J.24 Evaluating social protective factors for violence.pdf

Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=cdip20

Development in Practice

ISSN: 0961-4524 (Print) 1364-9213 (Online) Journal homepage: https://www.tandfonline.com/loi/cdip20

Evaluating social protective factors for violence involvement in Honduras

Tom Hare, Laura E. Miller-Graff & Juan Carlos Guzman

To cite this article: Tom Hare, Laura E. Miller-Graff & Juan Carlos Guzman (2019): Evaluating social protective factors for violence involvement in Honduras, Development in Practice, DOI:

10.1080/09614524.2019.1659230

To link to this article: https://doi.org/10.1080/09614524.2019.1659230

Published online: 06 Sep 2019.

Submit your article to this journal

Article views: 16

View related articles

View Crossmark data https://www.tandfonline.com/action/journalInformation?journalCode=cdip20 https://www.tandfonline.com/loi/cdip20 https://www.tandfonline.com/action/showCitFormats?doi=10.1080/09614524.2019.1659230 https://doi.org/10.1080/09614524.2019.1659230 https://www.tandfonline.com/action/authorSubmission?journalCode=cdip20&show=instructions https://www.tandfonline.com/action/authorSubmission?journalCode=cdip20&show=instructions https://www.tandfonline.com/doi/mlt/10.1080/09614524.2019.1659230 https://www.tandfonline.com/doi/mlt/10.1080/09614524.2019.1659230 http://crossmark.crossref.org/dialog/?doi=10.1080/09614524.2019.1659230&domain=pdf&date_stamp=2019-09-06 http://crossmark.crossref.org/dialog/?doi=10.1080/09614524.2019.1659230&domain=pdf&date_stamp=2019-09-06

Evaluating social protective factors for violence involvement in Honduras Tom Hare, Laura E. Miller-Graff and Juan Carlos Guzman

ABSTRACT

Endemic levels of violence continue to stall social and economic development in Honduras. Significant emphasis has been placed on social protective factors such as resilience and cohesion in violence prevention policies and programmes. This article evaluates these factors using previously validated tools applied to over 1,200 non-incarcerated and incarcerated individuals. Results demonstrate that resilience and cohesion are not as strongly associated with violence involvement in these samples as hypothesised, while early behaviour problems have a greater association with current violence involvement. These results suggest the need for further investigation into critical violence prevention hypotheses.

ARTICLE HISTORY

Received 4 September 2018 Accepted 25 April 2019

KEYWORDS

Conflict and reconstruction;

Governance and public policy; Latin America and the Caribbean

Introduction

Violence negatively impacts economic, social, and human development in the Northern Triangle countries of Central America (Adams 2017; Moser and McIlwane 2006; UNDP 2009). International aid to Honduras, El Salvador, and Guatemala is focused largely on reducing violence given that these countries perennially rank among the most violent in the world, with homicide rates at three to four times the Latin American average (UNODC 2015).

To combat this development constraint, outside investment in security programmes reached US $1.7 billion between 2009–2011 (Inter-American Development Bank 2011), and an additional US$1 billion was earmarked by the United States for increased security in the region in 2015. The Alliance for Prosperity Plan 2016–2020 committed regional governments to improving security in the region as a top budget priority. All of this funding has supported both traditional law enforcement approaches to security, as well as social programmes that aim to prevent violence. Social pro-grammes for violence prevention include activities such as those to increase employment, improve parks, increase accountability for local officials, and efforts to reduce corruption and increase respect for human rights. Despite these investments, chronic violence continues to adversely impact individuals, communities, democratic governance, and economic growth in Central America (UNAH-

IUDPAS 2018).

In Honduras, violence is most often perceived to be associated with drug trafficking groups and criminal gangs. While drug trafficking groups promote insecurity and generalised fear, gangs “perme-ate… society to the degree that some believe they pose a threat to national security by undermining the ability of these states to govern, and to control their borders” (Bruneau and Dammert 2011, 211). Gangs recruit vulnerable youth and young adults in their communities to perpetuate violence. Any discus-sion of violence and development in Honduras must include the reality of gang influence on individ-uals, relationships, and communities, given that gang members in Central America are more likely than their non-gang involved peers to engage in violent behaviour (Olate, Salas-Wright, and

© 2019 Informa UK Limited, trading as Taylor & Francis Group

CONTACT Tom Hare thare@nd.edu

DEVELOPMENT IN PRACTICE

https://doi.org/10.1080/09614524.2019.1659230 http://crossmark.crossref.org/dialog/?doi=10.1080/09614524.2019.1659230&domain=pdf&date_stamp=2019-09-06 mailto:thare@nd.edu http://www.tandfonline.com

Vaughn 2012). Although much violence in Honduras is related to the social influence of gangs, there is substantial benefit in examining violence more broadly. An examination of any type of violence in isolation is unlikely to capture the full range of risk and resilience evident in a given context (Hamby and Grych 2012). This study therefore acknowledges the relationship between gangs and violence, but does not focus exclusively on gang violence. Rather, we focus on violence involvement (victimi-sation, non-violent and violent criminal behaviour) more generally.

