ruiz-Vega M (2012) Crime and Erosion of Trust. Evidence for Latin America (IDB Working Paper Series, No. IDB-WP-344). Available at https://www.econstor.eu/bitstream/10419/89042/1/IDB-WP-344.pdf (accessed 21 March 2019).

Creswell J (2014) Research Design: Qualitative, Quantitative, and Mixed Methods Approaches. 4th ed. California: Sage.

Development Bank of Latin America (2014) Towards a Safer Latin America: A New Perspective to Prevent and Control Crime. Available at http://scioteca.caf.com/bitstream/handle/123456789/708/RED2014-english-towards-a-safer-latin-america.pdf (accessed 21 March 2019).

Doyle C (2018) ‘Orthodox’ and ‘alternative’ explanations for the reduction of urban violence in Medellín, Colombia. Urban Research & Practice. DOI: 10.1080/17535069.2018.1434822.

Doyle C and McCarthy-Jones A (2017) Researching in volatile environments and the importance of adaptive methods for junior researchers. Qualitative Research Journal 17(4): 335–344. DOI: 10.1108/QRJ-03-2017-0005.

Dudley S (2017) Homicides in Guatemala: The Challenge and Lessons of Disaggregating Gang-related and Drug Trafficking-related Murders. Available at https://www.insightcrime.org/images/PDFs/2017/Gang-and-DTO-Homicides-in-Guatemala-Final-Report_CARSI-USAID-InSight-Crime (accessed 17 May 2018).

Duncan G and Peciado A (2018) Presentación: Combos de Medellín ¿Cómo Resolver el Problema de Grupos Armados y Violencia en Contextos Urbanos? Available at https://www.medellíncomovamos.org/download/presentacion-combos-de-Medellín-como-resolver-el-problema-de-grupos-armados-y-violencia-en-contextos-urbanos/ (accessed 3 September 2018).

El Colombiano (2015) Asesinado conductor de bus que participó en protesta por vacunas. El Colombiano, 9 October. Available at http://www.elcolombiano.com/antioquia/conductor-de-bus-que-participo-en-protesta-por-vacunas-fue-asesinado-BH2854982 (accessed 21 March 2019).

Escobar A (1995) Encountering Development: The Making of Unmaking of the Third World. New Jersey: Princeton University Press.

Ferrell J (2013) Cultural criminology. In McLaughlin E and Muncie J (eds) Criminological Perspectives: Essential Readings: 330–344. 3rd ed. London: Sage.

Garcia D (2015) Están asesinando a nuestros defensores de derechos humanos. El Espectador, 29 May. Available at http://www.elespectador.com/noticias/nacional/estan-asesinando-nuestros-defensores-de-derechos-humano-articulo-563430 (accessed 21 March 2019).

Gill H (2004) Finding a middle ground between extremes: Notes on researching transnational crime and violence. Anthropology Matters 6(2): 1–19. Available at https://anthropologymatters.com/index.php/anth_matters/article/view/100/196 (accessed 21 March 2019).

Goldstein DM (2014) Qualitative Research in Dangerous Places: Becoming an ‘Ethnographer’ of Violence and Personal Safety (DSD Working Papers on Research Security, No. 1). Available at https://www.ssrc.org/pages/qualitative-research-in-dangerous-places-becoming-an-ethnographer-of-violence-and-personal-safety/ (accessed 17 May 2018).

Gualdron Y (2012) Bandas cobrarían extorsión por usar escaleras eléctricas de Comuna 13. El Tiempo, 25 January. Available at http://www.eltiempo.com/archivo/documento/CMS-10998261 (accessed 21 March 2019).

Gutiérrez Sanín F and Jaramillo AM (2004) Crime, (counter-) insurgency and the privatization of security: The case of Medellín, Colombia. Environment and Urbanization 16(2): 17–30. DOI: 10.1177/095624780401600209.

Hidalgo WA (2013) ‘Vacunas’ personalizados. El Espectador, 10 February. Available at https://www.elespectador.com/noticias/nacional/vacunas-personalizadas-articulo-404060 (accessed 21 March 2019).

Human Rights Watch (1994) Generation Under Fire: Children and Violence in Colombia. Available at https://www.hrw.org/reports/1994/colombia/gener2.htm (accessed 21 March 2019).

Hume M (2007) Unpicking the threads: Emotion as central to the theory and practice of researching violence. Women’s Studies International Forum 30(2): 147–157. DOI: 10.1016/j.wsif.2007.01.002.

Igarapé Institute (2016) Homicide Monitor. Available at https://igarape.org.br/en/apps/homicide-monitor/ (accessed 17 May 2018).

Inter-American Development Bank (2009) Medellín: A City Transformed. Available at http://www.iadb.org/en/topics/citizen-security/impact-Medellín,5687.html (accessed 17 May 2018).

