november 2012 phoenix truce project final evaluation Final Evaluation of the Phoenix TRUCE Project: A Replication of Chicago CeaseFire Arizona State University Center for Violence Prevention and Community Safety Andrew M. Fox, M.S. David E. Choate, M.A. Charles M. Katz, Ph.D. Shirin Marvastian, M.A. Eric C. Hedberg, Ph.D. November 2012 Suggested citation: Fox, Andrew, David Choate, Charles Katz, Shirin Marvastian, and Eric Hedberg (2012). Final evaluation of the Phoenix TRUCE Project: A Replication of Chicago Ceasefire. Phoenix, AZ: Center for Violence Prevention and Community Safety, Arizona State University. About the Center for Violence Prevention & Community Safety Arizona State University, in order to deepen its commitment to the communities of Arizona and to society as a whole, has set a new standard for research universities, as modeled by the New American University. Accordingly, ASU is measured not by whom we exclude, but by whom we include. The University is pursuing research that considers the public good and is assuming a greater responsibility to our communities for economic, social, and cultural vitality. Social embeddedness – university-wide, interactive, and mutually-supportive partnerships with Arizona communities – is at the core of our development as a New American University. Toward the goal of social embeddedness, in response to the growing need of our communities to improve the public’s safety and well-being, in July 2005 ASU established the Center for Violence Prevention and Community Safety. The Center’s mission is to generate, share, and apply quality research and knowledge to create “best practice” standards. The Center evaluates policies and programs; analyzes and evaluates patterns and causes of violence; develops strategies and programs; develops a clearinghouse of research reports and “best practice” models; educates, trains, and provides technical assistance; and facilitates the development and construction of databases. For more information about the Center for Violence Prevention and Community Safety, please contact us using the information provided below. MAILING ADDRESS Center for Violence Prevention and Community Safety College of Public Programs Arizona State University Mail Code 3120 500 N. 3rd Street, Suite 200 Phoenix, Arizona 85004-2135 PHONE (602) 496 - 1470 WEB SITE http://cvpcs.asu.edu Acknowledgements Arizona State University’s Center for Violence Prevention & Community Safety would like to thank Hildy Saizow and Pam Goslar for their early efforts and cooperation with ASU staff in writing and submitting the proposal that resulted in the award from the Bureau of Justice Assistance (BJA). We also thank the leadership and staff at Chicanos Por La Causa, Inc. (CPLC) for their willingness to lead the Phoenix TRUCE Project, particularly Maria Parra and Renato Ramos. The support of the Phoenix Police Department (PPD), both as a TRUCE project partner, and as a data provider was instrumental to the project. From the PPD we would like to thank Commander Mike Kurtenbach, Lieutenant Sean Connolly, Connie Kostelac, Rene Brugman, Karen Kontac and the rest of the PPD Crime Analysis and Research Unit (CARU) who assisted with data. Additionally, we would like to thank the Arizona Department of Health Services who provided the Arizona State Trauma Registry data used in our analysis, particularly the generous support of Richard Porter and Dr. David Harden. In preparing the report we received valuable assistance from our research assistants Valerie Marin and Marisol Cortez, and photography assistance from Keri Katz. Finally, we thank all of the TRUCE outreach workers and violence interrupters, community members, and members of the faith-based community, without whose feedback, interviews, and survey participation, this report would have been much less useful. This evaluation was funded by Chicanos Por La Causa, Inc., through a grant from the Bureau of Justice Assistance (2009-SC-B9-0051). Opinions contained herein are those of the authors and do not represent the position of Chicanos Por La Causa, Inc., the Bureau of Justice Assistance, Office of Justice Programs, U.S. Department of Justice, or any of the partner agencies of the Phoenix TRUCE Project. This project was approved by the Institutional Review Board at Arizona State University (No. 100500). Center for Violence Prevention and Community Safety Staff Charles M. Katz, Ph.D. Watts Family Director & Professor Center for Violence Prevention & Community Safety School of Criminology & Criminal Justice Arizona State University David E. Choate, M.A. Associate Director of Operations Center for Violence Prevention & Community Safety Arizona State University Jody Arganbright, M.A. Business Operations Manager Center for Violence Prevention & Community Safety Arizona State University Shirin Marvastian, M.A. GIS Analyst Center for Violence Prevention & Community Safety Arizona State University Andrew M. Fox, M.S. Research Analyst Center for Violence Prevention & Community Safety Arizona State University © 2012 by the Arizona Board of Regents for and on behalf of Arizona State University and its Center for Violence Prevention and Community Safety. Center for Violence Prevention and Community Safety rd 500 N. 3 Street, NHI-1, Suite 200 Phoenix, AZ 85069-7100 (602) 496-1470 Web site: http://cvpcs.asu.edu This document may be copied and transmitted freely. No deletions, additions, or alterations of contents are permitted without the expressed written consent of the Center for Violence Prevention and Community Safety. TABLE OF CONTENTS EXECUTIVE SUMMARY .................................................................................................................... 