To date, effective, sustained violence prevention in the Honduran context has been elusive.

Although the homicide rate in Honduras declined in 2017 and 2018, no reliable way to slow the enlist-ing of young people into gangs has been found (Bruneau and Dammert 2011) and the repression of violence through heavy-handed policing has failed. There has been some success, however, in iden-tifying risk and protective factors for gang membership (Hennigan et al. 2014). Concerted effort in violence prevention, known as “citizen security”, has also shown some promise in democratising traditional public security approaches. Citizen security:

“expand[s] the focus of public security measures from a narrow preoccupation with safeguarding state insti-tutions to encompass a wider concern with preserving the safety and rights of citizens… [including] violence pre-vention activities focused on at-risk families and youth”. (Muggah 2017, 291)

Specifically, progress has been made to identify and target high-risk areas through hot-spot mapping and crime observatories. “Place-based” approaches are focusing on these high-risk areas at the municipal and neighbourhood level to provide a mix of police, social programmes, and community interventions simultaneously. Despite these advances, methods to identify and target individuals within those communities who might lack protective factors, such as a strong family unit or neigh-bourhood ties, are less developed. Further, little is known about what protective factors for individ-uals are most critical in this context. Given the large number of possible points of prevention and intervention, research evaluating the relative contribution of protective factors across social-ecologi-cal levels stands to make an important contribution to effective prevention in this context.

The identification of protective factors in the citizen security prevention approach takes into account the concept of “resilience” or:

“the capacity of individuals to navigate their way to health-sustaining resources, including opportunities to experience feelings of well-being, and a condition of the individual’s family, community and culture to provide these health resources and experiences in culturally meaningful ways”. (Ungar 2008, 225)

Resilience, in Ungar’s conceptualisation, does not therefore simply represent an individual’s ability to respond and adapt to adverse circumstances, but places individual functioning within a context-specific framework, where outcomes are necessarily understood as the product of both individual factors and context-specific resources. Advocates of citizen security and resilience therefore seek to better understand both individual and context-specific protective factors as they relate to preven-tion mechanisms in violent contexts such as Honduras (Grych, Hamby, and Banyard 2015; Ttofi et al. 2016). By providing empirical evidence on context-specific protective factors, we aim to con-tribute to that discussion. By testing components of an integrated framework for human develop-ment and violence prevention, we also aim to inform the academic and policy discussions on violence as a:

“development constraint with no blueprint solutions… [and a] divide between academic research that empha-sises the complexity of violence, and policy-focused analysis that seek to categorise types of violence, measure its costs, and identify its consequences in terms of homicide and victimization rates”. (Moser and McIlwane 2006, 100)

Social protective factors

The most widely used framework for human development and violence prevention is the social-eco-logical model. Based on human development and public health approaches, the social-ecological model examines nested domains of influence – individual, relationship, community, and societal –

2 T. HARE ET AL.

and their interactions (Adams 2017; Moser and McIlwane 2006; WHO 2002). This approach recognises that “no single factor explains why some individuals behave violently toward others”, and provides a fra-mework by which to assess the multidimensional and interrelated factors that contribute to violence involvement (Moser and McIlwane 2006, 95). As an epidemiological approach, the social-ecological model is used to inform programmes and policies that seek to modify an individual’s behaviour or perception, and influence the social context that impacts those behaviours and perceptions. The model is particularly useful when used to identify both risk and protective factors.

Sharing the approach used by Farrington and others (e.g. Grych, Hamby, and Banyard 2015) we conceptualise risk factors as “predictors of persistent serious offending”, while protective factors are understood as mitigating factors that reduce the risk of offending (Stouthamer-Loeber et al. 2002, 111–112; Ttofi et al. 2016). As opposed to a continuum where risk and protective factors are at oppo-site ends, protective factors reduce risk. The inclusion and evaluation of multiple possible protective effects is especially important among social factors in areas of high violence where exposure to certain risk factors is nearly universal. In these instances, it is important to distinguish the factors that mitigate risk versus identifying what might lay at the positive end of the continuum alone.

For example, the presence of a gang in a neighbourhood is true for all residents in that neighbour-hood, and presents an elevated risk of both victimisation and violence through the recruitment of young adults. The non-presence of the gang would lie at the positive end of the continuum.

However, factors that reduce the risk of victimisation or participating in the gang’s violent acts (sense of security/belonging with family and friends) are potential protective factors that warrant examination. Our use of protective factors is purposeful given that we seek to understand the associ-ation between certain elements of the social domain and violence involvement.

While most studies on the region focus on violent offending alone, we include victimisation in our definition of violence involvement to determine the association between social protective factors and risk of victimisation. We envision this broad understanding of violence involvement as important given that victimisation and perpetration are linked behaviours, especially when con-sidering a broad understanding of violence. That is, individuals may both experience and perpetrate violence in their lifetime, with violence victimisation in some contexts predicting violence per-petration in others (e.g. community violence victimisation has been shown to predict both dating violence victimisation and perpetration in the US; Malik, Sorenson, and Aneshensel 1997).

