Problem Analysis in Community Violence Assessment: Revealing Early Childhood Trauma as a Driver of Youth and Gang Violence

Abstract

Problem analysis conducted by a university-based research partner can provide communities with data-driven options to address the local drivers of serious youth and gang violence. Situated in Worcester, Massachusetts, this article describes how subsequent to early childhood trauma being identified as a potential driver of adolescent and young adult violence; problem analysis using local data confirmed that being the victim or witness of a traumatic incident before the age of 12 years was significantly correlated with involvement in violence in adolescence or young adulthood. Although there is a robust literature on the relationship between early childhood trauma and later delinquency, local decision makers did not consider this knowledge actionable until the research partner used the city’s own police records to demonstrate the extent of the problem in the city. Rigorous problem analysis, conducted collaboratively between practitioners and an academic research partner, helped to compel local change and ensured that strategies addressed the right risk factors and directed service to the appropriate target population.

Keywords

community violence assessment problem analysis practitioner–academic partnership early childhood trauma action research

Introduction

Youth violence—including gang-related violence—is a pressing public health concern for cities across the United States. According to the Centers for Disease Control and Prevention (CDC), in 2014, homicide was the third leading cause of death among youth aged 10 to 24 years, and violence was a major cause of nonfatal injuries among youth. In 2014, a total of 431,264 young people aged 10 to 24 years were treated and released from emergency departments for nonfatal injuries sustained from assaults. The 2015 Youth Risk Behavior Survey revealed that among youth in Grades 9 to 12, 22.6% reported being in a physical fight 12 months preceding the survey, and 16.2% reported carrying a weapon (gun, knife or club) on one or more days in the 30 days preceding the survey. The CDC estimated that each year youth homicides and assault-related injuries result in an estimated US$16 billion in medical and work loss costs (CDC, 2012). Although difficult to track definitively, the Federal Bureau of Investigation (FBI) estimates that 13% of the nation’s homicides are gang related. Because the causes of youth violence are multifaceted and complex, there is growing recognition that relying on law enforcement and suppression alone is an inadequate response to this public health problem.

The Comprehensive Gang Model (CGM) is a multisector, collaborative approach that can address the numerous risk factors that contribute to gang and youth violence in a community (Gebo, Bond, & Campos, 2015; Spergel, 1995). The CGM, as a framework and not a prescribed program, responds to a community’s particular risk factors by strategically directing intervention, opportunity provision, and suppression resources toward gang-involved and high-risk youth while mobilizing the community and increasing organizational capacity through training, policy, and procedure change. The CGM has been shown to decrease gang-related robberies and reduce levels of violent crime and drug arrests among gang members (McGarrell et al., 2012). Yet due to the complexity of the CGM, many communities have faced implementation challenges that have weakened its effectiveness (Gebo et al., 2015; Howell, 2012).

Implementation challenges can occur when communities adopt the CGM without first conducting a comprehensive and methodically rigorous analysis of their youth and gang violence problem (Braga & Hureau, 2012; Braga, McDevitt, & Pierce, 2006; Gebo et al., 2015; Howell, 2010). Absent a thorough assessment, communities are unable to identify the appropriate target population for prevention, intervention, and suppression strategies and may not understand organizational factors that are inadvertently contributing to the gang and youth violence problem. Without pertinent data, communities risk wasting resources on replicating what has been traditionally done, but perhaps has not worked, rather than developing data-driven responses to a community’s particular dynamics and risk factors (Braga & Hureau, 2012). The lack of problem analysis also leaves a community without meaningful baseline data to track progress and strategy effectiveness. The Office of Juvenile Justice and Delinquency Prevention (OJJDP) produced a detailed guide for a community gang assessment process (OJJDP, 2009); yet communities may lack the capacity to conduct the rigorous analysis needed to develop a community assessment that can guide strategy development and implementation.

To address this limitation, the field of criminal justice is increasingly moving toward the inclusion of university-based research partners in comprehensive approaches to public safety issues (Braga & Hureau, 2012; Burkhardt et al., 2017; Rojek, Smith, & Alpert, 2012; Worden, McLean, & Bonner, 2014). The inclusion of research partners has been credited with reducing youth gun use (Braga et al., 2006), assisting police in interactions with individuals with mental illness (Burkhardt et al., 2017), and increasing community understanding of gang presence and activities (Takata & Tyler, 1995). Referred to as “real time social science,“ the inclusion of academic research partners supports community members and practitioners in a five-phase action research cycle consisting of problem identification, problem analysis, strategy development, assessment of strategy performance, and strategy modification (Braga & Hureau, 2012; see Figure 1).