1 THE CEASEFIRE MODEL ............................................................................................................... 2 PRIOR RESEARCH ON THE CEASEFIRE MODEL ............................................................................ 5 THE PHOENIX TRUCE PROJECT...................................................................................................... 12 OUTREACH WORKERS AND VIOLENCE INTERRUPTERS ............................................................ 12 CORE COMPONENTS OF THE PHOENIX TRUCE PROJECT .......................................................... 14 CHICANOS POR LA CAUSA AND THE TRUCE PARTNERS ........................................................... 17 PROJECT SETTING...................................................................................................................... 18 PROCESS EVALUATION .................................................................................................................. 21 PROCESS EVALUATION METHODOLOGY .................................................................................. 21 PROCESS EVALUATION FINDINGS ............................................................................................. 25 Core Component 1: Community Mobilization ...................................................................... 25 Core Component 2: Youth Outreach and Intervention ........................................................ 33 Core Component 3: Faith-Based Leader Involvement.......................................................... 61 Core Component 4: Public Education ................................................................................... 62 Core Component 5: Criminal Justice Participation ............................................................... 66 Other Impediments to Implementation ............................................................................... 70 IMPACT EVALUATION ................................................................................................................... 76 IMPACT EVALUATION METHODOLOGY .................................................................................... 76 ANALYTIC STRATEGY ................................................................................................................. 79 TRENDS IN POLICE INCIDENT REPORTS AND EMERGENCY DEPARTMENT DATA ..................... 81 TIME-SERIES ANALYSIS .............................................................................................................. 89 CONCLUSIONS AND RECOMMENDATIONS................................................................................... 96 REFERENCES ................................................................................................................................ 106 LIST OF APPENDICES ................................................................................................................... 108 EXECUTIVE SUMMARY INTRODUCTION The Phoenix TRUCE Project is modeled after the Chicago CeaseFire program, and as such has adopted a public health approach in responding to violence in the community. TRUCE emphasizes the use of outreach staff embedded in the community who identify community members who are at imminent risk of being either a victim or perpetrator of violence, particularly gun violence. The project is a data-driven, and its core components include, community mobilization and youth outreach. Each of its components addresses a different facet of the violence problem with the goal of preventing shootings. Both in the short term and the long term, CeaseFire’s purpose is to reduce the number of violent shootings in a community. The Phoenix TRUCE project was led by Chicanos Por La Causa, Inc. in partnership with Arizonans for Gun Safety, St. Joseph’s Hospital and Medical Center, the Phoenix Police Department, and Arizona State University’s Center for Violence Prevention and Community Safety. The CeaseFire model was implemented in Hermoso Park, a 1.5 square mile neighborhood located in South Phoenix. This report evaluates the implementation and impact of the TRUCE project from June 2010 to December 2011. Major findings are outlined below. PROGRAM IMPLEMENTATION  Outreach staff engaged in a substantial number (n=58) of conflict mediations. The disputants were typically gang-involved (70.7%), historically violent (63.8%), and young (91.4%), which are three criteria the model uses to define individuals as high-risk for gun violence involvement.       All of the clients who were recruited for Phoenix TRUCE (n=118) met at least four of the seven client selection criteria, indicating that the project abided by selection criteria established by national experts. Using eleven different forms of media and more than 11,000 individual items, TRUCE saturated the community with educational materials. Additionally, more than 25% of respondents who knew a shooting victim were knowledgeable about TRUCE. Advisory Board meetings were not held consistently, and the Board did not play a major role in establishing a strategic direction for the project. TRUCE did not establish a coordinated and collaborative relationship with the faithbased community. A Risk Reduction Plan was not completed for about two-thirds of program youth. Police regularly attended stakeholder meetings and provided liaisons to the project; however, a routinized process for information sharing was not developed. PROGRAM IMPACT     Time-series analysis indicated that program implementation corresponded to a decrease of more than 16 assaults on average per month, controlling for the comparison areas and the trends in the data. Time-series analysis indicated that program implementation corresponded to an increase of 3.2 shootings on average per month, controlling for the comparison areas and the trends in the data. The time-series analysis indicated that the more conflicts mediated and the more people involved in mediated conflicts, the greater the decreases in assaults in the target area. A number of the dosage effects were related to slight increases in shootings in the timeseries analysis, including number of clients, number of contacts (both home and on the street), and hours spent with clients. 