As such, exploring how social protective factors are associated with both victimisation and per-petration of violence more accurately reflects the complexity of violence involvement and conse-quently provides more comprehensive insight into the social factors most likely to exert positive impacts on individuals. While there is a substantial literature that explores risk and protective effects, individual factors (Ttofi et al. 2016), and individual offending in the United States and a growing literature on Central America (Olate, Salas-Wright, and Vaughn 2012), few studies have included social factors, and fewer still have included victimisation in Central America. Meanwhile, social factors are consistently used in citizen security programmes in Honduras and throughout Central America to reduce violence involvement (Cano and Rojido 2016). These programmes work at the relationship and community levels of the social-ecological framework to promote social control, social cohesion, and trust at the neighbourhood level. Despite their prevalence, rela-tively little is known about the impact of community interventions on violence involvement (Hare 2018). Where evidence is available, results are mixed on whether these programmes increase trust and cohesion, or only result in increased participation and other programmatic outcomes versus impact (Hansen-Nord et al. 2016).

Given the need to empirically understand the association between violence involvement and social conditions in high violence areas to improve violence reduction programmes and policies, we explore here the social protective factors associated with violence involvement (victimisation, non-violent and violent offences) among young adults in Honduras. We hypothesise a significant association between social protective factors (i.e. resilience and cohesion) and violence involvement.

This hypothesis is congruent with the social-ecological model where relationship and community

DEVELOPMENT IN PRACTICE 3

level factors are associated with violence involvement. The hypothesis also aligns with social disor-ganisation and collective efficacy theories, which posit that strong cohesion, trust, and shared values can mitigate violence (Sampson, Raudenbush, and Earls 1997).

Methods

To test this hypothesis, a convenience sample of 936 young adults were interviewed from nine high-crime neighbourhoods in and around Tegucigalpa and 277 incarcerated individuals were interviewed in two prisons (one holding males, the other holding females) located near Tegucigalpa. The age group of both samples (16–34 years old) mirrors the most violence-involved age range for Honduras (UNAH-IUDPAS 2018). For security reasons, our local coordinator obtained permission before entering any neighbourhood, and open or public spaces such as community centres and schools were chosen.

We then asked community leaders and local organisations to invite young adults to come to com-plete the interview. For the incarcerated population, prison management selected cellblocks, and vol-unteers participated from those blocks. Surveys were administered in common spaces on prison grounds in the presence of guards. In all locations, surveys were conducted as far from others as poss-ible to ensure privacy and confidentiality, however complete privacy was not always possible given space constraints and prison regulations.

Local enumerators administered the surveys in Spanish on tablets for the non-incarcerated popu-lation, and on paper for the incarcerated population given restrictions on electronic devices inside the jails. Survey supervisors participated in three days of training and enumerators participated in two days of training. All supervisors and enumerators had previous experience and were contracted through a local survey firm. The training covered the background of the project and ensured under-standing of the questions and tablets among the enumerators. The survey firm and the researchers provided supervision during data collection in communities and prisons.

Measures

In addition to questions on violence involvement, we validated two widely-used tools that examine components of the relationship and community domains in Honduras (Hare, Guzman, and Miller- Graff 2018). The tools were validated as part of a process to develop an empirically defensible risk of violence assessment for young adults. The full instrument, including scales on emotion regulation, depression, post-traumatic stress, self-esteem, self-reported delinquency, resilience, social cohesion, and collective efficacy, took 40 min to one hour to complete. Scales were translated from English to Spanish, back translated, checked for understanding with a local youth group, and then piloted before data for the sample discussed here were collected.

The pilot involved 100 individuals both in communities and prisons to test survey understanding.

Changes after this test included removing detailed follow-up questions on items with no variation or low response rates (i.e. stealing and drug use). After capturing the full sample, we evaluated whether measures retained similar structure to the originally developed measures using confirmatory factor analysis to ensure construct validity. We also examined the factor loading to determine the extent to which the item contributes to the overall latent construct. Finally, internal reliabilities were calcu-lated using Cronbach’s alpha to measure how closely related sets of items (questions) are as a group.

The values of the alpha for each measure below met or exceeded the standard for the social sciences (Cronbach’s alpha > .70). Thus, we determined that there is strong measurement validity among the tool’s components in the target population.

Violence involvement

Two binary items asked whether or not the subject or anyone from their home had been victimised in the past six months. The six-month timeframe was used to reduce telescoping and to align the

4 T. HARE ET AL.

timeline with current perceptions of protective factors. A positive response to either question was coded as “1” while a negative response to both was coded “0”. The question and timeframe was included with the incarcerated population given that victimisation within Honduran prisons is common.