Figure 1.

Action research cycle.

Academic research partners are particularly important in the problem analysis phase. Problem analysis involves in-depth, multidimensional, systematic assessment of crime problems at the local level, including an examination of underlying factors that lead to crime and disorder to develop informed responses (Boba, 2003; Braga et al., 2006). This analysis is necessary to develop relevant and effective strategies; yet it has been found to be the weakest and most overlooked phase of the action research cycle. Reasons for this are that problem analysis requires different forms of data, knowledge, analysis capacities, and skills than traditional crime analysis (Boba, 2003; Santos, 2014). Rigorous and comprehensive problem analysis is difficult to achieve if police cannot build these capacities internally and/or lack partnerships with external researchers.

This article features a Worcester, Massachusetts, initiative based on the CGM called Project Sure Steps and this initiative’s problem analysis phase in its comprehensive community violence assessment. During the problem identification phase, early childhood trauma was identified as a potential risk factor in the local violence landscape that had previously gone unrecognized. The problem analysis phase revealed dimensions about the relationship between early childhood trauma and later involvement in violence that were critical to motivating local action and developing appropriate strategies. The major conclusions of this article are twofold. One, rigorous problem analysis—assisted by a research partner—is needed for effective strategy development. Two, problem analysis using local data can catalyze new stakeholders and community resources to be directed to youth violence prevention and intervention efforts.

Project Background

Launched in 2006, Project Sure Steps is a multisector approach to reduce gang violence, recidivism, and retaliation and to increase the education and employment opportunities for high-risk and gang-involved youth and young adults. Directed by a police sergeant, the project brings together partners from youth agencies, workforce development providers, mental health organizations, and the schools along with juvenile and criminal justice sectors and research partners from Clark University.

In 2006, when Project Sure Steps began, the city of Worcester (population over 185,000) had a variety of characteristics that rendered young people vulnerable to gang involvement and violence. Nearly one third of all youth in Worcester lived in poverty; the percentage was 44% for Hispanic youth. One third of Worcester households was headed by a single female, and the 4-year graduation rate was lower than that of the state of Massachusetts. Young people aged 16 to 24 years were disproportionately affected by unemployment and limited job prospects. They were also disproportionately affected by violence.

In an initial analysis of the city’s gang problem, Braga (2006) found that young people were committing a great deal of the violent crime in the city. For example, 38% of arrests for assault and battery, disorderly conduct, distribution and sales of drugs, and firearm-related incidents were of young people under the age of 24 years. Much of the violent crime was gang related. There were approximately 700 to 900 gang members in the city; the average age of a gang member was 17 years old; and gang activity was estimated to be responsible for 43% of homicides and 37% of nonfatal shootings. Black and Latino males were disproportionately the victims and perpetrators of violent and gang crime (Braga, 2006). In this context, Project Sure Steps combined targeted suppression for the most violent offenders, with the provision of case management, street outreach, youth employment, education support, and other engagement opportunities for high-risk youth.

Fast forward to 2013. Homicides in Worcester had been in the single digits each year since the project was launched. Gang turf remained limited to 2% of the city, and the pool of gang victims and offenders was relatively small. Worcester’s juvenile arrest rate had declined by 40% since 2006. Hundreds of high-risk youth had received workforce development training and employment, and many had reconnected with school through Project Sure Steps efforts. Community-based agencies, the schools, and the Worcester Police Department (WPD) were sharing information and working together. In spite of these successes, the victims of shootings and homicides continued to be disproportionately Black and Latino men between the ages of 17 and 27 years. In 2013, although only constituting roughly 20% of the juvenile population, Latino males accounted for 55% of all juvenile male arrests. We saw that the arrest rate for Latino males rise, as arrests for other juveniles fell.

It was these latter two data points that caused Project Sure Steps leadership great concern. They began to ask why—in spite of 6 years of programmatic effort—were these patterns in violent crime and juvenile arrest persisting. Project leadership asked the research partner team to take on a comprehensive community assessment of the factors driving youth and gang violence in Worcester. The action research team consisted of two faculty members who are social scientists, a project manager, and master’s level graduate students with specializations in youth development, geographic information sciences, econometrics, and community development.