1 INTRODUCTION In the mid-1990s, the University of Illinois School of Public Health, led by Gary Slutkin, established the Chicago CeaseFire program. From the start, the Chicago Ceasefire program relied on a public health model that focused on reducing retaliatory violence in high gun crime areas of Chicago through community mobilization, community awareness, outreach to youth, and intervention in violent conflicts between youth (Webster et al., 2009). The program sought to change the behavior of a small number of youth who were at high risk of being shot or of being a shooter in the near future (Skogan, Hartnett, Bump, and Dubois, 2008; 8-1). The highly publicized results of the project were remarkable. They showed that the number of shootings declined in several of the targeted neighborhoods by 16 to 35% and retaliatory gang homicides declined by 100% in each of five targeted areas (Skogan et al., 2008). Given the seriousness of the youth violence problem for local, national, and international communities (Pridemore, 2003), the Chicago CeaseFire program represents an important innovation in responding to youth violence (Chaskin, 2010). Reducing shootings by 16 to 35% and virtually eliminating retaliatory shootings in high-violence communities would not only have a substantial impact on the victims of violence through the reduction of injury and death, but would reduce fear of crime, emotional distress, and health care costs among all members of those communities. It would also result in increases in public services, property values, and overall quality of life for residents in high-violence neighborhoods (Krug, Dahlberg, Mercy, Zwi, and Lozano, 2002: 25). To date, however, the utility of the Chicago CeaseFire program for reducing youth violence is largely undetermined because there have been few 2 subsequent replications and evaluations. (For exceptions see Skogan et al., 2008; Webster et al., 2012a; Webster et al., 2012b; Wilson and Chermak, 2011). This report presents our findings on the Phoenix TRUCE project, a BJA funded Ceasefire replication site, which received technical assistance from Chicago CeaseFire project personnel. While there are a number of differences between Chicago CeaseFire and Phoenix TRUCE with respect to community (e.g., scope and nature of violence, neighborhood structure, community culture), program implementation (e.g., management, dosage), and evaluation (e.g., resources, available data), we believe that this will allow us to further understand the generalizability of the impact of the CeaseFire program on gun violence in a different context. The purpose of this report is to examine the impact of the Phoenix TRUCE project, which was implemented by Chicanos Por La Causa, Inc., one of the largest Hispanic Community Development organizations in the United States, on violence and shootings through a quasiexperimental longitudinal design. In the sections below, we present the Ceasefire Model, prior research on the model, and our methodology for evaluating the Phoenix TRUCE Project. THE CEASEFIRE MODEL Today, the Chicago CeaseFire program is a popular strategy for addressing gun violence. It has been implemented in several American cities, including Baltimore, Kansas City, New Orleans, New York City, Albany, Buffalo, Yonkers, Oakland, Philadelphia, Phoenix, and Columbus, as well as in several locations throughout the world, including Iraq, England, South Africa, and Trinidad and Tobago (ceasefirechicago.org, 2012). As research has shown, not everyone in a given community engages in violence. CeaseFire was designed to focus on those select few who are the most at risk for “being shot or being a shooter” in the near future. It also 3 attempts to change norms about violence among targeted individuals and neighborhoods by altering how they perceive the short- and long-term prospects of violence (Skogan et al., 2008). The Chicago CeaseFire model calls for the implementation of several core strategies for the purpose of having an impact on the decision-making of those who are involved in shootings, reducing the risk that they will engage in violence, and changing the norms toward violence in the community.. One of the primary elements of the CeaseFire model is identification and detection (Maguire, 2012). The model is centered around the concept of enhancing a community’s capacity to identify and target its high violence neighborhoods and individuals. Prior research suggests that a small number of people, places, and guns (Sherman and Rogan, 1995) are disproportionally responsible for gun violence. By targeting scarce resources on these focal points, implementers can direct interventions toward those people, places, and guns that need the greatest attention. Prior research has shown that programs and strategies that misdiagnose the problem are significantly less likely to impact violence (Katz and Choate, 2006). The CeaseFire model, therefore, emphasizes thorough analysis to identify those individuals and neighborhoods that are most at risk for imminent violence (Webster, Vernick, and Mendel, 2009). Another core component of the CeaseFire Model is community mobilization, which has been carried out through the involvement of faith-based leaders, community coalitions, and public campaigns against violence. Community