Non-violent and violent offences in the last six months were captured using the International Self- Report Delinquency (ISRD) study. Questions on non-violent offences were: “Have you damaged some-thing on purpose?”, “Have you stolen anything from someone or somewhere?”, and “Have you sold drugs, served as an intermediary or lookout for a drug sale?” Questions on violent offences were: “Did you threaten anyone with a gun or with hitting them in order to get money or something else?”, “Did you participate in a group fight in a public place?”, and “Did you hit someone on purpose or hurt them with a stick or knife?” The question and timeframe was included with the incarcerated popu-lation given that continued violent and non-violent activity within Honduran prisons is common.

While there is some debate over self-report measures of deviant behaviour, they have shown a con-sistent correlation with police data in other contexts that outweighs concerns about systematic bias (Krohn et al. 2010).

Resilience

The Adult Resilience Measure (ARM) was designed to measure the resources available to individuals that may bolster their resilience. The measure includes individual, relational, communal and cultural items and has been used in multiple contexts across the world. The ARM uses the social-ecological framework to consider the influence of relationships on an individual, and the individual’s perception of those relationships and context. We validated the ARM in Honduras after some reduction of items that did not contribute to the validation (Cronbach’s alpha .83).

Questions on resilience included Likert-scale responses to: “I feel supported bymy friends”, “I know where to go if I need help”, “My friends care about me when times are hard (for example, when I am ill or in trouble)”, “People think that I am fun to be with”, “I feel secure when I am with my family”, “I like my family’s culture and the way my family celebrates things (e.g. holidays)”, “My family is supportive towards me”, “I am proud of my ethnic background (for example, I am proud of where my family comes from or know a lot about my family’s history)”, “I am treated fairly”, and “I feel that I belong in my community”.

Social control and cohesion

The Community Cohesion and Efficacy Survey (CCES) is used to measure social cohesion and willing-ness to intervene on behalf of the common good, primarily in the United States (Sampson, Rauden-bush, and Earls 1997). The CCES comprises two factors: social cohesion and trust, and social control.

We validated the CCES in Honduras, retaining all original items (Cronbach’s alpha .70).

Questions on social cohesion included responses to a five-point Likert scale ranging from “strongly agree” to “strongly disagree” and included items such as: “This is a tight-knit neighbourhood”, “The people in this neighbourhood can be trusted”, and “The people in this neighbourhood do not share the same values.” Questions on social control included responses to the same scale and included items such as: “People around here are willing to help their neighbours”, and “People around here generally do not get along.”

Control variables

Measures of historical experience of family, delinquency, and drug and alcohol abuse come from Violence Risk Appraisal Guide (VRAG). The VRAG is an actuarial tool designed to assess risk of violent recidivism. It has been extensively validated for the prediction of violent recidivism and consistently performed well in meta-analyses (Yang, Wong, and Coid 2010). Additional items

DEVELOPMENT IN PRACTICE 5

that examine current drug and alcohol abuse are from the ISRD and the Centers for Disease Control.

Analytic plan

In order to evaluate the association between social protective factors (i.e. resilience and cohesion) and violence involvement, we conducted logistic regression models for both non-incarcerated and incarcerated populations – one for each bivariate outcome (violent offending, non-violent offending, victimisation). Because non-incarcerated individuals were drawn from within the same community, results were clustered by neighbourhood of data collection to produce robust standard errors. Hypothesised social protective factors, including resilience and social cohesion, were entered into the model as independent variables, controlling for socio-demographic factors (i.e. age, sex, number of children, employment, education) and risk factors (i.e. presence of both parents, behaviour problems in childhood, drinking, and drug use). Coefficients produced were odds ratios where a value of >1 represents an increased risk of the dependent variable and a value of <1 represents a decreased risk of the dependent variable. Significant results were defined as those coefficients whose 95% confi-dence intervals did not include 1.

Results

Descriptive statistics for the sample disaggregated by incarceration status are presented in Table 1.

By design, the majority of the sample were non-incarcerated individuals (n = 936) while the balance were incarcerated (n = 277). Overall, there are not large enough demographic differences to declare the two samples unbalanced by examining the difference between columns in Table 1.

The notable exception to this is number of children, where non-incarcerated participants had an average of .7 children and incarcerated participants had 1.4 children on average (t = −2.3, p < .05).

Victimisation does not differ between the two samples, but self-reporting of violent and non-violent offence did differ as expected, with higher reports among the incarcerated population (t = −4.4 and t =−2.7 respectively, p < .05). Scores for resilience (ARM) were 34.6 for the non-incar-cerated population and 32.7 for the incarcerated population, and these scores and the difference

Table 1. Descriptive characteristics by incarceration status.