The research team followed procedures recommended by the OJJDP (2009) to analyze community gang and youth violence problems. We convened a leadership group of key decision makers including individuals from the schools, the city, business, and the project’s advisory committee. The team adopted Howell’s (2012) framework to identify Worcester-specific youth violence risk factors across age and ecological domains, including individual, family, school, peer group, and neighborhood/community.

The assessment drew on available secondary data including the 2010 Census, Massachusetts Department of Early and Secondary Education, and the Worcester Community Health Assessment to identify unaddressed factors that could potentially be driving violence in Worcester. We examined the police records of 100 young men between 14 and 24 years old who had been identified by the police department as victims or perpetrators of gun or knife violence and referred to a special program to connect these men to employment, education, and behavioral health supports. By reviewing their case records, we learned that over 30% of these individuals had their first police contact before the age of 12, the majority of the time as a victim or witness. At least one third of these young men were fathers of infants and very young children. This glimpse into the cyclical and generational nature of violence among this group compelled the leadership team to want to know more about the relationship between early trauma and later involvement in violence among Worcester males and what could be done to interrupt this cycle.

Guided by the project’s leadership group, the research team assembled these data into a report that was disseminated to community stakeholders including city government, youth-serving agencies, parents, and youth for comment. Over the course of 4 months, the research team organized discussions and community meetings to gain public input into the findings, to prioritize the issues, and to guide strategic planning. Hundreds of residents of all ages deliberated over the findings and potential priorities for action. Public engagement prioritized the issue of early childhood trauma for more in-depth problem analysis.

Method

Once community members identified early childhood trauma as a priority area, the research team’s problem analysis consisted of a literature review on the relationship between early childhood trauma and later delinquency and a quantitative analysis of a larger police data set. The literature revealed a robust connection between early childhood trauma and later involvement in delinquency including violence; yet to foster change in community conditions, it was important to demonstrate these connections in the local context.

Quantitative analysis was based on an extract from the city’s police database consisting of males under the age of 28. The extract included 25,375 males involved in a total of 98,914 incidents. Due to changes that occurred in WPDs data management system, males over the age of 27 were unlikely to have comprehensive records of early childhood experiences. The lead author of this article is Criminal Justice Information Services (CJIS) certified to handle law enforcement data. The research team was able to explore the hypotheses generated in the problem identification phase where the city did not have the capacity to perform this type of analysis internally.

WPD assigned each individual in the data set a unique numeric code to protect their identities while still allowing tracking of the same individual across incidents and over time. WPD coded individuals based on their role in incidents. The roles included in this analysis were “victims,” “witnesses,” or “arrests.” Of the individuals in the data set, 6,790 had ever been arrested, or 26.9%; 11,083 had ever been a victim, or 44%; and 2,526 had ever been a witness, or 9.95% (see Table 1). The mean age of the individuals was 22 and the median age was 23. The mean age of first incident was 14 and the median age of first incident was 16.

Table 1.

Worcester Police Dataset (N = 25,375).

	%
Breakdown by gender
Male	100
Female	0
Breakdown by current age
0-10 years old	7.42
10-15 year old	8.6
16-18 years old	8.8
19-22 years old	21.34
23-27 years old	53.82
Breakdown by race
Black or African American	7.9
Hispanic/Latino	15.33
White	28.15
Other Race	2.75
Race Missing	45.87
Breakdown by childhood police contact (age 12 and under)
No childhood police contact	80.53
One contact	9.97
Two contacts	6.15
Three plus contacts	3.35
Police contact as victim	18.24
Police contact as witness	3.78
Police contact as witness and victim	2.2
Roles
Witness (ever)	9.95
Victim (ever)	44.04
Ever arrested	26.76
One arrest	8.15
Two plus arrests	18.61
Ever violent	32.04
Violent arrest	9.82

The main independent variable was witness- and victim-based police contact before the age of 12, which we refer to as Early Police Contact. This variable is a proxy for early childhood trauma. An individual was defined as having Early Police Contact if he had at least one recorded incident of police contact before the age of 12 where his role was coded as a witness or a victim. This includes incidents such as an ambulance call, assault and battery, assault with a dangerous weapon, injured person, or to check on the welfare of a child. In the data set, 4,940 individuals, or 19.47%, experienced at least one police encounter before the age of 12 as a witness or victim.

