Abstract
In this article, we explore patterns of prison violence in five Latin American countries: Argentina, Brazil, Chile, El Salvador, and Peru. Drawing on data from prisoner surveys conducted in 49 facilities with over 4,400 prisoners, we analyze the association between facility-level and individual-level rates of experiences of violence and the extent of perceived criminal activity committed in or ordered from inside prisons. Contrary to classical theory, neither poor prison conditions nor prior delinquent experience is directly associated with increased violence. Rather, we demonstrate that prison facilities with more widespread criminal activity inside have higher rates of prison violence. Further, within a given facility, prisoners with closer ties to criminal activity have more pre-incarceration criminal exposure and are also more likely to experience violence inside prison; this reflects research on victim–offender overlap. At a general level, our study shows that involvement in the sub-rosa economy of the prison increases one’s risk of violence in prison. We consider how common features of Latin American prisons—scarce state-provided resources, permeability to people on the outside, and more prisoner-led governance—explain these dynamics of violence inside prisons. Where prisoner-led governance is more consolidated—such as in Brazil and El Salvador—violence appears to be less common, even if criminal activity is prevalent, compared to countries where prison governance is combined or contested between authorities and prisoners. These findings suggest that prison violence reduction policies should respond to the real needs and strategies of incarcerated people rather than simply impose more control.
Keywords
Violence inside prisons is one of the most serious concerns for justice systems: It causes harm to individuals and a generalized sense of insecurity, and it amounts to additional punishment and suffering for prisoners, on top of the deprivation of liberty. In Latin America, the most visible and brutal forms of prison violence make the news (LaSusa, 2016), but more routine violence inside prisons happens with little documentation or scrutiny. This is in part because prison data in the region are generally weak. But it is also because there is less consistent formal management of daily life inside many Latin American prisons, and staff often manage prisons by relying on some degree of collaboration with incarcerated people (Darke & Garces, 2017). In this context, some acts of violence may be deliberate and “accepted” as part of the prisoner subculture, as also occurs in U.S. prisons (Skarbek, 2014; Trammell, 2009). Inside prisons, groups that manage illicit transactions exert violence strategically and instrumentally, to facilitate their activities (Gundur, 2018; Insight Crime, 2017; Skarbek, 2014). As a result, official records on violent incidents are likely incomplete at best.
To address this gap, this study uses prisoners’ self-reported experiences of violence inside prison, in 49 facilities across five countries: Argentina, Brazil, Chile, El Salvador, and Peru. Building upon theories of prison violence in the Global North, which consider prison conditions, prisoners’ prior experiences, and management tactics, we add a focus on the internal, informal governance and social dynamics among prisoners. Using data from surveys of prisoners, we take a measure of the extent of self-reported criminal activity occurring inside prison facilities, as one reflection of variation in management methods and degree of prisoner organization. We analyze how in-prison criminal activity levels shape prisoners’ experiences of violence and victimization inside prison. In our study, neither poor prison conditions nor prior criminal history, on their own, explains variation in prisoners’ experiences of violence. We find, though, that facilities with higher degrees of in-prison perceived criminal activity have higher levels of violence experienced by prisoners. Further, individual prisoners who are more involved in criminal activity inside the prison are more likely to experience violence inside, compared to other prisoners in the same prison. This aligns with research in the Global North that shows that people who perpetrate crime and violence are more likely to also be victims of crime and violence inside prison (victim–offender overlap; Pyrooz, Moule, & Decker, 2014; Rufino, Fox, & Kercher, 2012; Schreck & Stewart, 2011).
We argue that this demonstrates the importance of accounting for prisoner culture and the sub-rosa economy in relatively unregulated Latin American prisons. Although our study does not directly measure facility governance types, we contend that a plausible explanation for the association between in-prison criminal activity and in-prison violence may be that the illicit prison economy is a reflection of the formal, informal, and hybrid governance strategies found in Latin American prisons.
Review of Research on Prison Violence and Governance
Latin American prisons are as varied as U.S. prisons—ranging in political and social contexts, infrastructure and resources, programs and services, and internal culture and social interactions. There are some major differences, though. First, Latin American prisons are, on average, far more overcrowded, with many countries at double or triple capacity and some well over 500% overcrowded (Darke & Karam, 2016). This is due in part to much higher rates of pretrial detention. The rate in the United States is about 21%, while the rates in Latin America range from 13% in Costa Rica to 77% in Paraguay, with most countries around 50% (Institute for Criminal Policy Research, 2018). Ten of the 20 countries worldwide with the highest pretrial detention rates are in Latin America (Schönteich & Varenik, 2014). This situation is due to harsher drug laws since the 1990s, lack of alternatives to pretrial detention, and problems and poor capacity in investigative and judicial processes, among other factors. A second difference between regions is that there is typically no distinction between prisons and jails in Latin America. People in pretrial detention are, with some exceptions, held in the same facilities and conditions as sentenced people. Third, in most Latin American countries, the services and programs inside prisons are very uneven, ranging from extensive to zero. Similarly, parole systems and supports for people postrelease are minimal (usually run by churches) and, in many places, nonexistent. Finally, the management of prisons in Latin America is, with some notable exceptions like Costa Rica, in the hands of police or military institutions, with some civilians in certain roles. Some systems are federations, with state-level and federal-level prison systems (e.g., Mexico, Argentina), whereas many are national, with a single prison system (e.g., Colombia, Chile). In general, prisons in the region are more permeable and unregulated than U.S. prisons, with much lower staff-to-inmate ratios.
Due to all these factors, there is limited research on prison violence in Latin America. Scholars of prison life in the region describe elaborate arrangements of rules and both violent and nonviolent enforcement tactics (Antillano, 2015; Biondi, 2017; Darke & Garces, 2017). Because most research on prisons in Latin America is ethnographic and within single facilities, there are few studies analyzing violence and misconduct across multiple facilities or years (notable exceptions exist for Chile [Sanhueza, 2014] and Mexico [Bergman, Fondevila, Vilalta, & Azaola, 2014]).
In general, official data from prison institutions in Latin America cover population numbers and demographic traits but rarely include reliable records on programs, risk assessments, misconduct, violence incidents, or recidivism. Much North American and European research on prison violence relies on official data about misconduct (Harer & Langan, 2001; Morris, Carriaga, Diamond, Piquero, & Piquero, 2012; Scott, Petrossian, & Mellow, 2018; Wooldredge, Griffin, & Pratt, 2001) or, for facility-specific detail, surveys with prison wardens (Bierie, 2011; Hensley, Koscheski, & Tewksbury, 2003; Tartaro, 2002). Other scholars use self-reported data—surveys or interviews with prisoners—because this provides a more complete and honest picture (Celinska & Sung, 2014; Hochstetler & DeLisi, 2005; Lahm, 2009; Rocheleau, 2013; Tasca, Griffin, & Rodriguez, 2010). Although standardization is helpful, the principal risk with official data is that it may misrepresent or omit a significant portion of violence that occurs because both prisoners and prison officials have incentives to minimize reported violence (Byrne & Hummer, 2007). In Latin American prisons, variation in institutional style and capacity generates official data that are difficult to compare cross-nationally. The key limitation of self-reported data is, of course, that prisoners may understand a given category or question differently—such as “physical assault.”
Deprivation Theory, Importation Theory, and Integrations
Most contemporary research upholds some integrated version of the two classic prison violence theories—deprivation and importation—noting that factors from both theories, plus management factors, matter. Deprivation theory (Clemmer, 1940; Sykes, 1958) proposes that the main explanation for violence lies in the nature, extent, and variation in the “pains of imprisonment,” such as family contact, heterosexual sex, safety, variety in food and daily routine, and opportunities for self-determination. Such suffering is shaped by one’s previous experiences and expectations (Hochstetler & DeLisi, 2005), and people respond differently to the same conditions (Blevins, Listwan, Cullen, & Lero Jonson, 2010; Morris et al., 2012). Research has found associations between worse conditions and higher rates of serious violence, specifically emphasizing the effects of poor conditions (Bierie, 2011), social and emotional “strain” (Morris et al., 2012; Rocheleau, 2013), and reduced visitation (Cochran, 2012). Strain theory more generally posits that people with limited options for meeting their basic needs may resort to violence or crime, including in prison (Agnew, 2006; Blevins et al., 2010; Merton, 1938). The severity and range of “deprivations” in Latin American prisons can be dramatic, given the material constraints, but these prisons often permit outside comforts, even for poorer prisoners, such as conjugal visits or food brought in by families.
Overcrowding is extreme in Latin American prisons and often stands as a proxy for harsh conditions generally. However, research findings are mixed about whether overcrowding increases violence in U.S. prisons (Steiner, Butler, & Ellison, 2014; Tartaro, 2002). Some find that overcrowding effects are moderated by other factors such as prisoners’ age (Lahm, 2008), facility differences (Camp, Gaes, Langan, & Saylor, 2003; Hochstetler & DeLisi, 2005), nonviolent misconduct (Franklin, Franklin, & Pratt, 2006), or the legitimacy of authority systems (McCorkle, Miethe, & Drass, 1995; Useem & Piehl, 2006).
Importation theory, originally articulated by Irwin and Cressey (1962), contends that the preexisting experiences and attitudes that prisoners bring into prison, as well as demographic traits, shape violence patterns. Generally, research finds that prisoners who are male, younger (particularly under 25), have more severe conviction types, 1 have a history of criminal activity or previous violence in custody, and who have spent less time in prison are more likely to commit violent misconduct (Cunningham, Sorensen, Vigen, & Woods, 2011; Lahm, 2008; Steiner et al., 2014). Other factors such as gang membership, security or actuarial risk classification, race/ethnicity, mental illness, and longer term exposure to delinquent peers show mixed results across studies (Hochstetler & DeLisi, 2005; Meyer, 2010; Steiner et al., 2014).
Most contemporary research integrates both theories, accounting for facility conditions, individual traits, and other factors (Blevins et al., 2010; Hochstetler & DeLisi, 2005; Huebner, 2003; McCorkle et al., 1995; Wooldredge et al., 2001), or aggregating individual indicators at the facility level (Camp et al., 2003). Notably, one of the few studies to use system-wide quantitative data on prison violence in Latin America (Sanhueza, 2014) takes this approach.