Non-incarcerated Incarcerated Difference

Mean t-value Mean t-value Mean t-value

Demographics Age 22.4 69.3* 23.7 18.9* −1.3 −1 Female (%) 45.7 9.5* 46.2 1.7 −0.5 # Number of children 0.7 15.6* 1.4 4.9* −0.7 −2.3* Employed^ (%) 20.5 9.2* 24.7 3.0* −4.3 −0.5 Attend school^ (%) 28.3 11.7* 35.5 3.6* −7.2 −0.7

Violence involvement Victim (%) 33.2 15.4* 37.7 10.9* −4.5 −1.2 Violent offence (%) 17.8 10.0* 46.4 6.9* −28.6 −4.4* Non-violent offence (%) 5.5 8.1* 18.5 3.7* −13 −2.7*

ARM 34.6 140.2* 32.7 43.0* 1.9 2.4*

CCES 22.6 48.4* 21.7 30.8* 0.9 1.1

Control Live with both parents < 15 (%) 58.7 26.4* 44.6 7.5* 14.1 2.2* Behaviour problems < 15 2.6 17.1* 4.9 8.4* −2.3 −4.0* Alcohol problem (%) 15.6 8.4* 15.9 3.8* −0.3 −0.1 Illegal drug problem (%) 7.5 4.3* 21.4 4.0* −13.9 −2.7*

N 936 277

Notes: #rounds to zero; *p-value < 0.5; ^prior to incarceration.

6 T. HARE ET AL.

between them are statistically significant (t = 2.4, p < .05). Differences between the two populations on the scores for social control and cohesion (CCES) were not statistically significant.

Control variables are included to help consider the size of the association between social protec-tive factors and violence involvement in relation to other commonly referenced risk and protective factors. Living with both parents (t = 2.2, p < .05), early behaviour problems (t =−4.0, p < .05) and current drug problems (t =−2.7, p < .05) had statistically significant differences between the two samples.

In order to test the hypothesis that there is a significant association between social protective factors (i.e. resilience and cohesion) and violence involvement, Table 2 presents the odds ratios for the independent variables for each of the categories of violence involvement by incarceration status.

Victimisation

Results of the regression show that having larger numbers of children is associated with lower odds of victimisation in both non-incarcerated and incarcerated samples (potentially due to the need to stay home to care for children and therefore less exposure to risk on the streets). Age is also associated with increased victimisation, but only in the incarcerated sample. In line with the hypothesis, resili-ence is associated with slightly lower odds of victimisation for both samples. Having behaviour early life problems is associated with higher odds of being a victim in both samples. No other pre-dictors, including cohesion, are large and statistically significant for victimisation.

Non-violent offences

In line with the hypothesis, higher levels of resilience reduce the odds of non-violent offences for the non-incarcerated population. The association between cohesion and non-violent offences is not stat-istically significant in either sample and thus fails to support the hypothesis. Early life behaviour pro-blems are associated with much higher odds of non-violent offences across groups. No other predictors are statistically significant for non-violent offences.

Violent offences

Contrary to our hypothesis, the association between resilience and cohesion with violent offences is not statistically significant for either incarcerated or non-incarcerated individuals. As with non-violent

Table 2. Social protective factor association with victimisation, non-violent, and violent offences by incarceration status.

Non-incarcerated Incarcerated

Victim Non-violent offence

Violent offence Victim

Non-violent offence

Violent offence

Demographics Age 1.01 0.95 0.94 1.09* 1.04 0.94 Female 1.00 0.99 0.99* 1.01* 1.01* 1.00 Children 0.82* 0.98 1.18 0.76* 0.64 1.10 Employed^ 1.00 1.00 1.00 1.01* 1.00 1.00 Attend school^ 1.00 1.00 1.00 1.00 0.99 0.99

ARM 0.96* 0.92* 0.96 0.94* 0.98 0.98

CCES 1.03 1.00 1.02 0.98 1.01 0.99

Control Live with both parents <

1.00 1.00 1.00 1.00 0.99* 1.00

Behaviour problems < 15 1.14* 1.34* 1.33* 1.17* 1.35* 1.38* Alcohol problem 1.00 1.00 1.01 1.00 1.01 1.01 Illegal drug problem 1.00 1.01 1.01* 0.99 1.01 1.01 n 902 902 902 267 267 267

Notes: *p < 0.05; ^prior to incarceration.

DEVELOPMENT IN PRACTICE 7

offences, having early life behaviour problems greatly increases the odds of violent offence across groups. No other predictors are large and statistically significant for violent offences.

Results summary

General trends in Table 2 show that resilience is associated with lower odds of violence involvement among non-incarcerated individuals (8% lower odds of a non-violent offence) and victimisation among incarcerated individuals (6% lower odds of being a victim), while cohesion not associated with violence involvement for either incarcerated or non-incarcerated individuals. There is support for the hypothesis that an increase in resilience is associated with a decrease in violence involvement across groups (especially for victimisation). However, there is no support in these data for the hypoth-esis that an increase in cohesion is associated with a decrease violence involvement. Finally, neither resilience nor cohesion is as strongly associated with violence involvement as early life behaviour pro-blems across groups.

Discussion

Understanding protective factors associated with a lower likelihood of victimisation or committing a violent or non-violent offence is particularly urgent in Honduras and its Central American neighbours.