Early Police Contact was tested in multiple models. For the first version of the model, Early Police Contact was a dummy variable equal to 1 if the individual was reported as a victim or witness in the police data set before the age of 12 and 0 if they were not. A second model was used to explore differences in the effects of witness- and victim-based trauma. For this purpose, Early Police Contact was represented by four categorical groups. The first group had a police contact before the age of 12 and was coded as a victim, the second group had a police contact before the age of 12 and was coded as a witness, and the third group had police contacts before the age of 12 and was coded as both a victim and a witness. Individuals who did not have a police contact before the age of 12 were the fourth group and were used as the base group.

We tested the relationship between Early Police Contact and three variables: Violent Incidents, Involved in Violence, and Perpetration of Violence. Violent Incidents is a nonnegative variable ranging from 0 to 20 representing the number of violent incidents recorded for an individual after the age of 12 regardless of whether he was the victim, witness, or the alleged perpetrator. This variable was used to test whether individuals who experience childhood trauma were also involved in more violence. The dummy variable Involved in Violence is equal to 1 if the individual was involved in at least one incident of violence over the age of 12, meaning the individual has a value of 1 or more for Violence Incidents. Overall, 8,113 individuals, or 32%, of the sample were involved in at least one violent incident over the age of 12. Perpetration of Violence is a dummy variable equal to 1 if an individual has been arrested at least once for a violent incident and 0 if they have not.

The correlation between Early Police Contact and Violent Incidents was tested using a tobit model with a restricted dependent variable to control for the nonnegative nature of the variable. The tobit model is as follows:

Violent Incidents * = x β + u, u | x ~ N o r m a l (0, σ^{2}) .

where Violent Incidents = max (0,20); β = β₁ Early Police Contact + β₂ Demographics.

This model also controlled for available demographics and informed whether early police contact increases the risk of multiple incidents. The largest portion of individuals in the data set (41%) was involved in one incident, followed by individuals involved in two incidents (22%); over 20% of the individuals were involved in five or more incidents. Of the 8,113 individuals recorded as being involved in a violent incident, 2,493 were arrested for their involvement, or 9.82% of the individuals, in the data set.

A dprobit model was used to understand whether early police contact is associated with increased likelihood for involvement in violence in any role, whether as a perpetrator, a witness, or a victim later in life. The dprobit displays the estimated marginal effect at the sample means allowing the marginal effects of the independent variables across the models to be compared. The dprobit model used the binary dependent variable, Involved in Violence. The model is shown below:

\Pr (Involved in Violence = 1) = f (Early Police Contact, Race, Age) .

The dprobit model was also used to understand whether incidents of trauma affect the risk of perpetrating violence later in life. This was classified using the dependent variable Perpetration of Violence. Both models were also run with the categorical variables for Early Police Contact to understand the differing effects of witness- and victim-based trauma.

The other variables included in each of the models are control variables, including race, gender, and age. Gender is restricted to males. Race is implemented as a control using the following categories of race: Black; Hispanic; Other Race, Race Missing; and White. White is the control category for race in the models. Although socioeconomic status, mental health, family structure, and neighborhood are also important factors in studying delinquency and gang involvement, they were not included in the models as they are not available using the Worcester Police Department data.

Findings

In the estimated models for the effect of childhood trauma on violence, we found that early trauma had a positive and highly statistically significant impact on violence. A male who was coded as a witness or a victim in an early police contact was 20.9% more likely to perpetrate, be a victim, or be a witness to violence later in life than a male who had no documented early childhood police contacts, statistically significant at the 1% level. A male who had an early police contact in the role of witness was 21.2% more likely to be involved in a violent incident later in life as a victim, witness, or arrested, statistically significant at the 1% level. Counterintuitively, a male who had early police contact as a victim was 2.3% less likely to be involved in a violent incident later in life. A male who had an early police contact as both a victim and witness was 49.2% more likely to be involved in a violent incident later in life than a male who did not experience any trauma, statistically significant at the 1% level (see Table 2).

Table 2.

Effects of Childhood Trauma on Involvement in Violence and Violent Arrests.

Variables	Involvement in violence	Perpetration of violence	Involvement in violence	Perpetration of violence
Variables	(1)	(2)	(3)	(4)
Victim	−0.0230**	0.0101***
Witness	0.212***	−0.00179
Victim and witness	0.492***	0.156***
Early police contact			0.209***	0.0258***
Observations	25,375	25,362	25,375	25,362

Source. Results of probit estimations from Worcester Police Dataset (2015).