Prison Management Considerations
Contemporary research also integrates staff and management factors into combined theories, although it is difficult to categorize this in a comparative way across settings (Mears, 2008). Variations in staff–inmate ratios and demographic composition of staff sometimes matter, but not in consistent ways (Camp et al., 2003; Reisig, 1998). But what matters most is the social culture between staff and prisoners and among prisoners (Bosworth, 2003; Bottoms, 1999; Carrabine, 2005; Liebling, 2004). Balanced approaches by formal management entities—using positive and negative incentives and formal and informal ways of handling problems—are generally more successful at keeping order and at achieving other goals (Byrne & Hummer, 2007; Cooke, Wozniak, & Johnstone, 2008; Craig, 2004). Outside oversight by courts can generate improvements in both material conditions and in order and discipline, at least for the short term (Boylan & Mocan, 2013; Taylor, 2013). Of course, formal management strategies shape and are shaped by prisoners’ culture, organizations, and reactions, too (Sparks & Bottoms, 1995). Informal prisoner-led governance factors are discussed in more detail below.
International Research on Prison Violence
Most research on prison violence focuses on systems in the Anglo-American North, and there are few cross-national studies at the global level (Arbach-Lucioni, Martinez-García, & Andrés-Pueyo, 2012; Singh et al., 2014). A system-wide study in Chile found that factors noted in Global North research—such as age, prior history, staff-prisoner interactions, and conditions—are also significant in Chile, but also that definitions of quality conditions and relationships must be adapted to the local context (Sanhueza, Valenzuela, & de los Ángeles Smith, 2015). Qualitative studies emphasize that the nature and extent of violence depends largely on which organized group inside controls the use of violence and how they wield this role (Butler, Slade, & Dias, 2018; Darke, 2018; Symkovych, 2017). This aligns with North American research on prison gangs (Skarbek, 2014), inmate codes and subcultures (Trammell, 2009), and prisoners’ understanding of threats from both formal and informal authorities (Maier & Ricciardelli, 2018).
Informal Governance and Gangs in Prison
In Latin American prisons, given the lack of amenities and relatively low staff–inmate ratio, prisoner-led organizations tend to have more prominence and control over daily life and resource distribution. Across the region, prisoners form groups to organize for collective needs, such as cooking and distributing food, allocating sleeping space, handling the logistics of family visit days, and sometimes quasi-administrative roles like counts or leading classes (Darke, 2018; Darke & Garces, 2017; Darke & Karam, 2016; Macaulay, 2013). Within these groups, some are based on quasi-democratic processes, such as voting for representatives, while others are based on local affiliations by neighborhood, cellblock, or church group (Antillano, 2015; Darke, 2018; Skarbek, 2016). Some prisoner-led groups in Latin American prisons align with Buentello, Fong, and Vogel’s (1991) framework of evolution from clique to protection group to predator group (see Gundur, 2018), while in some cases, the formal administration collaborates with factions of inmates in order to hold control, analogous to what Irwin (1980) describes in pre-1970s California prisons. In Brazil in particular, prisoner-led organizations quickly transformed into gangs (specifically the Primeiro Comando Capital) that eventually also gained power in the streets (Butler et al., 2018; Lessing, 2016; Nunes Dias & Darke, 2016). Meanwhile, in Central America, street gangs (such as MS-13 and 18) subject to police crackdowns fortified themselves inside prisons, as they are separated by facility (Insight Crime, 2017; Rosen & Cruz, 2018).
Prison gangs are prominent in studies of prison violence (Pyrooz, Decker, & Fleisher, 2011), largely through their role in running illicit economies inside (Skarbek, 2014). A major overview of gangs in Latin American prisons identifies at least six types of gangs, based on their goals and ties to outside communities (Insight Crime, 2017). In line with U.S. studies (Skarbek, 2014; Worrall & Morris, 2012), the way that gangs operate inside prison matters more than whether an individual person is formally a member, since all prisoners and staff must navigate gang dynamics. Prison gangs regulate and enforce economic and social transactions where information and trust are severely constrained (Skarbek, 2012). Their structure and tactics (monopoly, vertical, horizontal) depend largely on how the government seeks to control them and on whether the gang coheres through ethnic, ideological, and/or economic ties (Butler et al., 2018; Gundur, 2018). When gangs have leverage with people who are connected to outside communities, as is common in Brazil and Central America, they can force compliance by prisoners to the rules they impose inside prison facilities by threatening consequences on the street (Lessing, 2016; Nunes Dias & Darke, 2016). Some groups form inside prison for specific reasons, such as protection or for managing drug sales (Gundur, 2018). In some countries, such as Venezuela, gang-like groups have near-total control inside the facilities (Antillano, 2015), while in Central America, they negotiate power with staff (Weegels, 2018) and with street gangs (Rosen & Cruz, 2018), and in Chile and Argentina, prisoner-led governance groups exist but are not tied to gangs (Insight Crime, 2017; Sanhueza, 2014).
Regardless of the type of group that holds power in informally governed prisons, the rules they set—the inmate code or subculture, in U.S. parlance—affect what kinds of violence are permitted (Trammell, 2009). Rules typically prohibit snitching, fighting, stealing, disobeying the hierarchy of the prisoner governance system, and anything that disrupts the illicit markets that operate inside the facility. These markets involve legal goods (space, food, water, communications, and other commodities, sometimes sex) and illegal goods (drugs, weapons, sometimes sex, etc.) (Darke, 2018). When formal institutional recourses are not available for enforcing contracts and resolving interpersonal disputes, groups of prisoners may use threats and violence to do so (on criminal governance and violence generally, see Barnes, 2017; on this dynamic inside prisons, see Skarbek, 2011). Such dynamics are noted by prison scholars in the United States (Gundur 2018; Skarbek, 2014; Trammell, 2009), Canada (Maier & Ricciardelli, 2018; Ross, 2017), the former Soviet Union (Piacentini & Slade, 2015; Symkovych, 2017), Brazil (Biondi, 2017; Darke, 2018), and Venezuela (Antillano, 2015).
This Study: Hypotheses, Methods, and Data
This study explores the question of whether and how the extent and nature of criminal activity inside a prison facility may explain variation in experiences of violence inside. The central question of this study is: How does criminal activity inside a prison influence the degree and types of violence that prisoners experience? Further, it analyzes which kinds of prisoners are closer to criminal activity and how this proximity is associated with their experiences of violence and victimization. To do so, we use data from surveys of prisoners, conducted in 49 facilities in five Latin American countries (Argentina, 9 prisons; Brazil, 9 prisons; Chile, 8 prisons; El Salvador, 9 prisons, and Peru, 14 prisons).
This study has two hypotheses:
The following predictors of experiencing prison violence are included as controls: individual demographics, criminal history, and conviction or sentence type and two measures of facility conditions. The first is a measure of overcrowding, and the second is a prison conditions index (per facility), calculated based on access to 19 basic services, such as a toothbrush, clean water, and school. It also controls for an estimated “security level” of the prison, based on the aggregated percentage of prisoners with homicide and kidnapping charges, since institutional security designations and inmate risk assessments are not widespread (see descriptive statistics in Appendix B).
The main outcome variable in this study is physical violence, as reported by prisoners. We also include sexual violence through indirect experience (observation of this happening to another person) since this encourages a more candid response—though measures of sexual violence are highly susceptible to survey design and stigma (Wolff, Blitz, Shi, Bachman, & Siegel, 2006). We also measure prisoners’ self-reported experiences of property theft, as a nonviolent type of victimization. 2 Other forms of misconduct are excluded.
The main predictor variable is criminal activity directed from or committed inside facilities. We acknowledge that this includes two types of criminal activity in a single question. The most common types of crime reported in the survey are extortion, drug selling, robbery, kidnapping, and homicide. But more minor crimes related to the illicit economy likely occur as well. Overall, we assume that a portion of prisoners in any given facility will be involved in this type of activity. Official data are used to build only a few variables, mainly related to prison population characteristics and overcrowding rates.
Data
The data in this article come from the Survey of Incarcerated Populations in Latin America, a research project designed and managed by scholars from Centro de Investigación y Docencia Económicas (CIDE Mexico), including the coauthor on this article (Fondevila) and CELIV (Universidad Tres de Febrero, Argentina; Bergman, Amaya, Fondevila, & Vilalta, 2015). The team administered the survey in 2012–2013 in prisons in Argentina, Brazil, Chile, El Salvador, and Peru. These countries were selected for two principal reasons. First, they were the countries where government partnerships and funding enabled data collection. Second, they have characteristics in their prison systems that allow for comparisons. Argentina, Chile, and Peru share similar systems with enough funding to provide basic services for inmates. Brazil and El Salvador have overcrowded prisons that lack many basic services, and state control varies from partial to minimal, as gangs and inmates’ groups provide some services an organization. Some features make for interesting single-country analysis. For instance, Chilean prisons tend to be quite violent, as low-end criminals almost never end up in jail due to alternatives to incarceration and other mechanisms. Gangs in Argentina and Peru are almost nonexistent and prison control tends to be handled by ad hoc prisoner groups in each prison. Brazilian prisons have a mix of nonviolent criminals, gang members, and very violent criminals, as imprisonment (and pretrial detention) is the most common punishment in the criminal justice system.
The survey questions are based on CIDE’s long-standing survey of prisoners in Mexico (Bergman, Fondevila, et al., 2014). The overall methodology of the survey follows established protocols and permits specific adaptations required for particular national or local contexts (Bergman, Flom, & Vilalta, 2014). The surveys for these five countries were funded by the United Nations Development Program (UNDP) and are published as part of the 2013–2014 United Nations (UN) Human Development Report for Central America (UNDP, 2013). 3
The total sample across the five countries is 4,437 surveyed inmates in 49 facilities. We used a multistage sample process, through stratified random sampling of prison facilities (creating strata using prisons from each provincial area [for Chile and El Salvador] or metropolitan area [for Sao Paulo, Buenos Aires, and Lima]), with systematic selection of observations and with a gender quota (explained below). 4 We used two sampling frames. First, for the selection of prison facilities, we aimed for purposive variation in types of facilities and also used feasibility factors (e.g., our project budget and the distance between facilities). Selection by security level was not feasible, as most Latin American prisons do not use such classifications for entire facilities. Second, within each facility, we selected from the inmates who were willing to participate in each unit or area. In facilities with separate units inside, we surveyed proportional numbers of individuals in each unit. Although pretrial detention populations are high in Latin American prisons, our sample includes only sentenced inmates. We selected individual respondents using the official list of inmates and a systematic jump procedure, with a random starting point, in three steps. 5 If a selected inmate refused to participate, then we chose the following inmate on the list, using the systematic jump procedure. The sample was also stratified by inmate sex: Because women make up only 5% of the total incarcerated population, we oversampled women in the surveys. The estimates of key variables are different across countries and regions due to the effects of survey design and response rates of each facility’s population. 6
Survey Development and Administration Process
The survey administration team (hired by UNDP through private pollsters) completed the surveys with inmates through an in-person interview, asking questions verbally and recording answers on paper, without recording names. On average, in each country, the interviewer team was composed of between 12 and 15 interviewers who were trained by the lead researcher in two 4-hr sessions. Due to the prison authorities’ rules and restrictions, the team used paper-and-pencil interviewing. The survey team typically spent 1 or 2 days in each facility. The interviews lasted roughly 45 min (each) and took place in a specifically designated location within each facility or unit—away from any guards or prison staff, but with explicit permission from the authorities.