mobilization efforts are largely concentrated on altering the norms and values of the community away from violence and toward the peaceful resolution of conflict. As such, community mobilization tactics often include the development 4 of community coalitions with the involvement of schools, churches, businesses, residents, and public health and criminal justice agencies. Together, these groups build community commitment toward reducing violence, address underlying structural causes of neighborhood violence, and coordinate anti-violence services that successfully change cultural and behavioral norms and increase community capacity to respond to violence (Skogan et al., 2008; Webster et al. 2012b). The most widely recognized aspect of mobilization efforts are typically found in their public education campaigns, which rely on flyers, posters, pamphlets, t-shirts, and other forms of media to promote nonviolence in the community. In addition to violence identification and community mobilization, the CeaseFire model emphasizes the role of outreach workers and violence interrupters. These individuals possess unique street knowledge and credibility. They are typically from the targeted neighborhood and are often former gang members and drug dealers who were involved in serious criminality and violence. Outreach workers hold more complex and personalized relationships with clients. They are largely responsible for case management, conflict mediation, and mentoring. They help clients find jobs, counsel them on alternative methods for dealing with conflict, and help them address underlying risk factors for violence. Outreach workers work with gangs, subgroups within gangs, and other violent youth to help them transform their values and norms (Skogan et al., 2008; Webster et al., 2012b). Violence interrupters are more specialized and are largely responsible for identifying and responding to retaliatory violence before it intensifies (Ritter, 2009). They typically possess more legitimacy among street youth than outreach workers, often because the violence interrupters were more criminally involved and have a working knowledge of who is currently 5 involved in violence. While outreach workers are typically full-time employees of the program, violence interrupters may be volunteers, or if they are employed by the program, it is on a parttime basis with minimal pay (Skogan et al. 2008). PRIOR RESEARCH ON THE CEASEFIRE MODEL The Phoenix TRUCE Project was modeled after Chicago CeaseFire, which has been replicated in a number of different cities in the United States and internationally. However, there have been only a few formal evaluations of the program. Outside of Chicago, for example, only three sites, Newark, Baltimore, and Pittsburg, have been formally evaluated. As will be discussed below, some have argued that there were substantial implementation differences between Chicago CeaseFire and the programs replicated in Newark and Pittsburg, and that findings attributable to the latter programs should not necessarily be generalized to the CeaseFire model (Kane, 2012). Below, we review the results of the four prior evaluations, discussing their results and implications for understanding the effectiveness of CeaseFire. Chicago CeaseFire. Chicago Ceasefire began in 1999 and eventually was implemented in 27 target areas. Skogan et al.’s 2008 evaluation reported that outreach workers played a key role in the program. The outreach workers managed caseloads of high-risk individuals who they determined were the most likely to shoot or to be shot at. The impact evaluation of Chicago CeaseFire relied on 16 years of data to compare seven targeted areas with seven matched comparison areas with respect to trends in violence, specifically shootings and killings (Skogan et al., 2008). Given the complexity of the project - multiple sites, city-wide trends - time-series analyses were employed to examine its impact. Additionally, the authors used social network analysis to examine whether gang homicide networks weakened over time. 6 The authors reported that in four sites CeaseFire was associated with “distinct and statistically significant declines” in actual and attempted shootings. Skogan et al. reported that the declines in these areas ranged from 16 to 35%; six of the sites reportedly grew “noticeably safer” as a result of the CeaseFire program. Increased safety was measured by decreases in size and intensity of shooting hotspots; in four of the six sites the decreases, according to the evaluators, were directly tied to the introduction of CeaseFire. Additionally, using social network analysis, the authors reported that in two sites gang homicide density decreased, in four sites retaliatory gang killings decreased, and in three sites the average gang involvement in homicide decreased. The evaluators concluded that the program had the strong potential to reduce shootings and killings (Skogan et al., 2008). Maguire (2012), however, suggested that those results were not as straightforward as they might have appeared. Through a lengthy post-hoc analysis of the evaluation, Maguire points out that it focused on three outcome measures in seven zones for a total of 21 outcome measures. Through a re-presentation of these results, Maguire illustrated that “among the 21 sets of outcomes, 12 favor the comparison areas…, 8 favor the treatment areas…, and 1 favors neither...” (p. 8-10). Maguire also noted that the social network analysis examining reciprocal shootings was similarly interpreted in an optimistic fashion. Among the eight sites where retaliatory shootings were examined, retaliatory shootings declined by half. In the other half, no demonstrable change was exhibited. It is unclear what