Note. Marginal effects are shown instead of coefficients. Involvement in Violence = 1 if an individual has at least once incident of violence recorded in the Worcester Police Dataset, 0 if they do not. Perpetration of Violence = 1 if an individual has been arrested for a violent incident recorded in the Worcester Police Dataset, 0 if they have not. Victim = 1 if an individual has a recorded incident of victimization before the age of 12, 0 if they do not. Witness = 1 if an individual has a recorded incident as a witness before the age of 12, 0 if they do not. Victim and Witness = 1 if an individual has recorded incidents as both a victim and a witness before the age of 12, 0 if they do not. Early Police Contact = 1 if an individual has any recorded incident of childhood trauma (victim or witness before the age of 12), 0 if they do not. The base group is individuals who have no recorded incidents of childhood trauma. Race and age are controlled for in both models.

p < .1. **p < .05. ***p < .01.

A male who had an early police contact was 2.58% more likely to perpetrate violence later in life than an individual who had no documented early police contacts, statistically significant at the 1% level. An individual who was a victim in an early police contact was 1% more likely to perpetrate violence later in life, statistically significant at the 1% level. The correlation between witness-based early police contacts and Perpetration of Violence was not statistically significant. An individual who experienced childhood trauma as both a witness and victim was 15.6% more likely to perpetrate violence later in life, statistically significant at the 1% level (see Table 2).

In Table 2, we see that any incident involving early police contact increased an individual’s likelihood of perpetrating violence; this supports the findings from the probit model. Any incident involving early police contact increased the number of violent incidents later in life by .8 for the average individual, statistically significant at the 1% level.

The number of violent incidents later in life was increased by a larger magnitude if an individual had solely experienced witness-based trauma, an increase in violent incidents of .531 (p < .01), than if an individual had solely experienced victim-based trauma, an increase in violent incidents of .458 (p < .01). The number of violent incidents was not surprisingly increased by the largest magnitude for individuals who had experienced both witness- and victim-based childhood trauma. This group had 3.078 more incidents of violence than the base group, individuals who had no trauma, statistically significant at the 1% level (see Table 3).

Table 3.

Effects of Childhood Trauma on Total Violent Incidents.

Variables	Number of violent incidents	Number of violent incidents
Variables	(1)	(2)
Victim	0.458***
Witness	0.531***
Victim and witness	3.078***
Early police contact		0.800***
σ	1.291***	1.342***
Observations	25,375	25,375

Source. Results of tobit estimations from Worcester Police Dataset (2015).

Note. Estimation using tobit model where the dependent variable, Violent Incidents is constrained from 0 to 20, representing the number of violent incidents. Standard errors in parentheses. Victim = 1 if an individual has a recorded incident of victimization before the age of 12, 0 if they do not. Witness = 1 if an individual has a recorded incident as a witness before the age of 12, 0 if they do not. Victim and Witness = 1 if an individual has recorded incidents as both a victim and a witness before the age of 12, 0 if they do not. Race is controlled for in the model. The base group for age is 0 to 10. Restricted to males only.

p < .1. **p < .05. ***p < .01.

Worcester’s Response

After using these findings to secure a small planning grant from the local community foundation, the Project Leadership Team and the research partner convened a working group made up of pediatricians, psychologists, early child care education professionals, domestic violence victim advocates, youth mobile crisis team behavioral health specialists, the Worcester Police Department Crisis Intervention Team and Gang Unit officers, and a representative from the city manager’s office. This team spent 9 months developing an intervention consisting of a police–community health worker (CHW) co-responder model that directs child and family intervention at the point of trauma as well as a robust referral network for longer term family support for both witnesses and victims (see Figure 2). The intervention—referred to locally as Worcester ACTs (Worcester Addresses Childhood Trauma)—is envisioned to start before symptoms even have time to manifest in a child. The working group felt that CHWs with specialized trauma-informed training would be best suited in this co-responder model in that they would have a greater likelihood of success gaining the trust of these very high-risk families who tend to distrust social service agencies and institutions.

Figure 2.

Worcester ACTs: Early childhood co-responder model.

Co-responding models have been shown to reduce the likelihood of injuries to children, youth, bystanders, and police during police encounters (Reuland, Draper, & Norton, 2012) and facilitate connecting children and youth to mental health and trauma services (Markey, Usher, Gruttadaro, Honberg, & Cochran, 2011). Several local funders have committed resources to support the development of Worcester ACTs. In addition, the findings of this problem analysis convinced police leadership to include in-service training about the impact of trauma on child development. To date, all 400+ officers in the Worcester Police Department have received a first round of this training. Early childhood trauma is also a special module in the training of Crisis Intervention Team Officers and all new recruits.