The survey questionnaire has 310 (closed) questions that cover sociodemographic, criminal, institutional, living standards, and other issues. The questionnaire is based in part on the National Inmate Survey of the Bureau of Justice Statistics in the United States, especially on the main categories of criminal justice history and prison conditions and programs. The prison administrators had to authorize all questions and screened out some sensitive topics, such as a direct question on staff corruption. As such, some questions had to be written in a more indirect way. The relevant question in this study was adjusted for this reason: “Do you know of crimes being committed inside or being directed from inside prison?”
There was an average refusal rate of 0.4% of inmates summoned for the interview (during the second day, the rejection rate increases considerably to 2%). This low refusal rate is likely due to the fact that this was the first research project in most facilities and prisoners were eager to share their views and, also, the prison staff gave ample space and autonomy to the research team.
As in any prison research, the formal and informal surveillance that prisoners experience influences the way they participate in interviews and surveys. Prison conditions and length of stay may influence perceptions and what people report to researchers (Bulman, Garcia, & Hernon, 2012). Because it is difficult to isolate these effects, it is advisable that studies of prisons develop a (weak) quasi-experimental methodology to compare groups and explain their differences. In this study, we compare prisoners’ experiences of violence based on factors affecting groups of prisoners: facility conditions, overcrowding levels, extent of programs in a facility, and degree of criminal activity inside a facility. We use various individual-level variables as controls.
Missing Data
Although the response rate out of total potential respondents was quite high (over 98%), not all respondents answered every question in the survey. It is difficult to ensure that surveyed inmates answer all questions, perhaps due to fear of retaliation or a desire to present a certain image to the surveyor. Despite taking many precautions for privacy, prisoners may still hesitate to answer some questions. We did not note any pattern of respondents skipping particular questions—such as those to do with violence or drugs—more frequently than others. We did not impute data for questions where responses were missing.
For this study, the key question is the one that asks about a person’s knowledge of crimes inside prison. On this question, 83.7% of respondents provided an answer (3,709 valid responses of 4,437 surveys). This limits the present analysis to a maximum of 3,709 surveys. Within this set, in each analysis, if an answer was missing on another relevant variable, we dropped the case from the analysis. In the multivariate analysis (Appendix D), the smallest number of valid cases was 3,411. In the bivariate analyses of individual variables, the smallest number of valid cases was 594 (on a question about recent use of crack/cocaine).
Variables
Following our hypotheses outlined above, we explore experiences of violence and victimization as outcome variables. These questions are all dichotomous (yes/no). Prison violence: In the last 6 months, have you been beaten? Sexual violence: Have you seen other inmates been forced to have sexual relations against their will?
7
Property theft victimization: Have your personal possessions been stolen in prison?
We include property theft as a dimension of victimization even though it is not necessarily violent. The key independent predictor variable is the extent of crimes committed from inside or inside the prison. This is measured through this question (dichotomous): ○ Do you know of crimes being committed inside or being directed from inside prison?
We acknowledge that there are limitations to a dichotomous question. When a higher proportion of prisoners answers “yes” to the question, this suggests a more extensive and active illicit economy and level of organization.
Each of the 49 prisons was classified into one of five groups according to the percentage of inmates that responded positively to this question about crimes inside. The five groups are defined as follows: a very low level of crimes (less than 10% of inmates answered “yes”), low (between 10% and 20%), medium (between 20% and 30%), high (between 30% and 40%), and a very high level of crimes (more than 40% of inmates answered “yes”). In these categories, 11 prisons were classified in “very low,” 17 in “low,” 8 in “medium,” 7 in “high,” and 6 in “very high.”
In the logistic regression, we include other factors at the facility and individual level.
Facility factors
○ Percentage of sentenced for homicide and kidnapping in the facility (0–100%). ○ Overcrowding (1/0): Are there inmates in your cell who do not have a bed? ○ Prison conditions: A score based on what percentage of prisoners answered yes to questions on specific amenities and services.
The percentage of people sentenced for violent crimes is an estimation of the “security level” of a facility, in a crude sense. The latter two factors—overcrowding and the prison conditions index—reflect dimensions of the “pains of imprisonment,” part of deprivation theory.
Individual factors
Demographic factors are controls; we test the criminal justice history factors as part of importation theory. ○ sex, ○ age, ○ conviction type (coded for the type of crime), ○ sentence length, ○ time in prison so far, ○ individual drug use in prison (yes/no), ○ participation in prison programs (yes/no), ○ prior experiences with violence (yes/no), ○ prior experiences with weapons (yes/no), ○ was in a juvenile detention institution (yes/no), ○ had contact with gangs during childhood (yes/no), and ○ robbed people or sold drugs as a juvenile (yes/no).
See Table 1 for descriptive statistics of the sample. A description of the statistics of all variables can be found in Appendix B.
Descriptive Statistics: Sample by Country and Enrollment Process.
Note. Data on prison population rate and occupancy level are from Institute for Criminal Policy Research (www.prisonstudies.org). Data on prison population are from the corresponding prison authority.
Findings
Descriptive Findings: Extent of Criminal Activity Inside Prison
Across all five countries in our study, almost one in four inmates answered affirmatively that they knew of criminal activity inside the prison (23.5%). This proportion is higher in Chile and Peru, with 36.7% and 33.5%, respectively, and lowest in El Salvador, where only 11.1% of respondents answered that they have knowledge of crimes inside prison. When disaggregating by type of crime they knew about being committed in or directed from prison, the most common types, across all five countries, are extortion (9.0%), followed by drug trafficking (5.8%), robbery (5.2%), kidnapping (4.9%), and homicide (2.5%). These rates may seem low individually, but the aggregate level of crimes is higher, though admittedly there are reasons for prisoners not to report this honestly, particularly in gang-controlled facilities (see Appendix A).
This shows the variation in the approximate level of criminal activity inside prisons, as reported by inmates, across facilities and across national systems. Notably, within each country, there is a wide range from 5.9% to 35% in Argentina, from 0% to 16% in Brazil, from 10% to 57% in Chile, from 0% to 11% in El Salvador, and from 11% to 33.5% in Peru (see Table 2 for details). As outlined in the review of relevant research, it is reasonable to assume the prisoners who report knowledge of organization of crimes in prison are closer to—though not necessarily participating in—the prisoners who are directly involved in the criminal activity. Due to the risks of disruption or punishment, those involved in crimes have reason to keep this information limited to people who are involved and/or trusted. The rest of the inmates in the facility may be generally aware that illicit or criminal acts occur, but they do not have actual knowledge of these crimes. Many also likely deliberately distance themselves from such information. Research in both North America and Latin America suggests that committing crime inside prisons requires collaboration and organization among a group of prisoners—but this study does not directly measure the nature or strength of that organization. Rather, it analyzes the extent of in-prison criminal activity and of violence and property theft inside, at the facility level and the individual level.
Summary of Prisons per Country, Disaggregated by Level of In-Prison Crimes.
Prisoners With Knowledge of Criminal Activity Inside Prison Report Higher Criminal Exposure and Involvement Pre-Incarceration and During Incarceration
Prisoners with knowledge of criminal activity inside prison report experiences of previous criminal activity and exposure significantly more (p < .01) than do prisoners with no knowledge of criminal activity in prison (see Appendix C). These indicators include pre-incarceration criminal exposure 8 and pre-incarceration criminal activity and severity. 9 Moreover, those involved in crimes inside prison were more likely to have been unemployed or working with law enforcement prior to arrest. Not surprisingly, prisoners who know about crimes inside prison have spent longer in prison (over 10 years), compared to prisoners who do not know about crimes inside. This difference may be due to having longer sentences due to charge severity. But a more plausible explanation is that the longer one spends in prison, the more one has awareness of crimes committed inside and opportunities to be involved in such crimes. Moreover, time inside prison builds trust and relationships that can be economically beneficial during and after prison. Because punishment for in-prison crimes is unlikely and formal parole decisions are discretionary, the incentives to participate in the prison illicit market may outweigh the incentives of parole criteria. Prior and current substance use 10 is also associated with knowledge of in-prison crime—which makes sense because drug sales are a major part of most illicit markets inside prison (see Appendix C for detailed statistics).
When disaggregating by country, there is some notable variation in terms of which specific risk factor is significantly associated with knowledge of in-prison crime. For example, gang involvement matters in Brazil, where 25.8% of prisoners who grew up with gangs report in-prison criminal activity, compared to only 8.2% of those who did not grow up with gangs in their neighborhood; the difference is modest in Argentina (18.8% vs. 20.1%). Also, in El Salvador, inmates charged with a gang-related crime are twice as likely to have knowledge of crimes (10.2% vs. 21.8%), which is not the case in Chile (36.5% vs. 40%). In Peru, previous violent crime matters: more than half (52.7%) of inmates who have wounded or killed someone report in-prison crimes, compared to 30.4% whose have never wounded or killed anyone, whereas in El Salvador, the difference is marginal (10.2% vs. 11.8%). Experience with firearms is associated with knowledge of in-prison crime in Argentina, Brazil, and Chile, with 22.3% vs. 9.5%, 19.6% vs. 9%, and 41.9% vs. 22.3%, respectively, but not in El Salvador (12.8% vs. 9.4%).
In summary, almost one in four inmates in Latin America has knowledge of crimes being organized from inside the prison. This group of prisoners also has higher involvement in and exposure to criminal activity and peers, both before incarceration—including during childhood—and during incarceration. In short, they appear to continue participation in illegal activities once inside prison. These factors all represent the key elements of importation theory—the experiences and values that an individual brings with him or her into incarceration. Building on importation theory, we tested whether people with more previous exposure to crime and violence experience significantly more violence inside prison. This direct association was not significant—which suggests that simple importation theory is not a sufficient explanation. However, it is important to note that prior exposure to crime and violence is related to knowledge of crimes inside prison, which is the principal predictor variable we analyze, in the following section, first at the facility level and then at the individual level.