factors might have resulted in a decline in reciprocal shootings in half the areas, but not in the others. For example, Project Safe Neighborhoods (PSN) was carried out at the same time and in 50% of the same locations as Chicago CeaseFire (Papachristos, Meares, 7 & Fagan, 2007). In fact, the authors concluded, “after controlling for the social, demographic, and PSN factors, no statistically significant effect in the declining homicide rates during the observation period can be attributable purely to the presence of Operation Ceasefire in the PSN treatment areas” (Papachristos et al., 2007: 264). These findings when taken together suggest that, contrary to popular belief, the CeaseFire program may not have been as effective in reducing violence as first promoted. Operation Ceasefire Newark. Ceasefire Newark was one of the first near-replication sites in the nation. Because of the high rates of gun violence in Newark, stakeholders were committed to implementing two promising strategies at the time: Boston’s Operation Ceasefire and Chicago CeaseFire. These were to be implemented in Newark as Operation Ceasefire, a hybrid of the Boston and Chicago programs, for the purpose of reducing gun violence in Newark’s most violent neighborhoods. In 2004, a working group examined available police data and identified a two-squaremile area, later referred to as the Ceasefire Zone (CF Zone),where Operation Ceasefire would later take place. The CF Zone was characterized as having higher rates of gun violence than other neighborhoods in Newark (Boyle, Lanterman, Pascarella, and Cheng, 2010). Operation Ceasefire formally began in May 2005. The program included five elements of Chicago CeaseFire: public education, community mobilization, youth outreach, faith-based leader involvement, and criminal justice participation. Newark CeaseFire differed from its predecessor in a few important ways. First, it did not make use of violence interrupters. Second, outreach workers did not employ a case management approach to their clients. Third, as in Chicago, outreach workers might be ex-offenders, but they might also be church 8 congregants (Boyle et al., 2010: 107). As in Boston, the police played a strong role in Newark CeaseFire. Shooting teams were responsible for the “aggressive investigation of shootings” in the CF Zone. Police were also responsible for notifying CeaseFire partner organizations of shootings for the purpose of mobilizing resources, and for working with parole officers to monitor parolees in the CF Zone. While Boston Ceasefire emphasized the importance of “pulling levers” and collective accountability, the Newark Ceasefire project did not employ this strategy, largely because Newark gangs and groups did not possess the organizational structure or cohesiveness that would warrant such an approach (Boyle et al., 2010). Boyle et al. (2010) examined the effectiveness of Operation Ceasefire through the analysis of admissions to a University Level 1 trauma center. Specifically, they examined the number of gunshot wounds in the CF Zone and in a comparison area over a 3-year period from January 1, 2004 through December 31, 2006. The comparison area was matched and subsequently selected based on the number of gunshot wounds and on census data reflecting resident ethnicity, age, household income, poverty, and housing vacancies. Time-series analysis illustrated that Operation Ceasefire had no significant impact on the number of gunshot wounds. Some advocates of the Chicago Ceasefire model, however, maintain that these findings are not reflective of the Chicago Ceasefire model, given the lack of implementation fidelity delivered in Newark (Kane, 2012). Baltimore’s Safe Streets Program. In 2007, the U.S. Department of Justice funded the Baltimore’s Safe Streets Program to implement the Chicago CeaseFire program in Baltimore. From the start, the only programmatic difference between Chicago and Baltimore was to be the use of violence interrupters. Violence interrupters were not employed as part of the Baltimore 9 project.Instead, outreach workers were responsible for conflict mediation as well as for intervening in the lives of at-risk youth. Over time, the program grew from one targeted community (McElderry Park) to five targeted communities (McElderry Park, Union Square, Ellwood Park, Madison-East-End, and Cherry Hills) (Webster et al., 2009; 2012a; 2012b). Daniel Webster and a team of public health scholars from Johns Hopkins served as the evaluation team. They produced a number of manuscripts documenting program fidelity and effectiveness in the Baltimore sites. The first such public document was an interim evaluation report based on data through October 2008, which focused only on the first three sites: McElderry Park, Union Square, and Ellwood Park. The authors reported mixed findings. For example, in McElderry Park homicides declined (particularly among those under age 30), but nonfatal shootings increased, particularly when contrasted with the comparison areas. In Union Square homicides and shootings actually increased in the target area. In Elwood Park homicides and shootings remained stable with the exception of nonfatal shootings, which declined in the target area (Webster et al., 2009). In 2012, the evaluators released the final report that examined the impact of the program in four sites—McElderry Park, Elwood Park, Madison-Eastend, and Cherry Hill—relying on data through December 2010. This time the results were more impressive, with three of four neighborhoods reporting a significant decline in homicides and nonfatal shootings. The evaluators claimed that over 112 months, the program in the four neighborhoods had prevented five homicides and 35 nonfatal shootings. Following