However, despite a large amount of prescriptive literature, and a growing literature on individual risk and protective factors in the region (Olate, Salas-Wright, and Vaughn 2012), very little empirical data exist on protective factors at the relational and contextual levels of the social-ecological framework.

This is due in part to fewer validated tools available to measure relationship, community, and societal factors in the sociological literature, compared to tools that measure individual factors in the psycho-logical literature (Grych, Hamby, and Banyard 2015). This dearth in empirical data may also relate a lack of reporting on null findings, despite their importance.

Even without good empirical support, citizen security policies and programmes in Central America frequently employ violence prevention mechanisms at the relational and contextual levels in neigh-bourhoods and communities. The strong focus on intervening with violence prevention mechanisms at these levels demands empirical analysis no matter what the findings. Programmes at the relational level seek to build resilience and cohesion through increased positive peer interactions. For youth, interactions include participation in sports, social activities, and after-school programmes. Develop-ment agencies and local government have also supported youth soccer and basketball teams to provide opportunities to play a sport while increasing teamwork and social skills. Social activities that ultimately seek to reduce violence include community clean ups and clubs for chess, music, and other hobbies. After-school programmes contain some mix of sports and social activities, as well as skill building in computer science and languages. For adults, violence prevention programmes include those that seek to increase participation in neighbourhood associations, clubs, and organis-ations. At the contextual level, violence prevention programmes seek to increase resilience and cohe-sion through participation in municipal committees and through the opening or rehabilitation of public spaces. Participation in municipal decision-making is believed to increase trust in local officials and provide important input on critical issues that impact the wellbeing and security of com-munity members. Public spaces are believed to be important to provide opportunities for community members to interact and develop sense of community around shared resources. All of these pro-grammes are carried out with the programme theory that if these interactions and participation increase resilience and cohesion, then violence involvement will decrease.

Results from our study found some support to suggest that higher levels of multi-systemic resili-ence, including individual, relational, and contextual resilience, is associated with lower levels of vio-lence involvement in this sample. Importantly, the study examined individuals’ total resilience across these systems rather than the contribution of resilience at each social ecological level, differentially.

First, these data suggest that a holistic understanding of resilience is important when working with

8 T. HARE ET AL.

violence-involved individuals. Prevention programmes would do well to consider how to diversify intervention components to reflect the core domains of resilience measured here. This includes fos-tering personal skills, social skills, peer and familial support systems, as well as helping individuals identify how they might engage in community resources they value (e.g. spiritual, educational, cul-tural). It is also important to recognise, however, that many individuals have demonstrated resilience amidst hardship. As such, it may be useful for programmes to conduct a baseline assessment with incoming constituents to assess unique patterns of strengths and weaknesses, tailoring available sup-ports accordingly.

The incorporation of resilience frameworks into practice has significant appeal to many prac-titioners and these data provide preliminary support for the relevance of this approach. It is impor-tant, however, that before embarking on resilience-focused interventions, practitioners carefully consider how best to assess change, considering carefully the theory undergirding the types of resi-lience facilitated and the types of positive adaptation such resilient processes should predict (Masten 2011). This likely requires programmes across sectors to consider coordinated efforts that may work synergistically (Masten 2011).

We did not, however, find strong support in these data for the hypothesis that higher levels of social cohesion are associated with lower levels of violence involvement. Social cohesion, as it was measured in the current study, represents the extent to which individuals felt a sense of trust and shared values with others in their neighbourhoods. Unlike contextual resilience, which assesses the extent to which individuals access and value resources at the community level, social cohesion reflected individual perceptions of the quality of the social context. That social cohesion was not associated with violence perpetration has important implications for development practitioners.

That is, despite the popularity of programmes aimed at facilitating a sense trust, shared values, and neighbourhood relationships, the efficacy of such efforts to reduce violence is unclear. Pro-gramme theories that employ cohesion should be examined critically to determine whether or not they are effective, and a greater emphasis on family and friend relationships as a component of resi-lience versus community building more generally might improve violence reduction effectiveness.

This research also highlights the fact that early life behaviour problems and drug use may be more strongly associated with violence involvement than are current social protective factors. These findings are consistent with previous work; early behavioural problems have been consistently shown to predict violence involvement (Swinford et al. 2000). In the context of Honduras, it is impor-tant to recognise that the pervasive violence in communities may explain the presence of some of these early behaviour problems. Given that behavioural problems are readily observable by care-givers and teachers, and that importance of early intervention has been consistently underscored in the research literature (Stormont 2002), programmes should consider support for behavioural health in early childhood as a means of addressing and remediating risk for later violence involve-ment. Adolescent drug use has also been found to predict future victimisation, but victimisation also predicts future illegal drug use (Weiner et al. 2005), suggesting complex reciprocal relationships over time. These reciprocal patterns of influence suggest the relevance of programmatic integration that both addresses youths’ victimisation experiences and works to remediate the physiological effects of drug use and addiction (Weiner et al. 2005). Similarly, this suggests that development prac-titioners would be well-served to consider including interventions that target youth under the age of 15 to address early victimisation and prevent behaviour problems and drug use as these factors are more associated with violence involvement among young adults than any social factors measured here.