Association Between Facility-Level Criminal Activity and Individual Experiences of Violence
Our analysis finds that prisoners who reside in facilities with higher levels of crimes organized inside report experiencing more violence and property theft—with the exception of the association with physical violence for three countries. Property theft: In the prisons with higher proportions of people reporting knowledge of crimes committed inside, more people report having their possessions stolen. Physical violence: In the prisons with higher proportion of people reporting knowledge of crimes committed inside, more people report being beaten by other prisoners (A) and by staff (B). This relationship is significant for Argentina, Brazil, and for the five countries combined, but not for Chile, El Salvador, and Peru. Sexual violence: In the prisons with higher proportion of people reporting knowledge of crimes committed, more people report experiencing nonconsensual sex. This relationship is significant for Argentina and the combination of countries, but not for the other countries. Notably, reports of sexual violence observed are higher in male facilities than female facilities.
Across all countries, in prisons with a very low level of crime inside, 17.9% of prisoners experienced property theft, while 50% of prisoners in prisons with a very high level of organization of crimes reported property theft. On physical violence, the proportion of inmates who report having been beaten goes from 3.4% in very low-crime facilities to 21.1% in very high-crime facilities. On sexual violence, the proportion of inmates who report having seen others forced to have sex rises from 2.1% in very low-crime facilities to 14.3% in very high-crime facilities (see Table 3 and Figure 1 for details). Per prison, the correlation between knowledge of crimes and physical violence is .629 (N = 49; p < .001), with property theft the correlation is .581 (N = 49; p < .001) and with sexual violence the correlation is .37 (N = 49; p = .008).
Percentage of Inmates Who Have Suffered Violence Within the Prison, Disaggregated by the Level of Crimes Committed Within the Prison.
Note. Test of difference of proportions with the immediately previous level of organization of crimes.
Level of significance: **.01 and *.05.
Percentage of inmates who have suffered violence within the prison, disaggregated by the level of crimes committed within the prison. Bars indicate a 95% confidence interval. Data from study survey.
Thus, Hypothesis 1 (the level of criminal activities organized inside a prison is positively associated with the level of violence suffered by prisoners) is not rejected.
As shown in Appendix B, violence levels vary widely among the countries. Brazil and El Salvador have the lowest levels of violence while Chile and Argentina have the highest. Nonetheless, when disaggregating the violence by level of organization of crimes, in all countries, we observe a tendency for more violence in the prisons with the highest levels of organization of crimes. For instance, in Argentina, the percentage of inmates who have been beaten in prisons with “very low level” of crimes inside is 10% and increases up to 19.6% in the prisons with “high level” of crimes inside. In Brazil, 1.5% of inmates in prisons with very low or low levels of crime organized inside have been beaten, while 15.4% report beating in prisons with medium levels of crimes organized inside. This pattern is similar in Chile (which has the most violent prisons): 10% were beaten in prisons with low levels of criminal activity inside, while 28.3% were beaten in prisons with high and very high levels of criminal activity inside. The exception is El Salvador, where physical violence does not correlate with any level of organization of crimes. We speculate that this is due to the unique role of gangs in Salvadoran prisons, with facilities almost entirely run by a given gang that exerts hierarchical control over both criminal activity and the use of violence among prisoners (see Figure 2).
Percentage of inmates who have knowledge of crimes organized inside the prison and percentage who have suffered physical violence, disaggregated by country.
Association Between Prison Conditions and Violence
Following deprivation theory, we also tested the direct association between prison conditions at the facility level and experiences of violence. To do so, we used the Prison Conditions Index, which is built based on how many individuals in each facility responded that they had access to a series of basic services. In aggregate, El Salvador’s prisons have the lowest conditions score and Argentina’s prisons have the highest conditions score (see Appendix B for details). In the mixed effects logistic regression of experiences of violence and theft and the prison conditions index (per facility), there are no significant associations. Similarly, we tested the association between the overcrowding rate of each facility and the prevalence of prisoners’ experiences of violence. This association was also not significant (see Appendix D for these results.)
Further, we tested the direct association between the “security level” of the prison and experiences of violence. As discussed above, prisons in the countries we studied are not classified as maximum, medium, and minimum security. Instead, we assigned each prison a score of “severity” based on the percentage of prisoners inside who had a sentence due to homicide and kidnapping—the most severe charges. We have no way to determine whether conditions or security measures are different for these facilities, but it is reasonable to assume that to the extent that the administration could exert stronger controls on prisoners’ movements and privileges, they would apply these on the prisons holding the people with the most violent charges. (Risk assessments on factors other than charge are not widely used.) Using this approach, we find no significant associations between the proportion of prisoners with severe violent crime charges and the prevalence of experiences of violence at the facility level (see Appendix D)
These findings—about the role of prison conditions, overcrowding, and of the “severity” of the facility—suggest that deprivation theory alone is not a sufficient explanation for the prevalence of violence inside prisons. Although this direct association is not the main hypothesis in this study, we nonetheless anticipated that harsher prison conditions and more overcrowding would be associated with more violence. Given that our findings do not show support for a direct link between violence in prison and individual criminal history (importation theory) or facility conditions (deprivation theory), we turn to other dynamics inside prisons, to build a more nuanced picture.
Individual-Level Analysis
Now we move to an individual analysis to test Hypothesis 2 (the violence is concentrated among certain types of prisoners inside a given facility). To analyze the individual characteristics associated with prison violence, we use a model of logistic regression. We replicated the same model for the property theft (nonviolent insecurity) and for two types of violence: physical and sexual (indirect)—all dichotomous variables. Whether a person suffered these types of violence is the dependent variable, and the predictor variables are (a) the individual’s knowledge of crimes organized from inside prison (individual—dichotomous) and (b) the level of criminality within the prison (from very low to very high, as per the levels described above).
Control variables for individual demographic and social characteristics are age, sex, education level, and history of exposure to criminal activity before the current sentence (friends who committed crimes, previous prison sentences, use of firearms, and use of drugs). Variables on the individual participation in community activities and the consumption of substances such as alcohol, marihuana, cocaine, and ecstasy in the last 6 months are also included. Dichotomous variables for 10 kinds of crimes were included to control for the severity of the crime for which the person is currently incarcerated (intentional homicide, manslaughter, sexual crimes, etc.). Finally, dichotomous variables for each of the five countries were included to control for differences in criminal and prison systems.
Our analysis finds that sex and age are more important than education level. There is a significant difference by sex for property theft—men experience higher property theft than women—but differences are not significant for the other types of violence. (This is rather surprising, given that one might expect women to report higher levels of sexual violence.) Younger inmates (up to 35 years old) suffer more physical violence compared to older inmates. But for sexual violence, older prisoners report higher observations. There are no significant differences by age for property theft. Individual education level does not affect the level for any of the types of violence or insecurity.
Few factors related to pre-incarceration exposure to criminal activity are significant. People’s previous delinquent peer ties and prior use of a weapon moderately increase the probability only for sexual violence but not for property theft or physical violence. Surprisingly, previous incarceration (as an adult or juvenile) is not significantly associated with individual experiences of violence. On substance use, past individual consumption of marijuana, but not other substances, moderately increases the probability of property theft, while past consumption of ecstasy moderately increases the probability of physical violence. (This may reflect the relatively low prevalence of cocaine and heroin.) Once inside the prison, if a prisoner consumes marihuana, alcohol, or cocaine, the likelihood that a person suffering violence, especially physical violence, increases.
The type of crime for which the person was sentenced affects their probability of experiencing physical violence and property theft but has a particularly significant effect on their probability of observing sexual violence inside prison. Those sentenced for crimes with major social impact and visibility—namely, intentional homicide, kidnapping, assault, sexual crimes, aggravated robbery, 11 small-scale drug selling, or simple theft—have observed more sexual violence than inmates sentenced for other crimes, such as manslaughter or fraud. The only sentence type that increases a person’s probability of experiencing or observing all types of violence is a sexual crimes sentence; this likely reflects the reality that people with this charge tend to be targeted by or marginalized by other prisoners, due to stigma. It is notable that type of crime does not seem to matter overall for the probability of experiencing theft or physical violence.
Conversely, individuals’ participation in various prosocial activities inside prison—sports, academics, and a job—overall does not have any significant relationship with their likelihood of experiencing violence or theft (see Appendix D). Participation in such activities depends on the level of opportunity and access to such programs, which varies dramatically by prison (and is reflected in the prison conditions index discussed above). But propensity and personal choice to participate also reflects “positive values,” from the perspective of importation theory. The only significant relationship is between participation in academic activities and likelihood of experiencing property theft, which likely reflects a higher socioeconomic position, and therefore may be more about crimes of opportunity instead of individual values.
Discussion
The first notable finding from our analysis is that a direct test of deprivation factors—prison conditions and overcrowding—and of importation factors—prior criminal activity/exposure—shows that these are not significant predictors of prisoners’ experiences of violence and theft during incarceration. This stands in contrast to classical North American and European theories in their simplest form but aligns with more recent research suggesting that more multifaceted analysis—accounting for factors like prison management styles—is required. Although a comprehensive, multidimensional analysis of management factors is outside the scope of this study, our analysis of criminal activity inside prison as a potentially important factor is part of this more complicated picture. On this front, our findings generally provide support for our hypotheses. First, we find that higher levels of perceived criminal activity inside facilities are associated with higher levels of prisoners’ experiences of violence. Second, at the individual level, prisoners who are more closely tied to criminal activity are more likely to experience violence, compared to others in the same facility. As explained above, we consider criminal activity inside prison to be not just an important factor on its own but also a reflection of certain informal governance dynamics in the prisoner culture and in prisoners’ negotiated power arrangements with staff. In the following section, we consider what our findings and the Latin American prisons context suggest in terms of understanding criminal activity inside prison and, taking a step further, informal prison governance issues, as relevant to the existing theories of prison violence (deprivation theory, importation theory, and an analysis of management factors).
Poor Prison Conditions as a Governance Challenge
In our study, there is no simple, direct link between poor prison conditions and prisoners’ experiences of violence and theft. This may be due in part to the reality that for most incarcerated people, poor material conditions and overcrowding inside prison do not differ much from the poor material conditions of their home communities or from what marginalized communities know to expect during detention. In other words, the bar for what counts as a real sense of “deprivation” is higher for people who are accustomed to poverty and state neglect.