the public success of the program, in late 2011, the U.S. Department of Justice provided an additional $2.2 million in 10 funding to support the Baltimore Safe Streets Program (Webster et al., 2012a; Webster et al., 2012b). These findings should be interpreted with caution, however. Three of the five program implementation sites (Ellwood Park, Union Square, and Madison-Eastend) were shut down shortly after being established. Several local news stories focusing on the program revealed that the Union Square site was closed after one year as a consequence of the city terminating its contract with the local NGO that managed the site. The city learned through local and federal law enforcement that a local street gang (the Black Guerilla Family ) had infiltrated the site. Gang members, one a gang leader, were working for the Union Hills Safe Streets site as outreach workers for the purpose of obtaining cover for their gang’s heroin distribution network (Kelly, 2010; Fenton, 2012). Two other sites—Ellwood Park and Madison-Eastend-closed in 2010, but little information about the reason for their closures has been revealed in the media or interim, final, or peer reviewed manuscripts (see Webster et al., 2009; Webster et al., 2012a; Webster et al., 2012b). These closures suggest that implementation and/or the sustainability of such complex programs might be difficult. Pittsburg’s One Vision One Life. Pittsburg’s One Vision One Life (otherwise known as One Vision) program was established as a consequence of several grassroots organizations joining to seek an innovative evidence-based response to violence in Allegheny County. Community leaders began by diagnosing the County’s violence problem and conducting research on best practices. Following consultation with partners and the police, the Pittsburg program was “modeled after (but [did] not mirror)” Chicago and Baltimore CeaseFire programs (Wilson and Chermak, 2011: 995). One Vision staff visited Chicago in 2004 and 2005 for the 11 purpose of replicating Chicago Ceasefire’s programming and data collection in Pittsburg (Wilson and Chermak, 2011). The One Vision program adopted a six-point plan to reduce shootings. The plan consisted of (1) mediating and intervening in conflicts, (2) providing outreach to at-risk youth, (3) building strong community coalitions, (4) publicizing a no-shooting message, (5) rapidly responding to violence in target areas, and (6) developing programming for at-risk youth (One Vision One Life, no date). The One Vision program was more similar to Baltimore than Chicago in its use of outreach workers. Community coordinators served as generalists, addressing a variety of community needs, and were responsible for conflict intervention, mediation, and other outreach work with at-risk youth. Others have also noted that One Vision was different from Chicago CeaseFire with respect to police participation in the project. Specifically, in Pittsburg, community coordinators rarely conferred with the police, and when they did, the experience was often negative (Maguire, 2012). A rigorous evaluation of the One Vision program was conducted by Wilson et al. (2011). The evaluators relied on a quasi-experimental design comparing three target areas to matched comparison areas. Wilson et al. found that the onset of the program either increased or did not affect homicides. Further, they reported that the program increased aggravated and gun assaults in the target areas (Wilson, Chermak, and McGarrell, 2011). Wilson and colleagues (2011) caution that the implementation of the Chicago CeaseFire model has varied by site and that researchers do not yet know which components, or aspects of each component, have the most impact on gun violence. For example, while outreach workers in Pittsburgh engaged in more community mobilization than outreach workers in other sites, it 12 is not clear whether or not the activity of the outreach worker plays a unique role in increasing or decreasing shootings. Thus, Wilson et al. recommended that future evaluations should continue to examine how the different levels of implementation are related to the impact of the program. THE PHOENIX TRUCE PROJECT The Phoenix TRUCE Project was modeled after the Chicago CeaseFire program, and as such, had adopted a public health approach in responding to violence in the community. TRUCE emphasizes the use of outreach staff embedded in the community, who identify community members who are at imminent risk of being either a victim or a perpetrator of violence, particularly of gun violence. By focusing attention on the gun violence, TRUCE seeks first and foremost to diminish shootings. Other beneficial potential reductions in crime and community improvement are expected, but are not necessarily the primary focus of the TRUCE effort. OUTREACH WORKERS AND VIOLENCE INTERRUPTERS TRUCE uses carefully selected outreach workers and “violence interrupters” with ties to the local community. TRUCE staff must be familiar with the neighborhoods where they work, know the people who live and frequent the area, and have sufficient credibility with those people to establish a dialogue. The outreach workers recruit members of the community who are identified as being at risk for gun violence, both potential victims and perpetrators. Outreach workers must foster trust within the community, especially with those who would be clients. By developing a relationship with the community, outreach workers become increasingly familiar with those individuals best suited to be recruited as clients. 