Overall, the association of protective factors with violence involvement appears to grow weaker the further a factor is from the individual level in the social ecological framework. It appears from these results that a programmatic focus on the relationship level (i.e. microsystem of family and close peers) holds more promise than a focus on the community level (i.e. neighbourhood and social system). These findings are consistent with both empirical and descriptive research that explores protective factors at the relationship and community levels in the United States and

DEVELOPMENT IN PRACTICE 9

elsewhere (Hansen-Nord et al. 2016; Hare 2018; Sampson 2012). Development programmes that intervene to increase resilience and cohesion and thus decrease violence through activities for youth, participation for adults, and improved spaces for public use in high violence areas may not reduce violence involvement. However, future research should explore other positive results of these programmes such as increased social capital, collaboration, and efficacy.

Limitations and future directions

Despite the numerous strengths of this study, including its large sample and its inclusion of both community and incarcerated populations from high-risk areas of Tegucigalpa, some limitations should be noted. First, the study was cross-sectional, and as such, violence involvement and resilience were assessed contemporaneously, with participants asked to report on violence involvement in the previous six months. Given this, even though some important implications can be drawn from the identified associations between resilience and violence involvement, the study results should not be viewed as causal in nature. Second, the non-incarcerated and incarcerated samples were not drawn at random. In both cases, people volunteered to participate in the study. Thus, generalisations to the population from these neighbourhoods or the incarcerated population are not necessarily stat-istically representative. Third, both populations were interviewed face-to-face, which may affect the degree at to which individuals are willing to report non-biased responses to criminal activities (though the face validity of difference in results between samples suggests some confidence in responses). We cannot rule out a downward bias on those questions. Future research should consider longitudinal evaluations that engage in multiple methods of assessment, especially for those domains most vulnerable to under-reporting.

This study highlights important future directions for related research questions and the results stand to make important contributions to both research and practice. Specifically, they suggest that development researchers and practitioners need to consider whether social protective factors function as expected in high violence situations. Absent evidence that can show a stronger associ-ation between relational and contextual resilience and violence involvement, or cohesion and vio-lence involvement, practitioners should consider approaches to preventing violence that focus more on individual resilience, early life experience, and early life behaviours.

Acknowledgements

Support for this research was provided by the United States Agency for International Development through awards AID- 522-TO-16-00002 and AID-OAA-I-15-00019 to Dexis Consulting. The authors’ views expressed in this publication do not necessarily reflect the views of the United States Agency for International Development, the United States Government, or Dexis Consulting.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

Support for this research was provided by the United States Agency for International Development through awards AID- 522-TO-16-00002 and AID-OAA-I-15-00019 to Dexis Consulting

Notes on contributors

Tom Hare is Senior Technical Associate at the Notre Dame Initiative for Global Development. His research examines vio-lence prevention policies, programmes, and theories in Central America. Hare is the author of Zonas Peligrosas (Fordham University Press), and several articles that critically assess local and international violence prevention efforts in the region.

10 T. HARE ET AL.

Laura Miller-Graff is Assistant Professor of Psychology and Peace Studies at the University of Notre Dame. Her research seeks to understand how various ecological systems interact to promote or inhibit healthful development following vio-lence exposure. Miller-Graff’s research also seeks to develop and evaluate transdiagnostic interventions in contexts where violence is systemic and chronic.

Juan Carlos Guzman is Senior Research Associate at the Notre Dame Initiative for Global Development. Guzman’s research focuses on the impact of development projects on human well-being in Central American and sub-Saharan African countries.

References

Adams, T. M. 2017. How Chronic Violence Affects Human Development, Social Relations, and the Practice of Citizenship.

Washington: Wilson Center.

Bruneau, T. C., and L. Dammert. 2011. Maras: Gang Violence and Security in Central America. Austin: University of Texas Press.

Cano, I., and E. Rojido. 2016. Mapeo de programas de prevención de homicidios en América Latina y el Caribe [Mapping of homicide prevention programs in Latin America and the Caribbean]. New York: Open Society Foundation.

Grych, J. H., S. Hamby, and V. Banyard. 2015. “The Resilience Portfolio Model: Understanding Healthy Adaptation in Victims of Violence.” Psychology of Violence 5 (4): 343–354.

Hamby, S., and J. Grych. 2012. The web of Violence: Exploring Connections Among Different Forms of Interpersonal Violence and Abuse. Berlin: Springer Science and Business Media.

Hansen-Nord, N., F. Kjaerulf, J. Almendarez, V. Morales Rodas, and J. Castro. 2016. “Reducing Violence in Poor Urban Areas of Honduras by Building Community Resilience Through Community-Based Interventions.” International Journal of Public Health 61 (8): 935–943.

Hare, T. 2018. Zonas Peligrosas: The Challenge of Creating Safe Neighbourhoods in Central America. New York: Fordham University Press.