Another explanation might be that prisoners self-organize to fill the gaps. Prior research in Latin American prisons and, to some extent, in North American prisons suggests that illicit economies and informal prisoner-led governance arrangements emerge in situations where prisoners’ needs are not being met by the formal administration. For example, prisoner-led governance groups may organize collecting money for purchasing outside food supplies or providing mattresses and electricity (Darke & Garces, 2017). Prisoner-led groups (gangs or otherwise) also organize to facilitate more illicit transactions, such as drugs and weapons, in response to prisoners’ demands for these (Skarbek, 2014); this demand suggests a need for sedation/intoxication and for protection, to survive incarceration.
If prisons are very overcrowded and underfunded, it is more challenging for these groups and for staff to keep order and discipline, even if they are responding to some concrete needs. The groups that hold power inside may use violence more liberally to enforce rules but may also develop strategies to reduce or mitigate conflicts over scarce resources and harsh conditions. Similarly, in overcrowded cells, prisoners’ actions and infractions are constantly observed by other prisoners, which could increase enforcement by formal and informal authorities but could also generate incentives to find day-to-day nonviolent ways of handling the pressures of overcrowding.
Further research is necessary to explore how different prisoners, at the individual level, and different prisoner-led organizations, at the facility level, cope with the “pains of imprisonment” such as a lack of basic amenities and space, in the Latin American context where these conditions are widespread and normalized. Moreover, it is unclear whether the insignificant effects of prosocial activities that may be available—such as sports, school, or jobs—on violence (in this study) reflect inconsistent access to such activities across or within facilities. Although access to such activities may serve as a way to cope with harsh prison conditions at the individual level, these do not necessarily resolve prisoners’ basic needs for space, food, money, and protection. In Latin American prisons, even individuals with decent access to pro-social activities likely need to participate in informal prisoner-led governance arrangements in order to cover the conditions that the state does not provide.
Prior and Current Exposure to Criminal Activity and Experiences of Violence
Regardless of facility differences in crime or violence, this study shows that violence inside prisons is concentrated among certain individuals. The prisoners who report knowledge of crimes inside—that is, are close to the groups running criminal activity inside—also are more likely to report experiences of violence. It is difficult to determine whether people with a propensity for violence choose to—or are recruited to—join up with groups involved in crimes or whether those who start committing crimes inside prison then later experience violence inside. Importation theory suggests that they seek out what they are accustomed to, while strain theory suggests that those with fewest options for survival are more likely to use force. Most likely, these links go in both directions and reinforce one another. Further research should explore interactions between prior criminal activity/exposure, current involvement in in-prison criminality, and experiences of violence, at the individual and group levels.
What’s more, where there are more people in this high-crime high-violence group, there is a more generalized effect on prison violence: All inmates experience higher levels of violence and theft. This points to the possibility that this group has an influence on the general culture of prisoners, regardless of any illicit market or governance factors. It is worth highlighting that the prisoners who are in this “concentrated” group also reflect the individual characteristics that U.S. and European research find to be significant for involvement in violence during incarceration: age (young), sex (male), prior involvement in and exposure to criminal activity, and prior incarceration. This provides some support for importation theory’s relevance in Latin American prisons, even when facility-level and structural factors also matter. Theories that emphasize exposure to criminality as a predictor of experiencing or committing violence in prison (see, e.g., Hochstetler & DeLisi, 2005) align with our data—though in Latin America, the exposure also occurs inside prison for some people. The fact that crime or charge type (at the individual level and aggregated at the facility level for severe crimes) is not a significant predictor of experiences of violence also fits this explanation: Risk levels are less about a single illegal action and more about one’s peers and social experiences.
Considering the Potential Effects of Informal Governance Arrangements
Our findings show that facilities with more widespread criminal activity also have more prevalence of violence and theft experienced by prisoners. As we have outlined, since a direct application of deprivation and importation theories does not explain variations in experiences of violence, it is important to consider how prisoners might respond differently to the conditions and cultures of prison facilities in Latin America. Illicit and criminal activities undertaken by some prisoners might be both a respond to the conditions of the facility and an influence on the culture of the prison more generally. We argue that a plausible explanation for at least some of the link between criminality and violence inside prison might be that prisoner-led informal governance mechanisms influence when and how prisoners use violence. Where criminal activity is more widespread, the relative power of the tactics for facilitating these activities—including violence as a rule-enforcing tactic—is stronger than in prisons with lower levels of criminal activity.
Economic research shows that people use violence as a way to enforce rules and contracts for transactions that are illegal or illicit and cannot rely on formal laws; this dynamic is exacerbated in prison settings where buying and selling “regular” goods is often prohibited, along with goods such as drugs or weapons. Therefore, the prison gangs or groups responsible for these transactions deploy violence strategically and regularly (Skarbek, 2014). Research on how organized criminal groups that run illicit markets outside prison highlights the importance of keeping order and avoiding high-visibility incidents of violence, in order to facilitate uninterrupted transactions and to reduce attention from authorities (Durán-Martínez, 2018; Lessing, 2016). This can involve protection rackets, in which those in power use violence deliberately to reinforce their power and to impose order, and government officials may participate in these or turn a blind eye (Gundur, 2018). In the same vein, when prisoner-led governance groups no longer meet prisoners’ needs, prisoners may use violence to disrupt their rule; prisoners may also use coordinated violence to exert pressure on formal authorities (Lessing, 2016; Weegels, 2018).
However, a key insight from this broader research on criminal groups’ use of violence is that groups deliberately hide the violence they use or choose low-visibility tactics—such as burying bodies—to avoid detection (Durán-Martinez, 2018). Periodically, confrontations between groups or with law enforcement may generate episodes or incidents of violence. These dynamics likely occur within the prison setting as well, particularly in more permeable prisons in Latin America where such arrangements can involve actors on the outside as well as on the inside. Therefore, where criminal governance structures cover most of a prison’s population or social environment, violence—both visible and covert or hidden—is likely more widespread—even if it does not appear in official data or public reports. More detail on the frequency, severity, or visibility of violence against prisoners, as well as changes over time, would be necessary to understand how the violence patterns stem from protection racket dynamics.
A feature of protection markets and criminal governance is that, in many cases, when they consolidate control, violence rates often drop, due to reduced conflict and more compliance with the (illicit) rules (Arias, 2017; Lessing, 2016). So, in the prison setting, this might mean that where criminal activity is more sophisticated and organized, there is less violence and more security, though under a strict regime. This might explain the fact that prisoners report much lower levels of violence, particularly physical violence, in prisons in El Salvador and Brazil—where gangs are more dominant. Although some of the most visible and brutal cases of violence in prisons have occurred in these prisons—such as the 2017 riots in prisons in Northern Brazil (Darke, 2018) and the killing or decapitation of gang members in El Salvador (Insight Crime, 2017)—these are the countries where gangs have the most consolidated control inside prisons (Lessing, 2016; Rosen & Cruz, 2018). In contrast, governments in Chile and Argentina have imposed more formal authority and have challenged the rule of informal groups inside prison; this could generate more conflict and violence at least in the medium term. One study in Chile (Sanhueza, 2014) finds that the nature of interactions between staff and prisoners is an important and understudied factor shaping violence. Notably, there is much less variation across countries in prisoners’ reports of having their belongings stolen. This may reflect that prisoners steal from one another in similar ways, outside of the structures and rules of the organized groups inside.
This is a contrast to U.S. prison gang dynamics, where authorities have much stronger control over organizations, which reduces the gang members’ ability to communicate with each other and among the ranks (Skarbek, 2014). In order to understand these factors more concretely, further research could develop a typology for the model of governance inside facilities, ranging from completely consolidated illicit or prisoner governance (with distinctions for gang vs. nongang formats) to completely consolidated formal authorities, and combinations in between, and analyze the association with violence rates.
Limitations
Beyond the inherent limitations of using self-report survey data in a cross-sectional analysis across very different locations, several limitations in this study are worth noting. First, this study lacks the data to measure in a direct way the types and variation in governance models across facilities, including formal management strategies or types (state institutions and staff type) and informal governance dynamics, such as facilities that are gang-controlled and those that have a hybrid or combined governance arrangement, between prisoners and staff. For this reason, our discussion on governance as a potential explanation is speculative and requires further research to substantiate. Our study is also missing direct measures of other relevant facility-level factors that are important in other research: facility size, staff–inmate ratio, access to family, and community ties.
Second, the extent of criminal activity inside is determined by prisoners’ knowledge of crimes—which depends on their position within the facility and their willingness to share this information, even in general terms, with the survey team. Even then, this is not a measure of actual crime inside prison. There is slippage in our survey question between crimes committed in prison and crimes directed from prison toward the outside, and this distinction may matter. Other more direct measures of criminal activity inside—such as information developed through qualitative fieldwork in each facility—would strengthen this measure. Third, due to stigma and lack of research on sexual violence in prison (or in general) in the region, it is likely that respondents underreported their observations on this topic. Finally, we were not able to do a multilevel modeling analysis in this study, due to limitations with sampling at the facility level and the type of variable generated by the survey question. Therefore, our analysis covers facility-level differences and individual-level differences, but it does not integrate this analysis in a nested way. This limits the precision of our comparisons.
Implications and Conclusion
Overall, this study highlights the ways that the different material and social realities of prisons in Latin America shape patterns of violence in ways that are distinct from prisons in the United States and Europe. Prisoners experience deprivations and harsh conditions in varying ways, but these do not directly explain their experiences of violence. Similarly, prisoners’ prior exposure to and involvement in criminal activity is relevant to their roles in culture and economies inside prison (e.g., to their knowledge of criminal activity inside) but does not directly explain their likelihood of experiencing violence in prison.
This study suggests that looking at criminal activity inside prisons provides one avenue of explanation. But we argue that criminal activity must be understood in the context of Latin American prisons, not just through a strict definition or assuming clear distinctions between what is legal and illegal. Criminal activity may emerge in different forms depending on prisoners’ needs and demands and on the formal management style of a given facility—that is, it may be a way of securing basic amenities for prisoners (food, clothing) or it may be a way of generating profits or running larger criminal operations (e.g., drug trafficking or extortion). This study points to the importance of analyzing illegal activity inside prisons as a potential influence on prison violence—and perhaps on other aspects of prisoner experiences.