13 Outreach workers use a set of seven criteria to guide client selection, assessing each potential client’s appropriateness for inclusion in the TRUCE caseload. Individuals meeting four or more of the seven criteria are considered high risk and good candidates for recruitment. The seven at-risk criteria are: 1. Gang involvement – Current members or associates are at greatest risk. 2. Key role in a gang – Highly invested gang members are more likely to be involved in more serious crimes, and more often. 3. Prior criminal history – Individuals with a history of serious offending, particularly for gun crimes, are at high risk. 4. Recently released from prison – Related to prior criminal history, the time immediately following re-entry is critical. 5. High-risk street activity – Involvement in drug sales and/or other crimes. 6. Recent victim of a shooting – Research supports a strong link between victimization and offending. Today’s victim, tomorrow’s offender – makes these individuals prime candidates for TRUCE clients. 7. Young – Principally between the ages of 16 and 25 years old. Outreach workers serve as mentors, counselors, and case managers for their clients. As outreach workers become familiar with their clients, identifying their individual needs and challenges, they connect them with services and opportunities that could mitigate their risk for gun violence. Violence interrupters work to mediate conflicts that may lead to or stem from gun violence. Going into the community immediately following a shooting, working with those that may seek retaliation, violence interrupters seek to intervene and prevent retaliatory shootings. The violence interrupter’s job requires a special set of skills. Often mediation takes place between members of rival gangs. Violence interrupters, like outreach workers, must be an 14 integral part of the local community. Knowing the people, places, and events that lead to and surround a shooting is essential to the mediation process. Violence interrupters work the streets following a shooting, speaking to residents, friends and relatives of the victim and perpetrator(s), gang members, faith leaders, and others with something to say about the shooting. Distributing public education materials, talking with people, and coordinating with outreach workers, violence interrupters are a critical component of the effort to reduce shootings. CORE COMPONENTS OF THE PHOENIX TRUCE PROJECT Replicating Skogan et al.’s (2008) work, the Phoenix TRUCE project was designed as a replication of the Chicago CeaseFire project. As a result, interventions carried out by outreach workers and violence interrupters were to be guided by theories derived from public health strategies and responses. The TRUCE clients should be recruited in accord with their at-risk status because of their diverse and deeply-rooted needs. The outreach workers were to customize services for each client, directing them to programs that addressed their particular needs, including drug and alcohol treatment, housing, vocational or job skills training, employment, literacy and GED education, prenatal and parenting assistance, food assistance, family conflict resolution, help leaving a gang, and emotional or mental health treatment. Taking a holistic approach to the individual’s well-being by addressing immediate concerns should lead to overall well-being and a reduction in risk for gun violence. The broader activities of outreach workers, violence interrupters, and other TRUCE personnel should be focused on public education and community mobilization. Public education and community mobilization efforts are essential to TRUCE, both for its near-term 15 effectiveness and its long-term sustainability. The public education component at its most basic level would make the public aware of TRUCE, its mission, the outreach workers, and how the community can become involved as a partner in reducing shootings in their neighborhoods. Among other things, the public education component would include providing educational materials and other takeaways that TRUCE staff could distribute, giving TRUCE a public identity, lending credibility and support to the efforts of the outreach workers and violence interrupters.. Community mobilization focuses on the inclusion of the community as a whole. While the efforts of outreach workers should be focused on those community members at greatest risk for gun violence, the support of the entire community is vital to making their efforts successful and sustainable. Ideally, TRUCE would encourage community mobilization by informing and interacting with residents, the faith-based community, service organizations(both public and private), local area businesses, and other interested partners. Staging public displays – peace marches following shootings, midnight barbeques to provide youth with a safe alternative to being on the streets late at night, and other community reinvestment activities - would help establish rapport between TRUCE and the community, which in turn could open the way to begin changing the cycle of violence. Communities where gun violence is most prevalent need to change community norms related to violence. Improving the understanding of social and personal costs of violence, particularly gun violence, and making people aware of consequences and alternatives, can have a profound impact on changing those norms.. This kind of outreach can be particularly effective when involving youth and the faith-based community. 