Discussion

It is well documented that children who experience childhood trauma—either as a victim or witness—are at greater risk to engage in serious delinquent behaviors in adolescence, including violence (Dahlberg & Potter, 2001; Howard, Feigelman, Li, Cross, & Rachuba, 2002; Smith & Thornberry, 1995; Widom & Maxfield, 2001; Wood, Foy, Layne, Pynoos, & James, 2002; see Berg, Stewart, Schreck, & Simons, 2012; Jennings, Piquero, & Reingle, 2012; Lauritsen, Sampson, & Laub, 1991; Ousey, Wilcox, & Fisher, 2011; Posick & Gould, 2015 for evidence of the offender–victim overlap). Trauma in early childhood has detrimental effects on brain development in areas that regulate fear response, impulse control, reasoning, planning, and academic learning (Eckenrode, Laird, & Doris, 1993; Herrenkohl, Hong, Klika, Herrenkohl, & Russo, 2013). These effects on the brain can cause children to have extreme reactions to seemingly low-stress incidents. Hypervigilance and exaggerated reactions result from the stress response system activating more frequently and for longer periods than is necessary (Fisher, Gunnar, Chamberlain, & Reid, 2000). Long-term, unaddressed trauma, also known as toxic stress, is associated with mental and physical health disorders as well as overall shorter life expectancy as adults (Felitti et al., 1998; Shonkoff & Garner, 2012). Although the brain continues to develop over the life span, some stress-related changes are resistant to reversal. Therefore, early intervention to address the effects of trauma while the child’s brain is still sufficiently “plastic” or open to influence is essential (Davidson & McEwen, 2012).

Worcester’s comprehensive community assessment exposed drivers of youth and gang violence that previously had gone unaddressed. Revealing early childhood trauma as one of those drivers helped to explain why certain patterns in Worcester’s youth violence problem have persisted in spite of years of programmatic intervention. In-depth problem analysis allowed city decision makers to understand that there was a significant correlation between recorded incidents in which boys were a witness or victim to a traumatic event and later involvement in violence as a victim, witness, or being arrested. As a research partner, we were able to establish that boys who experienced both witness- and victim-based trauma had a significantly higher likelihood of engaging in violence later in life than boys who had not experienced early trauma. These boys were over 49% more likely to have a violent incident later in life and predicted to be involved in roughly three more recorded incidents involving violence as adolescents or young adults than boys who had no early police-recorded incidents, controlling for age and race. As the majority of individuals in the data set only had one recorded incidence of violence, the serious risk of early exposure to violence cannot be emphasized enough.

Of particular surprise to the decision makers was the predictive nature of witnessing violence. Boys who only witnessed violence were more likely to experience violence later in life than boys who were victims only. We hypothesize the reason for this counterintuitive finding because victims are more likely to receive an intervention than witnesses. This finding about witnessing violence proved particularly powerful in catalyzing new stakeholders to come to the table. Although the literature is robust about the relationship between early childhood trauma and later involvement in delinquency, it was not until we demonstrated this relationship locally that it became actionable evidence for Worcester stakeholders. Phase two, problem analysis, proved critical to the action research cycle.

Limitations

The data is limited to police documented incidents that occurred within the jurisdiction of the city in this study. This means that incidents of victimization, witness, or arrest outside of Worcester are unknown and cannot be factored into the regression analysis. Furthermore, traumatic incidents that individuals in this data set experienced to which police were not called are also not captured in this analysis. In addition, many individuals were missing information on race. This limits the ability of race to be analyzed; however, race was examined during problem identification. Finally, this is a data set of men with police contact, excluding all other males who may have been exposed to violence but who never had contact with the police in Worcester. These factors limit the findings of this study in terms of generalizability to other populations but not in terms of usefulness for informing the current patterns and practices in Worcester. In spite of these documented limitations, the results were compelling and revealed to local decision makers that not only was victimization associated with cycles of violence but also with incidents of trauma as a witness. This information was used to inform the strategy development for an intervention to break the cycle of violence in Worcester that included witnesses, which had traditionally been overlooked.