On first glance, it may appear that, given the findings of this study, an appropriate way to reduce violence among prisoners would be to target criminal activity carried out by groups of prisoners. Of course, both staff and prisoners have an interest in downplaying the connection between illicit activities by prisoners inside and violence patterns, in order to avoid punishment and to maintain profits. A potential unintended consequence of an institutional focus on violence reduction is that the more organized groups may choose to hide their violence more thoroughly. We argue that any intervention should be based on a clear analysis of why and how the criminal activity exists and why and how its participants may use violence against other prisoners. The answers to these questions are beyond the scope of this study and should be part of future research, particularly qualitative and ethnographic research. To the extent that violence occurs as part of managing criminal transactions in the prison context, whether tied to gangs or to other forms of cooperation, reducing this violence requires understanding the reasons behind it. The implication for policy makers is that violence-reduction efforts should seek to understand and engage strategically with prisoner-led governance logics to address key causes of violence, rather than relying solely on top-down infrastructure improvements, individual security classifications, disciplinary schemes, or separation of groups.
Footnotes
Appendix A
Percentage of Inmates Who Have Knowledge of Organization of Crimes Within the Prison, Disaggregated by Country, Prison, and Type of Crime.
| Country/Prison | N | Prison Conditions Index | % Accused of Homicide | % Accused of Kidnapping | % Know of Crimesa | Group of Level of Crimes | % Extortion | % Traffic | % Robbery | % Kidnapping | % Homicide |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Argentina | 456 | .536 | 11.8 | 1.9 | 19.1 | 5.2 | 3.1 | 6.2 | 7.0 | 2.7 | |
| 48 | 39 | .575 | 2.2 | 2.2 | 35.9 | High | 4.3 | 4.3 | 15.2 | 4.3 | 0.0 |
| 42 | 57 | .453 | 6.7 | 3.3 | 26.3 | Medium | 10.0 | 5.0 | 15.0 | 15.0 | 11.7 |
| 15 | 50 | .472 | 12.9 | 1.6 | 26.0 | Medium | 4.8 | 0.0 | 6.5 | 8.1 | 0.0 |
| 2 | 81 | .499 | 22.0 | 4.4 | 19.8 | Low | 6.6 | 2.2 | 6.6 | 6.6 | 3.3 |
| 30 | 98 | .442 | 16.5 | 2.6 | 19.4 | Low | 8.7 | 6.1 | 5.2 | 8.7 | 3.5 |
| 8 | 28 | .620 | 9.4 | 3.1 | 10.7 | Low | 0.0 | 3.1 | 0.0 | 0.0 | 0.0 |
| 33 | 40 | .574 | 7.1 | 0.0 | 7.5 | Very low | 0.0 | 2.4 | 0.0 | 4.8 | 0.0 |
| 9 | 29 | .612 | 20.6 | 0.0 | 6.9 | Very low | 0.0 | 0.0 | 0.0 | 2.9 | 0.0 |
| 51 | 34 | .577 | 8.8 | 0.0 | 5.9 | Very low | 0.0 | 0.0 | 0.0 | 2.9 | 0.0 |
| Brazil | 594 | .403 | 9.7 | 1.7 | 16.0 | 1.1 | 5.2 | 4.3 | 2.0 | 2.3 | |
| GUARULHOS 1 | 56 | .472 | 27.5 | 1.1 | 28.6 | Medium | 1.1 | 9.9 | 9.9 | 0.0 | 1.1 |
| GUARULHOS 2 | 105 | .359 | 6.5 | 3.3 | 17.1 | Low | 0.8 | 4.1 | 4.1 | 0.8 | 4.1 |
| FRANCO DA ROCHA | 128 | .299 | 4.0 | 1.1 | 16.4 | Low | 0.0 | 8.5 | 2.8 | 1.1 | 0.0 |
| HORTOLANDIA | 56 | .310 | 5.8 | 1.2 | 16.1 | Low | 1.2 | 2.3 | 5.8 | 1.2 | 1.2 |
| SOROCABA | 113 | .459 | 5.8 | 1.7 | 15.9 | Low | 3.3 | 4.1 | 2.5 | 5.8 | 6.6 |
| SANTA ANA | 90 | .538 | 7.8 | 2.9 | 11.1 | Low | 1.0 | 1.9 | 3.9 | 2.9 | 1.0 |
| PINHEIROS 1 | 20 | .260 | 14.3 | 0.0 | 10.0 | Low | 0.0 | 4.8 | 0.0 | 4.8 | 4.8 |
| PINHEIROS 2 | 21 | .299 | 16.0 | 4.0 | 4.8 | Very low | 0.0 | 0.0 | 4.0 | 0.0 | 0.0 |
| PINHEIROS 3 | 5 | .632 | 0.0 | 0.0 | 0.0 | Very low | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
| Chile | 662 | .464 | 5.7 | 1.0 | 36.7 | 3.9 | 18.1 | 10.0 | 3.2 | 1.6 | |
| CONCEPCION | 30 | .442 | 4.3 | 0.0 | 56.7 | Very high | 4.3 | 13.0 | 7.2 | 5.8 | 1.4 |
| CCP COLINA | 127 | .528 | 9.0 | 1.4 | 55.1 | Very high | 7.6 | 13.1 | 17.9 | 2.8 | 1.4 |
| CPFF STGO | 46 | .487 | 10.6 | 2.1 | 43.5 | Very high | 10.6 | 36.2 | 19.1 | 8.5 | 8.5 |
| CDF SANTIAGO | 39 | .463 | 9.8 | 0.0 | 35.9 | High | 7.3 | 34.1 | 9.8 | 9.8 | 2.4 |
| Valparaíso | 167 | .439 | 6.6 | 1.0 | 34.7 | High | 0.0 | 18.2 | 4.0 | 2.5 | 0.5 |
| CDP STGO. SUR | 151 | .466 | 5.5 | 1.1 | 30.5 | High | 2.7 | 20.8 | 14.2 | 1.6 | 2.2 |
| CDP PTE ALTO | 74 | .475 | 0.0 | 2.2 | 20.3 | Medium | 4.4 | 12.2 | 1.1 | 2.2 | 0.0 |
| Quillota | 28 | .414 | 0.0 | 0.0 | 10.7 | Low | 0.0 | 3.3 | 3.3 | 0.0 | 0.0 |
| El Salvador | 986 | .317 | 35.2 | 4.4 | 11.1 | 6.9 | 0.8 | 0.5 | 1.5 | 1.6 | |
| Centro de Readaptación para mujeres (Ilopango) | 105 | .345 | 5.9 | 7.4 | 26.7 | Medium | 18.4 | 1.5 | 2.2 | 1.5 | 3.7 |
| Centro Penitenciarío de San Miguel | 109 | .310 | 21.0 | 1.7 | 18.3 | Low | 10.9 | 0.0 | 0.8 | 4.2 | 3.4 |
| Centro Penitenciario de Apanteos | 251 | .388 | 18.2 | 7.4 | 15.5 | Low | 9.8 | 2.0 | 0.3 | 2.7 | 1.7 |
| Penitenciaría Oriental San Vicente | 93 | .311 | 28.2 | 3.9 | 7.5 | Very low | 4.9 | 0.0 | 0.0 | 1.0 | 1.0 |
| Centro Penitenciario de Sonsonate | 54 | .210 | 38.2 | 4.4 | 7.4 | Very low | 2.9 | 1.5 | 0.0 | 1.5 | 0.0 |
| Centro Penitenciario de de Izalco | 91 | .449 | 43.4 | 1.9 | 4.4 | Very low | 1.9 | 0.0 | 0.9 | 0.0 | 0.9 |
| Centro Penitenciario de Quezaltepeque | 105 | .306 | 33.9 | 2.5 | 3.8 | Very low | 1.7 | 0.0 | 0.0 | 0.0 | 0.8 |
| Centro de Cumplimiento de Penas Ciudad Barrios | 90 | .259 | 74.5 | 8.5 | 3.3 | Very low | 1.9 | 0.0 | 0.0 | 0.0 | 0.9 |
| Centro Penitenciario de de Chalatenango | 88 | .280 | 53.8 | 1.9 | 0.0 | Very low | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
| Peru | 1,009 | .488 | 5.2 | 2.0 | 33.5 | 21.1 | 4.1 | 6.6 | 10.1 | 4.1 | |
| E.P. TRUJILLO | 118 | .435 | 9.0 | 5.2 | 52.5 | Very high | 42.5 | 0.7 | 6.7 | 11.2 | 11.9 |
| E.P. PICSI | 85 | .404 | 6.3 | 1.0 | 49.4 | Very high | 40.6 | 2.1 | 1.0 | 19.8 | 1.0 |
| E.P.LURIGANCHO | 237 | .506 | 9.1 | 2.6 | 46.0 | Very high | 28.5 | 8.8 | 11.3 | 18.2 | 5.8 |
| E.P. SOCABAYA—AREQUIPA | 76 | .515 | 4.2 | 3.2 | 32.9 | High | 21.1 | 0.0 | 8.4 | 4.2 | 0.0 |
| E.P. SANTA MONICA | 47 | .469 | 2.0 | 4.0 | 31.9 | High | 14.0 | 6.0 | 8.0 | 12.0 | 4.0 |
| E.P. HUANCAYO—JUNIN | 79 | .471 | 9.9 | 2.2 | 31.6 | High | 16.5 | 3.3 | 11.0 | 7.7 | 2.2 |
| E.P. MIGUEL CASTRO CASTRO | 40 | .444 | 4.5 | 6.8 | 25.0 | Medium | 20.5 | 2.3 | 4.5 | 13.6 | 4.5 |
| E.P. VARONES | 4 | .579 | 0.0 | 0.0 | 25.0 | Medium | 16.7 | 0.0 | 0.0 | 0.0 | 0.0 |
| E.P. CALLAO | 51 | .502 | 5.3 | 0.0 | 21.6 | Medium | 12.3 | 8.8 | 5.3 | 7.0 | 5.3 |
| E.P. SANANGUILLO—TARAPOTO | 39 | .558 | 4.3 | 0.0 | 17.9 | Low | 8.5 | 2.1 | 4.3 | 2.1 | 0.0 |
| E.P. YANAMILLA/AYACUCHO | 63 | .441 | 6.7 | 1.1 | 15.9 | Low | 6.7 | 1.1 | 2.2 | 1.1 | 2.2 |
| E.P. SAN JORGE LIMA | 20 | .589 | 0.0 | 0.0 | 15.0 | Low | 11.5 | 0.0 | 0.0 | 0.0 | 0.0 |
| E.P. QUENCORO | 97 | .469 | 8.3 | 1.5 | 12.4 | Low | 3.0 | 4.5 | 3.8 | 3.8 | 3.0 |
| E.P. ANCON II | 53 | .449 | 3.2 | 0.0 | 11.3 | Low | 6.3 | 4.8 | 4.8 | 6.3 | 3.2 |
| Total | 3,707 | .446 | 12.9 | 2.2 | 23.5 | 9.0 | 5.8 | 5.2 | 4.9 | 2.5 |
aThis is the percentage of people who report knowledge of any crimes being committed in or directed from prison. The specific types of crimes in the other columns is not a comprehensive list; these are just the most commonly reported types. So the total percentage is higher than the sum of these types.