Hare, T., J. C. Guzman, and L. Miller-Graff. 2018. “Identifying High-Risk Young Adults for Violence Prevention: A Validation of Psychometric and Social Scales in Honduras.” Journal of Crime and Justice 41 (2): 627–642.

Hennigan, K. H., C. L. Maxson, D. C. Sloane, K. A. Kolnick, and F. Vindel. 2014. “Identifying High-Risk Youth for Secondary Gang Prevention.” Journal of Crime and Justice 37 (1): 104–128.

Inter-American Development Bank. 2011. “MAPEO de las intervenciones de seguridad ciudadana en Centroamérica financiadas por la cooperación internacional” [Mapping of citizen security initiatives in Central Americafinanced by international agencies]. Accessed April 24, 2019. www.wola.org.

Krohn, M. D., T. P. Thornberry, C. L. Gibson, and J. M. Baldwin. 2010. “The Development and Impact of Self-Report Measures of Crime and Delinquency.” Journal of Quantitative Criminology 26 (4): 509–525.

Malik, S., S. B. Sorenson, and C. S. Aneshensel. 1997. “Community and Dating Violence Among Adolescents: Perpetration and Victimization.” Journal of Adolescent Health 21 (5): 291–302.

Masten, A. S. 2011. “Resilience in Children Threatened by Extreme Adversity: Frameworks for Research, Practice, and Translational Synergy.” Development and Psychopathology 23 (2): 493–506.

Moser, C., and C. McIlwane. 2006. “Latin American Urban Violence as a Development Concern: Towards a Framework for Violence Reduction.” World Development 34 (1): 89–112.

Muggah, R. 2017. “The rise of citizen security in Latin America and the Caribbean.” In Alternative Pathways to Sustainable Development: Lessons from Latin America, International Development Policy Series No.9, 291-322. Geneva, Boston:

Graduate Institute Publications.

Olate, R., C. P. Salas-Wright, and M. G. Vaughn. 2012. “Predictors of Violence and Delinquency Among High Risk Youth Gang Members in San Salvador, El Salvador.” International Social Work 55 (3): 383–401.

Sampson, R. 2012. Great American City: Chicago and the Enduring Neighbourhood Effect. Chicago: University of Chicago Press.

Sampson, R., S. Raudenbush, and F. Earls. 1997. “Neighbourhoods and Violent Crime: A Multilevel Study of Collective

Efficacy.” Science 277 (5328): 918–924.

Stormont, M. 2002. “Externalizing Behavior Problems in Young Children: Contributing Factors and Early Intervention.”

Psychology in the Schools 39 (2): 127–138.

Stouthamer-Loeber, M., R. Loeber, E. Wei, D. P. Farrington, and P. H. Wikström. 2002. “Risk and Promotive Effects in the

Explanation of Persistent Serious Delinquency in Boys.” Journal of Consulting and Clinical Psychology 70 (1): 111–123.

Swinford, S. P., A. DeMaris, S. A. Cernkovich, and P. C. Giordano. 2000. “Harsh Physical Discipline in Childhood and

Violence in Later Romantic Involvements: The Mediating Role of Problem Behaviours.” Journal of Marriage and Family 62 (2): 508–519.

Ttofi, M., D. Farrington, A. Piquero, and M. DeLisi. 2016. “Protective Factors Against Offending and Violence: Results From Prospective Longitudinal Studies.” Journal of Criminal Justice 45: 1–3.

UNAH-IUDPAS (Universidad Nacional Autónoma de Honduras – Instituto Universitario en Democracia, Paz y Seguridad).

2018. “Observatorio de la violencia, enero-diciembre 2017.” Accessed April 24, 2019. https://iudpas.unah.edu.hn/ observatorio-de-la-violencia/boletines-del-observatorio-2/boletines-nacionales/.

DEVELOPMENT IN PRACTICE 11

http://www.wola.org https://iudpas.unah.edu.hn/observatorio-de-la-violencia/boletines-del-observatorio-2/boletines-nacionales/ https://iudpas.unah.edu.hn/observatorio-de-la-violencia/boletines-del-observatorio-2/boletines-nacionales/

Ungar, M. 2008. “Resilience Across Cultures.” British Journal of Social Work 38 (2): 218–235.

United Nations Development Programme. 2009. Opening Spaces to Citizen Security and Human Development: Human

Development Report for Central America. New York: UNDP.

United Nations Office on Drugs and Crime. 2015. Global Study on Homicide. Vienna: UNODC.

Weiner, M. D., S. Sussman, P. Sun, and C. Dent. 2005. “Explaining the Link Between Violence Perpetration, Victimization and Drug Use.” Addictive Behaviors 30 (6): 1261–1266.

World Health Organisation. 2002. “World Report on Violence and Health.” Geneva: WHO.

Yang, M., S. C. P. Wong, and J. Coid. 2010. “The Efficacy of Violence Prediction: A Meta-Analytic Comparison of Nine Risk

Assessment Tools.” Psychological Bulletin 136 (5): 740–767.

12 T. HARE ET AL.