Conclusion

Youth and gang violence are major public health problems requiring comprehensive approaches that are responsive to local dynamics. “Real-time social science,” as part of an action research cycle that includes rigorous problem identification and analysis conducted collaboratively between practitioners and an academic research partner, can help communities overcome some of the implementation challenges inherent in comprehensive youth and gang violence reduction approaches. Although the analysis is collaborative, practitioners and academics play different roles in the process.

Law enforcement and other stakeholders suggest researchable questions that arise from their practice. They have access to needed data. They have practice-based hypotheses about problem causes and can provide insight into reasons why interventions may or may not work. The academic research partner has theory, content, methodological, and analytical expertise that they use collaboratively and iteratively to respond to and build on practitioner-partners’ knowledge and analysis. In the absence of a research partner, the problem analysis phase has traditionally been the weakest component of this action research cycle leading communities to adopt new or faddish programs without full consideration about the extent to which the program is relevant to local conditions and/or to continue to use the same strategies that have not worked (Boba, 2003). In this context, the Worcester, Massachusetts, case offers an example of how robust problem analysis can reveal new options and evidence to address the local drivers of serious youth and gang violence.

This type of “real time social science” is a form of community engaged scholarship that offers university faculty an opportunity to work in mutually beneficial partnership with practitioners and other community stakeholders to achieve practical, data-driven solutions to pressing social issues (Boba, 2003; Boyer, 1996; Burkhardt et al., 2017). As research partners transform data into actionable information, new partners can be mobilized to engage in the dialogue on youth violence. New partners bring different knowledge, strategies, and resources. By disrupting the traditional dialogue on youth and gang violence, we are also seeing the possibility of disrupting persistent patterns and generational cycles of this complex societal problem.

Footnotes

Acknowledgements

The authors acknowledge the Worcester Police Department’s Crime Analysis Unit for their assistance securing the incident data for this article, specifically Deputy Chief Sean Fleming and Principal Crime Analyst, Tiana Antul. The authors also thank Jim Gomes from Clark University’s Mosakowski Institute for Public Enterprise for committing Institute resources to this research and his help in reviewing drafts of this article. Finally, the authors thank the blind reviewers for their contributions that led to a significantly clearer and stronger article.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Berg

Stewart

Schreck

Simons

(2012). The victim-offender overlap in context: Examining the role of neighborhood street culture. Criminology, 50, 359-390.

Boba

(2003). Problem analysis in policing. Washington, DC: Police Foundation.

Boyer

(1996). The scholarship of engagement. Journal of Public Service & Outreach, 1(1), 11-20.

Braga

(2006). Homicide and serious gun violence in Worcester, Massachusetts (Unpublished report). Cambridge, MA: John F. Kennedy School of Government.

Braga

Hureau

(2012). Strategic problem analysis to guide comprehensive gang violence reduction strategies. In Gebo

Bond

(Eds.), Looking beyond suppression: Community strategies to reduce gang violence (pp. 129-152). Lexington, MA: Lexington Books.

Braga

McDevitt

Pierce

(2006). Understanding and preventing gang violence: Problem analysis and response development in Lowell, Massachusetts. Police Quarterly, 9, 20-46.

Burkhardt

B. C.

Akins

Sassaman

Jackson

Elwer

Lanfear

. . . Stevens

(2017). University researcher and law enforcement collaboration: Lessons from a study of justice-involved persons with suspected mental illness. International Journal of Offender Therapy and Comparative Criminology, 61, 508-525. doi:10.1177/0306624X15599393

Centers for Disease Control and Prevention. (2012). Data sheet. Retrieved from https://www.cdc.gov/violenceprevention/pdf/yv-datasheet-a.pdf

Dahlberg

Potter

(2001). Youth violence: Developmental pathways and prevention challenges. American Journal of Preventative Medicine, 20, 3-14.

10.

Davidson

McEwen

(2012). Social influences on neuroplasticity: Stress and interventions to promote well-being. Nature Neuroscience, 15, 689-695.

11.

Eckenrode

Laird

Doris

(1993). School performance and disciplinary problems among abused and neglected children. Developmental Psychology, 29, 53-62.

12.

Felitti

Anda

Nordenberg

Williamson

Spitz

Edwards

. . . Marks

(1998). Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults: The Adverse Childhood Experiences (ACE) Study. American Journal of Preventative Medicine, 14, 245-258.

13.