Appendix B
Summary of Descriptive Statistics.
| N | Argentina 516 | Brazil 751 | Chile 805 | El Salvador 1,160 | Peru 1,205 |
|---|---|---|---|---|---|
| Violence inside the prison (% who answered yes) | |||||
| Beaten in the last 6 months | 18.2 | 4.5 | 26.1 | 3.5 | 14.4 |
| Theft of personal items | 31.8 | 30.1 | 38.3 | 31.8 | 47.0 |
| Forced to sexual intercourse | 2.3 | 0.3 | 1.3 | 0.5 | 0.9 |
| Prison conditions indexa (deprivation variable, lower score means harsher conditions) | |||||
| Mean | 0.509 | 0.385 | 0.468 | 0.333 | 0.372 |
| Standard deviation | 0.175 | 0.175 | 0.142 | 0.136 | 0.146 |
| Q1 | 0.389 | 0.235 | 0.368 | 0.222 | 0.267 |
| Q2 (median) | 0.500 | 0.368 | 0.474 | 0.333 | 0.375 |
| Q3 | 0.618 | 0.526 | 0.579 | 0.421 | 0.465 |
| Prison conditions (% who answered yes) | |||||
| Has access to a television | 39.7 | 43.7 | 53.4 | 48.4 | 43.3 |
| Has enough water to drink | 40.7 | 30.9 | 56.0 | 35.3 | 39.5 |
| Toilets are clean | 43.4 | 39.4 | 38.1 | 29.6 | 42.2 |
| Has access to books | 45.0 | 19.7 | 42.2 | 28.7 | 41.2 |
| The institution provides medical attention | 25.8 | 21.6 | 47.7 | 24.7 | 41.2 |
| Family members receive good treatment | 61.2 | 26.5 | 23.0 | 35.1 | 29.5 |
| Has access to newspapers | 36.0 | 18.5 | 40.5 | 24.3 | 42.3 |
| Has access to magazines | 37.8 | 34.0 | 38.1 | 15.9 | 34.1 |
| Has access to radio | 39.3 | 30.6 | 45.8 | 12.8 | 32.0 |
| The institution provides medications | 21.3 | 21.3 | 42.6 | 12.5 | 28.5 |
| Has access to public telephone | 50.2 | 0.4 | 17.8 | 18.3 | 47.6 |
| Food quantity is enough | 12.0 | 19.8 | 31.4 | 8.3 | 23.5 |
| Medical care is good | 12.2 | 6.9 | 16.3 | 5.9 | 10.7 |
| Food quality is good | 11.0 | 6.1 | 15.5 | 3.4 | 8.5 |
| Has access to cell phone | 7.0 | 1.5 | 14.2 | 6.4 | 3.5 |
| The institution provides toilet paper | 6.4 | 18.1 | 1.0 | 4.4 | 0.7 |
| The institution provides toothpaste | 7.6 | 12.9 | 0.7 | 4.1 | 0.6 |
| The institution provides soap | 7.2 | 13.7 | 0.7 | 4.4 | 0.6 |
| The institution provides toothbrush | 7.0 | 8.3 | 0.9 | 3.5 | 0.7 |
| Participation in activities | |||||
| Proportion that participates in sport activities | 33.7 | 22.2 | 28.4 | 31.7 | 36.8 |
| Proportion that participates in academic activities | 29.7 | 9.6 | 30.3 | 26.9 | 20.0 |
| Proportion that participates in cleaning activities | 22.3 | 17.0 | 31.3 | 30.6 | 34.9 |
| Proportion that has a job | 29.3 | 24.9 | 34.2 | 24.2 | 36.2 |
| Overpopulation (% who answered no; deprivation variable) | |||||
| Each inmate has a bed | 9.3 | 89.4 | 24.6 | 92.9 | 52.6 |
| Individual factors (importation factors) | |||||
| Proportion of men | 79.1 | 86.3 | 86.1 | 81.9 | 87.0 |
| Proportion that receives a visit at least once a week | 20.7 | 21.6 | 52.2 | 15.6 | 32.7 |
| Proportion that calls family least once a week | 91.3 | 3.9 | 40.1 | 25.5 | 74.9 |
| Proportion that consumed drugs or alcohol recently | 9.9 | 13.2 | 20.7 | 1.6 | 15.1 |
| Proportion that was in a juvenile detention institution | 19.4 | 18.8 | 40.2 | 11.9 | 6.9 |
| Proportion that had contact with gangs during infancy | 68.4 | 46.2 | 63.9 | 46.0 | 40.4 |
| Proportion that robbed or sold drugs when was a minor | 44.4 | 35.9 | 59.0 | 12.6 | 15.8 |
| Respondent age | |||||
| Proportion on inmates of 35 years or less | 61.2 | 65.8 | 60.8 | 68.5 | 45.1 |
| Average | 34.6 | 33.5 | 35.1 | 33.5 | 38.9 |
| Standard deviation | 10.3 | 10.6 | 11.2 | 10.3 | 11.0 |
| Years attended school | |||||
| Average | 15.5 | 17.5 | 15.6 | 15.7 | 16.6 |
| Standard deviation | 3.9 | 5.5 | 5.1 | 5.2 | 5.4 |
| Years of imprisonment (sentence) | |||||
| Average | 9.3 | 10.8 | 7.9 | 18.5 | 12.3 |
| Standard deviation | 11.9 | 11.6 | 9.4 | 17.9 | 8.1 |
| Proportion of inmates accused of each crime | |||||
| Intentional homicide | 13.2 | 8.9 | 5.8 | 31.4 | 3.3 |
| Manslaughter | 7.6 | 1.2 | 2.0 | 4.8 | 1.0 |
| Kidnapping | 1.9 | 1.9 | 0.7 | 4.0 | 0.7 |
| Assault | 1.2 | 1.3 | 1.5 | 1.0 | 6.5 |
| Sexual crimes | 7.9 | 14.1 | 7.2 | 10.9 | 2.9 |
| Theft | 50.8 | 30.7 | 55.9 | 11.7 | 48.0 |
| Drug trafficking | 14.0 | 32.8 | 19.6 | 9.4 | 14.1 |
| Illegal arms possession | 1.0 | 0.4 | 1.0 | 1.9 | 2.3 |
| Aggravated theft | 0.2 | 6.0 | 2.4 | 1.1 | 16.0 |
| Extortion | 0.2 | 0.3 | 0.0 | 19.3 | 0.3 |
| Obstruction of justice | 0.1 | 0.0 | 0.0 | n.d. | 0.0 |
| Other crimes | 2.0 | 2.4 | 3.9 | 4.5 | 4.9 |
aThe index represents the proportion of the 19 services described in “prison conditions” to which the inmates have access (if inmates have access to all 19 services, their index is 1; if they have access to none of the services, their index is 0). The Kuder–Richardson (KR20) value to assess the internal consistency of the 19 items was 0.646 (moderate reliability).
Appendix C
Comparing Prisoners With and Without Knowledge of Crime Inside by Individual Characteristics.
| Survey Question | Response Group | Knowledge of Crimes Being Organized (%) | Z Value (One Tail) |
|---|---|---|---|
| Childhood | |||
| Your best friends committed crimes | Yes (N = 1,780) | 29.5 | 8.240 (.000) |
| No (N = 1,861) | 17.9 | ||
| Your friends at school committed crimes | Yes (N = 903) | 33.2 | 7.372 (.000) |
| No (N = 2,491) | 20.9 | ||
| In the neighborhood where you lived there were youth gangs and criminal bands | Yes (N = 1,844) | 28.1 | 6.441 (.000) |
| No (N = 1,817) | 19.1 | ||
| Was in a juvenile detention center | Yes (N = 641) | 34.3 | 5.423 (.000) |
| No (N = 3,035) | 22.9 | ||
| Delinquent life | |||
| Ever worked? | Yes (N = 3,375) | 22.9 | 2.532 (.005) |
| No (N = 325) | 29.2 | ||
| Was in the army or police | Yes (N = 466) | 29.1 | 3.094 (.001) |
| No (N = 3,214) | 22.6 | ||
| Committed robbery (of people) or sold drugs before age 18 | Yes (N = 1,061) | 33.8 | 9.402 (.000) |
| No (N = 2,632) | 19.3 | ||
| Committed the abovementioned crimes with other people | Yes (N = 961) | 37.4 | 4.541 (.000) |
| No (N = 394) | 24.6 | ||
| Sold drugs without being arrested | Yes (N = 269) | 32.3 | 4.868 (.000) |
| No (N = 445) | 16.6 | ||
| Previously convicted of another crime | Yes (N = 1,155) | 29.9 | 6.329 (.000) |
| No (N = 2,516) | 20.4 | ||
| The crime for which you are convicted was ordered by a criminal group | Yes (N = 175) | 30.2 | 3.635 (.000) |
| No (N = 2,436) | 18.9 | ||
| Did anyone else participate in the crime you are charged with? | Yes (N = 1,964) | 26.4 | 4.483 (.000) |
| No (N = 1,691) | 20.1 | ||
| Preincarceration substance use | |||
| How frequently did you use marijuana in the 6 months prior to arrest? | Every day (N = 663) | 27.9 | 1.949 (.025) |
| Didn’t consume (N = 400) | 22.5 | ||
| How frequently did you use crack/cocaine in the 6 months prior to arrest? | Every day (N = 327) | 33.0 | 0.997 (.159) |
| Didn’t consume (N = 267) | 31.8 | ||
| Did you ever consume inhalable drugs? | Yes (N = 468) | 32.4 | 4.915 (.000) |
| No (N = 3,159) | 22.1 | ||
| Substance use during incarceration | |||
| In the last month, have you consumed alcohol or drugs? | Yes (N = 424) | 41.7 | 9.466 (.000) |
| No (N = 3,255) | 21.0 | ||
| In the last month, you have consumed alcohol | Yes (N = 182) | 42.3 | 7.297 (.000) |
| No (N = 2,637) | 19.5 | ||
| In the last month, you have consumed marijuana | Yes (N = 253) | 39.5 | 8.030 (.000) |
| No (N = 2,483) | 18.2 | ||
| In the last month, you have consumed cocaine | Yes (N = 72) | 51.3 | 6.607 (.000) |
| No (N = 2,666) | 19.5 | ||
| You have observed other inmates consuming alcohol or drugs | Yes (N = 2,426) | 28.9 | 11.226 (.000) |
| No (N = 1,158) | 11.9 | ||
| Collusion with authorities | |||
| In the crime you are charged with, an authority/official participated as an accomplice | Yes (N = 122) | 36.8 | 2.874 (.002) |
| No (N = 1,890) | 25.1 | ||
| What role did the authority/official have in the crime? | Organizer (N = 44) | 38.6 | 0.432 (.332) |
| Not investigating/giving info (N = 58) | 34.4 | ||
| Possession and use of weapons | |||
| Carried a weapon at the time of arrest | No (N = 2,423) | 20.6 | 4.774 (.000) |
| Firearm (N = 970) | 28.2 | ||
| Have ever held a firearm in your hands | Yes (N = 2,082) | 28.2 | 7.775 (.000) |
| No (N = 1,607) | 17.3 | ||
| Have ever held a long gun | Yes (N = 904) | 33.4 | 8.058 (.000) |
| No (N = 2,803) | 20.3 | ||
| Have ever held a handgun | Yes (N = 1,614) | 26.9 | 4.322 (.000) |
| No (N = 2,093) | 20.8 | ||
| Have tried to shoot another person | Yes (N = 798) | 35.4 | 5.821 (.000) |
| No (N = 1,258) | 23.6 | ||
| In the 6 months before arrest, had bought a firearm | Yes (N = 307) | 33.2 | 2.088 (.018) |
| No (N = 1,737) | 27.4 | ||
| Have injured or killed another person | Yes (N = 751) | 34.7 | 8.163 (.000) |
| No (N = 2,833) | 20.5 | ||
| Time in prison | |||
| Years inside the prison | Less than 10 (N = 3,363) | 22.5 | 4.629 (.000) |
| More than 10 (N = 320) | 34.0 | ||
Note. Yes = knowledge of crimes inside; no = no knowledge of crimes inside (with significance test in parenthesis).