Jaitman L and Guerrero Compeán R (2015) Closing Knowledge Gaps: Toward Evidence-based Crime Prevention Policies in Latin America and the Caribbean. Available at https://publications.iadb.org/en/publication/12322/closing-knowledge-gaps-toward-evidence-based-crime-prevention-policies-latin (accessed 21 March 2019).

Kesteren J van, Dijk J van and Mayhew P (2013) The international crime victims surveys: A retrospective. International Review of Victimology 20(1): 49–69. DOI: 10.1177/0269758013511742.

Krause K (2009) Beyond definition: Violence in a global perspective. Global Violence 10(4): 337–355. DOI: 10.1080/17440570903248270.

Lamb RD (2010) Microdynamics of Illegitimacy and Complex Urban Violence in Medellín, Colombia. PhD Thesis, University of Maryland, United States.

Latinobarómetro (2011) Banco de Datos. Available at http://www.latinobarometro.org/latContents.jsp (accessed 1 October 2018).

Laverde Palma JD (2015) Ana Cristina Restrepo, la última periodista que los violentos intentan amedrentar. El Espectador, 16 March. http://www.elespectador.com/noticias/judicial/ana-cristina-restrepo-ultima-periodista-los-violentos-i-articulo-549726.

Liamputtong P and Ezzy D (2005) Qualitative Research Methods. South Melbourne: Oxford University Press.

Marinho M and Tinoco D (2017) A new campaign against Latin America’s epidemic of homicide. Open Society Foundations, 23 May. https://www.opensocietyfoundations.org/voices/new-campaign-against-latin-america-s-epidemic-homicide.

Maris L and Ortega D (2014) Crime Perceptions and Reality in Latin America (CAF Working Paper, 2014/01). Available at http://scioteca.caf.com/handle/123456789/255.

McCord J (2003) Cures that harm: Unanticipated outcomes of crime prevention programs. The ANNALS of the American Academy of Political and Social Science 587(1): 16–30. DOI: 10.1177/0002716202250781.

McDermott J (2013) New Year massacre heralds more violence for Medellín. InSight Crime, 2 January. https://www.insightcrime.org/news-analysis/new-year-massacre-violence-Medellín.

McDermott J (2014) The Changing Face of Colombian Organized Crime. Available at https://library.fes.de/pdf-files/bueros/la-seguridad/11153.pdf.

McGee R and Pearce J (eds) (2009) Violence, social action and research. IDS Bulletin 40(3). http://www.ids.ac.uk/publication/violence-social-action-and-research.

Medellín Como Vamos (2015) Presentación: Informe de Calidad de Vida de Medellín, 2014. Available at https://www.medellíncomovamos.org/presentaci-n-informe-de-calidad-de-vida-de-medell-n-2014-2015 (accessed 17 May 2018).

Medellín Como Vamos (2017) Presentación: Informe de Calidad de Vida de Medellín, 2017. Available at https://www.medellincomovamos.org/download/presentacion-informe-de-calidad-de-vida-de-medellin-2017/ (accessed 17 May 2018).

Moser C (2004) Urban violence and insecurity: An introductory roadmap. Journal of Environment and Urbanization 16(2): 3–16. DOI: 10.1177/095624780401600220.

Moser C and McIlwaine C (1999) Participatory urban appraisal and its application for research on violence. Environment and Urbanization 11(2): 203–226. DOI: 10.1177/095624789901100217.

Moser C and McIlwaine C (2004) Encounters with Violence in Latin America: Urban Poor Perceptions from Colombia and Guatemala. New York: Routledge.

Muggah R (2012) Researching the Urban Dilemma: Urbanization, Poverty and Violence. Available at https://www.idrc.ca/sites/default/files/sp/Images/Researching-the-Urban-Dilemma-Baseline-study.pdf (accessed 21 March 2019).

Muggah R (2014) Deconstructing the fragile city: Exploring insecurity, violence and resilience. Environment and Urbanization 26(2): 345–358. DOI: 10.1177/0956247814533627.

Muggah R (2015) A manifesto for the fragile city. Journal of International Affairs 68(2): 19–36. Available at http://gsdrc.org/document-library/a-manifesto-for-the-fragile-city/ (accessed 21 March 2019).

Muggah R and Giannini R (2015) Interactive map tracks murder rate worldwide. Global Observatory, 19 May. Available at https://theglobalobservatory.org/2015/05/homicide-monitor-brazil-mapping/ (accessed 21 March 2019).

Muggah R and Szabó de Carvalho I (2014) Fear and backsliding in Rio. New York Times, 15 April. Available at https://www.nytimes.com/2014/04/16/opinion/fear-and-backsliding-in-rio.html (accessed 21 March 2019).

National Institute of Statistics and Geography (2017) Encuesta Nacional de Victimización y Percepción sobre Seguridad Pública (ENVIPE) 2017. Available at https://www.inegi.org.mx/programas/envipe/2017/ (accessed 21 March 2019).