Fisher

Gunnar

Chamberlain

Reid

(2000). Preventive intervention for maltreated preschool children: Impact on children’s behavior, neuroendocrine activity and foster parent functioning. Journal of the American Academy of Child & Adolescent Psychiatry, 39, 1356-1364.

14.

Gebo

Bond

Campos

(2015). The OJJDP comprehensive gang strategy. In Decker

S. H.

Pyrooz

D. C.

(Eds.), The handbook of gangs (pp. 392-405). New York, NY: John Wiley.

15.

Herrenkohl

Hong

Klika

J. B.

Herrenkohl

Russo

M. J.

(2013). Developmental impacts of child abuse and neglect related to adult mental health, substance use, and physical health. Journal of Family Violence, 28, 191-199. doi:10.1007/s10896-012-9474-9

16.

Howard

Feigelman

Cross

Rachuba

(2002). The relationship among violence victimization, witnessing violence, and youth distress. Journal of Adolescent Health, 31, 455-462.

17.

Howell

J. C.

(2010). Gang prevention: An overview of research and programs (Bulletin). Washington, DC: Office of Juvenile Justice and Delinquency Prevention, Office of Justice Programs, U.S. Department of Justice.

18.

Howell

J. C.

(2012). Gangs in America’s communities. Thousand Oaks, CA: SAGE.

19.

Jennings

Piquero

Reingle

(2012). On the overlap between victimization and offending: A review of the literature. Aggression and Violent Behavior, 17, 16-26.

20.

Lauritsen

Sampson

Laub

(1991). The link between offending and victimization among adolescents. Criminology, 29, 265-292.

21.

Markey

Usher

Gruttadaro

Honberg

Cochran

(2011). Responding to youth with mental health needs: A CIT for youth implementation manual. Arlington, VA: National Alliance on Mental Illness.

22.

McGarrell

E. F.

Corsaro

Melde

Hipple

Cobbina

Bynum

Perez

(2012). An assessment of the comprehensive anti-gang initiative: Final project report. Washington, DC: U.S. Department of Justice.

23.

Office of Juvenile Justice and Delinquency Prevention. (2009). A guide to assessing your community’s youth gang problem. Washington, DC: Institute for Intergovernmental Research.

24.

Ousey

Wilcox

Fisher

(2011). Something old, something new: Revisiting competing hypothesis of the victimization-offending relationship among adolescents. Journal of Quantitative Criminology, 27, 53-84.

25.

Posick

Gould

L. A.

(2015). On the general relationship between victimization and offending: Examining cultural contingencies. Journal of Criminal Justice, 43, 195-204.

26.

Reuland

Draper

Norton

(2012). Statewide law enforcement/mental health efforts: Strategies to support and sustain local initiatives. Council of State Governments Justice Center. Retrieved from https://www.bja.gov/Publications/CSG_StatewideLEMH.pdf

27.

Rojek

Smith

Alpert

(2012). The prevalence and characteristics of police practitioner-researcher partnerships. Police Quarterly, 15, 241-261.

28.

Santos

R. B.

(2014). The effectiveness of crime analysis for crime reduction: Cure or diagnosis? Journal of Contemporary Criminal Justice, 30, 147-168.

29.

Shonkoff

Garner

(2012). The lifelong effects of early childhood adversity and toxic stress. Pediatrics, 129(1), e232-e246.

30.

Smith

Thornberry

(1995). The relationship between childhood maltreatment and adolescent involvement in delinquency. Criminology, 33, 451-481.

31.

Spergel

(1995). The youth gang problem: A community approach. New York, NY: Oxford University Press.

32.

Takata

Tyler

(1995). Community-university based approach to gang intervention and delinquency prevention: Racine’s innovative model for small cities. Journal of Gang Research, 2, 25-38.

33.

Widom

Maxfield

(2001). An update on the “cycle of violence” (National Institute of Justice Research Brief).Washington, DC: U.S. Department of Justice.

34.

Wood

Foy

D. W.

Layne

Pynoos

James

C. B.

(2002). An examination of the relationships between violence exposure, posttraumatic stress symptomatology, and delinquent activity: An “ecopathological” model of delinquent behavior among incarcerated adolescents. Journal of Aggression, Maltreatment & Trauma, 6(1), 127-147.

35.

Worcester Police Department (2015). Dataset. Not publicly available.

36.

Worden

McLean

Bonner

(2014). Research partners in criminal justice: Notes from Syracuse. Criminal Justice Studies, 27, 278-293.