Appendix D
| Physical Violence | Proprietary Victimization | Sexual Violence | |||||||
|---|---|---|---|---|---|---|---|---|---|
| B | SE B | eB | B | SE B | eB | B | SE B | eB | |
| Has knowledge of crimes | 0.494*** | 0.128 | 1.64 | 0.677*** | 0.133 | 1.97 | 0.925*** | 0.148 | 2.52 |
| Level of organization of crimes (base = very low) | |||||||||
| Low | 0.907* | 0.425 | 2.48 | 0.832* | 0.326 | 2.30 | 0.601 | 0.446 | 1.82 |
| Medium | 0.855 | 0.484 | 2.35 | 0.967* | 0.385 | 2.63 | 0.048 | 0.533 | 1.05 |
| High | 1.216* | 0.501 | 3.37 | 1.254*** | 0.387 | 3.51 | 0.613 | 0.505 | 1.85 |
| Very high | 1.486** | 0.516 | 4.42 | 1.358*** | 0.386 | 3.89 | 0.956 | 0.493 | 2.60 |
| Personal characteristics | |||||||||
| Man | −0.043 | 0.249 | .96 | −1.02*** | 0.168 | 0.36 | −0.011 | 0.282 | 0.99 |
| Age up to 35 years old | 0.336* | 0.134 | 1.40 | −.066 | 0.089 | 0.94 | −0.501*** | 0.155 | 0.61 |
| Study level (base = less than primary) | |||||||||
| Primary school | −0.320 | 0.318 | 0.73 | 0.014 | 0.185 | 1.01 | −0.241 | 0.365 | 0.79 |
| Secondary | −0.431 | 0.334 | 0.65 | 0.081 | 0.194 | 1.08 | −0.067 | 0.379 | 0.94 |
| Technical | −0.418 | 0.424 | 0.66 | 0.441 | 0.237 | 1.55 | 0.250 | 0.459 | 1.28 |
| Undergraduate or graduate | −0.400 | 0.414 | 0.67 | 0.575* | 0.245 | 1.78 | 0.077 | 0.451 | 1.08 |
| History before current sentence | |||||||||
| Had friends who committed crimes | 0.245 | 0.145 | 1.28 | 0.172 | 0.094 | 1.19 | 0.408* | 0.171 | 1.50 |
| Was in a minors’ institution | 0.277 | 0.144 | 1.32 | −0.052 | 0.114 | 0.95 | 0.090 | 0.182 | 1.09 |
| Was imprisoned before | 0.260 | 0.142 | 1.30 | 0.062 | 0.103 | 1.06 | 0.094 | 0.170 | 1.10 |
| Had used a fire weapon | 0.170 | 0.150 | 1.19 | 0.122 | 0.096 | 1.13 | 0.517** | 0.177 | 1.68 |
| Used to consume marihuana | 0.307 | 0.162 | 1.36 | 0.293** | 0.107 | 1.34 | −0.126 | 0.187 | 0.88 |
| Used to consume cocaine | −0.099 | 0.144 | 0.91 | 0.073 | 0.105 | 1.08 | 0.073 | 0.174 | 1.08 |
| Used to consume ecstasy | 0.382* | 0.155 | 1.46 | 0.216 | 0.128 | 1.24 | 0.097 | 0.194 | 1.10 |
| Used to consume heroin | 0.583 | 0.332 | 1.79 | −0.317 | 0.301 | 0.73 | 0.469 | 0.381 | 1.60 |
| Sentence | |||||||||
| Sentence for intentional homicide | 0.492 | 0.252 | 1.64 | 0.045 | 0.167 | 1.05 | 0.969*** | 0.277 | 2.64 |
| Sentence for manslaughter | 0.728* | 0.331 | 2.07 | 0.404 | 0.237 | 1.50 | 0.369 | 0.438 | 1.45 |
| Sentence for kidnapping | 0.712 | 0.365 | 2.04 | 0.482 | 0.253 | 1.62 | 1.003* | 0.400 | 2.73 |
| Sentence for assault | 0.646 | 0.435 | 1.91 | 0.199 | 0.308 | 1.22 | 1.256** | 0.451 | 3.51 |
| Sentence for sexual crimes | 0.548* | 0.272 | 1.73 | 0.514** | 0.173 | 1.67 | 0.962*** | 0.292 | 2.62 |
| Sentence for aggravated robbery | 0.429 | 0.252 | 1.54 | −0.121 | 0.175 | 0.89 | 0.991*** | 0.273 | 2.69 |
| Sentence for crimes against health | 0.143 | 0.246 | 1.15 | −0.096 | 0.157 | 0.91 | 0.703** | 0.274 | 2.02 |
| Sentence for illegal arms possession | 0.092 | 0.266 | 1.10 | −0.050 | 0.191 | 0.95 | −0.534 | 0.358 | 0.59 |
| Sentence for simple robbery | 0.537* | 0.222 | 1.71 | −0.090 | 0.146 | 0.91 | 1.070*** | 0.241 | 2.92 |
| Sentence for fraud, embezzlement | 0.778 | 0.606 | 2.18 | 1.156* | 0.489 | 3.18 | −0.410 | 1.055 | 0.66 |
| Sentence for extorsion | 0.322 | 0.409 | 1.38 | −0.453* | 0.225 | 0.64 | 1.197** | 0.447 | 3.31 |
| Prison conditions and security level | |||||||||
| Prison conditions index | −0.226 | 1.761 | 0.80 | −0.994 | 1.350 | 0.37 | 1.212 | 1.794 | 3.36 |
| Overpopulation in cell | 0.031 | 0.533 | 1.03 | 0.432 | 0.323 | 1.54 | −0.609 | 0.391 | 0.54 |
| % of accused of homicide | −0.229 | 1.345 | 0.80 | −0.681 | 1.014 | 0.51 | −1.527 | 1.526 | 0.22 |
| % of accused of kidnapping | −6.751 | 6.146 | 0.00 | 1.201 | 4.920 | 3.32 | 0.948 | 6.289 | 2.58 |
| Community activities | |||||||||
| Participates in sport activities | 0.058 | 0.152 | 1.06 | −0.071 | 0.105 | 0.93 | −0.159 | 0.185 | 0.85 |
| Participates in academic activities | 0.052 | 0.167 | 1.05 | 0.248* | 0.114 | 1.28 | 0.197 | 0.200 | 1.22 |
| Has a job | −0.098 | 0.156 | 0.91 | −0.073 | 0.105 | 0.93 | −0.040 | 0.186 | 0.96 |
| Substances consumption during last month | |||||||||
| Alcohol | 0.588** | 0.227 | 1.80 | 0.230 | 0.195 | 1.26 | 0.027 | 0.279 | 1.03 |
| Marihuana | 0.634*** | 0.181 | 1.89 | 0.374* | 0.160 | 1.45 | 0.278 | 0.229 | 1.32 |
| Cocaine | 0.866** | 0.317 | 2.38 | −0.040 | 0.282 | 0.96 | 1.365*** | 0.326 | 3.92 |
| Ecstasy | −0.829 | 0.607 | 0.44 | −0.052 | 0.503 | 0.95 | −0.555 | 0.813 | 0.57 |
| Constant | −4.08*** | 1.145 | 0.02 | −1.565 | 1.190 | 0.21 | −4.75*** | 1.188 | 0.01 |
| Wald χ2 | 216.72 | 187.81 | 178.68 | ||||||
| df | 41 | 41 | 41 | ||||||
| N | 3,419 | 3,426 | 3,411 | ||||||
*p < .05. **p < .01. ***p < .001.
Acknowledgments
We would like to thank the anonymous reviewers and the journal editors, as well as Jeff Mellow, for their helpful comments on this article. We would also like to thank the research teams who helped with collecting the survey data, the government partners who facilitated access to prisons, the United Nations Development Program for its project support, and Centro de Investigación y Docencia Económicas (CIDE) and the Centro de Estudios Latinoamericanos sobre Inseguridad y Violencia (CELIV), Universidad Tres de Febrero.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The collection of data used in this article was funded in part by the United Nations Development Program and in part by the Centro de Investigación y Docencia Económicas (CIDE).