Nordstrom C and Robben A (1995) Fieldwork Under Fire: Contemporary Studies of Violence and Survival. Oakland: University of California Press.

Orjuela C (2010) The bullet in the living room: Linking security and development in a Colombo neighbourhood. Security Dialogue 41(1): 99–120. DOI: 10.1177/0967010609357039.

Patiño JE, Duque JC, Pardo-Pascual JE and Ruiz LA (2014) Using remote sensing to assess the relationship between crime and the urban layout. Applied Geography 55: 48–60. DOI: 10.1016/j.apgeog.2014.08.016.

RCN Radio (2017) Once policias han sido capturados en Medellín por vinculos con bandas. RCN Radio, 10 March. Available at https://www.rcnradio.com/locales/autoridades-capturado-11-policias-vinculos-bandas-criminales-Medellín/ (accessed 21 March 2019).

Reiner R, Livingstone S and Allen J (2003) From law and order to lynch mobs: Crime news since the Second World War. In Mason P (ed) Criminal Visions: Media Representations of Crime and Justice: 13–32. Cullompton: Willan.

Restrepo EM and Moreno AJ (2007) Bogotá: Crime or fear of crime? Revista Desarrollo y Sociedad 59: 165–214. DOI: 10.13043/dys.59.5.

Russo S, Roccato M and Vieno A (2013) Criminal victimization and crime risk perception: A multilevel longitudinal study. Social Indicators Research 112(3): 535–548. DOI: 10.1007/s11205-012-0050-8.

Rozema R (2008) Urban DDR-processes: Paramilitaries and criminal networks in Medellín, Colombia. Journal of Latin American Studies 40(3): 423–452. DOI: 10.1017/S0022216X08004392.

Salazar A and Jaramillo AM (1992) Medellín: Las Subculturas del Narcotráfico. Bogota: Cinep.

Santos BS (2014) Epistemologies of the South: Justice Against Epistemicide. London: Paradigm.

Seguridad, Justica y Paz (2017) Las 50 Ciudades Mas Violentias del Mundo 2016 + Metodologia. Available at http://www.seguridadjusticiaypaz.org.mx/biblioteca/download/6-prensa/239-las-50-ciudades-mas-violentas-del-mundo-2016-metodologia (accessed 17 May 2018).

Semana (2016) En el 80% de los barrios de Medellín hay extorsion. Semana, 8 February. Available at https://www.semana.com/nacion/articulo/seguridad-Medellín-en-el-80-de-los-barrios-hay-extorsion/459948 (accessed 21 March 2019).

Semana (2017) Capturado Gustavo Villegas, secretario de Seguridad de Medellín. Semana, 4 July. Available at http://www.semana.com/nacion/articulo/capturado-gustavo-villegas-secretario-de-seguridad-de-Medellín/531218 (accessed 21 March 2019).

Shenaz Hossein C (2016) Going local in downtown Kingston, Jamaica: Doing political ethnography and qualitative research in a volatile urban environment. Qualitative Research Journal 16(4): 345–361. DOI: 10.1108/QRJ-03-2016-0015.

United Nations Human Settlements Programme (2007) Global Report on Human Settlements 2007: Enhancing Urban Safety and Security. Available at http://www.poa-iss.org/CASAUpload/Members/Documents/22@Human_Settlements_2007.pdf (accessed 17 May 2018).

United Nations Office of Drugs and Crime (2013) Global Study on Homicide. Available at https://www.unodc.org/documents/gsh/pdfs/2014_GLOBAL_HOMICIDE_BOOK_web.pdf (accessed 17 May 2018).

Valdez A and Kaplan CD (1998) Reducing selection bias in the use of focus groups to investigate hidden populations: The case of Mexican-American gang members from South Texas. Drugs & Society 14(1–2): 209–224. DOI: 10.1300/J023v14n01_15.

Winton A (2004) Urban violence: A guide to the literature. Environment and Urbanization 16(2): 165–184. DOI: 10.1177/095624780401600208.

World Bank (2011) Violence in the City: Understanding and Supporting Community Responses to Urban Violence (Working Paper, No. 63888). Available at http://documents.worldbank.org/curated/en/2011/04/14831166/violence-city-understanding-supporting-community-responses-urban-violence (accessed 17 May 2018).

World Bank (2016) Urban violence: A challenge of epidemic proportions. World Bank, 6 September. Available at http://www.worldbank.org/en/news/feature/2016/09/06/urban-violence-a-challenge-of-epidemic-proportions (accessed 17 May 2018).

Zambrano Benavides D (2018) Hallan cadáver dentro de una maleta cera al Parque Lleras. El Colombiano, 15 January. Available at http://www.elcolombiano.com/antioquia/seguridad/cadaver-en-una-maleta-cerca-al-parque-lleras-en-Medellín-EX8011978 (accessed 21 March 2019).