16 Additionally, outreach workers and other TRUCE personnel should work with youth in the targeted neighborhoods, both in schools and in public areas. Outreach workers meet youth in the community – at parks, on the street, in stores – wherever youth gather. Contacting them in their community where they live and play allows TRUCE workers to build rapport with youth not yet at risk, with the goal of diverting them from becoming the next shooters or shooting victims.. The faith-based community is an effective avenue for community outreach, embeddedness, and mobilization. Churches and similar institutions are safe havens for those looking to avoid violence - a safe place where honest community dialogue can occur, mediation meetings can be held after a shooting, or people can gather for a peace march following a killing. Mobilizing the community from within would be an important component for TRUCE’s immediate and lasting success. The Chicago CeaseFire model originated without a hospital component, but later on, hospitals as partners became an integral part of the concept, serving a number of basic programmatic functions. The most pivotal for TRUCE was use of St. Joseph’s Hospital’s emergency department as an alternative center for recruiting potential clients. The emergency department received many gunshot victims arriving from the target neighborhood, due to its proximity and its level 1 trauma facility status. Additionally, the hospital served as a community outreach partner and contributed to the public education component.. Phoenix TRUCE included Saint Joseph’s Hospital and Medical Center in its program implementation from the outset. Representation from St. Joseph’s was consistent and committed throughout implementation. The original TRUCE implementation plan called for an 17 employee of the hospital to double as an outreach worker in the emergency department of the hospital. They were to serve as a liaison between the project and shooting victims and their families. (The implementation challenge and results of that decision are discussed later in the section presenting program impact findings.) CHICANOS POR LA CAUSA AND THE TRUCE PARTNERS The fiscal agent and lead of the Phoenix TRUCE project is Chicanos Por La Causa, Inc. (CPLC). CPLC is a statewide organization with 800 employees, 40 years of experience, and the resources needed to develop, implement, sustain, and manage programs and services. CPLC has 26 years of experience in community-based prevention. It is considered by many in the state to be the premier nongovernmental organization serving the Hispanic community. CPLC’s Director of Prevention is responsible for fiscal and contract management. The Program Manager oversees implementation and supervises project staff. The Prevention Services Coordinator is responsible for community partnerships and coordinating prevention efforts among partners. Outreach workers serve as mentors and counselors to youth at risk of being shot, connecting them with a broad range of services. Violence interrupters mediate conflicts between gangs and intervene during conflicts in order to stem the retaliatory violence that can threaten to break out following a shooting. CPLC received technical assistance from the Chicago Project for Violence Prevention (CPVP), which was responsible for the creation of the Chicago CeaseFire model. CPVP provided street outreach and violence interruption skills training to the Phoenix TRUCE team, as well as lending ongoing violence prevention program and planning support.. 18 CPLC works closely with local organizations to accomplish its project goals. Arizonans for Gun Safety (AzGS), a well-known community-based nonprofit organization, oversees the public education component of Phoenix TRUCE. St. Joseph’s Hospital and Medical Center provides support and advocacy for youth involved in violence to reduce recidivism related to violent injury and retaliation. The Phoenix Police Department provides advice, information and data about the nature of the crime problem. Last, Arizona State University’s Center for Violence Prevention and Community Safety served as the evaluation partner for the TRUCE project. PROJECT SETTING The Phoenix TRUCE target area is located in the South Mountain area, a cluster of neighborhoods on the south side of Phoenix, Arizona. The area is largely comprised of some of the city’s oldest homes and businesses, but also includes large swaths of new homes where farms have been and pockets of recently gentrified developments. The nature of deeply rooted inter-generational gangs, dilapidated properties, and an emerging and shifting neighborhood composition makes for a community in stressful transition. Efforts are being made to improve the quality of life for people who live in the area, but like many inner-city communities, South Mountain is challenged by high crime rates, a struggling economy, and urban decay. Most alarming is the disproportionate use of firearms to commit crime. The Arizona Department of Health Services (AZDHS) reported that in 2007, in the South Mountain area 80% of firearm-related deaths were homicides compared to 35% in the remainder of the state. Related, the South Mountain Police Precinct experienced a large increase in violent crime in 2007. During the first 11 months of 2007, 62 homicides were committed, a 41% increase over the previous year. Crime analysis revealed that a large amount 19 of the violence was occurring in four neighborhoods: 7-11 Fight Back Neighborhood, Lindo Park Neighborhood, South Vista Housing Development, and Hermoso Park Neighborhood. Comparing the number of Part One violent crimes in these areas to the rest of Phoenix for 2007 and 2008 showed significantly higher homicides, robberies, sexual assaults, and aggravated assaults in these neighborhoods in both years. Initially, Phoenix TRUCE sought implementation in all four neighborhoods. The Chicago CeaseFire Technical Assistance team argued that including all of them would broaden the geographic scope of the target area beyond the capacity of the available resources. Phoenix and Chicago staff debated the matter at length until the Chicago team declared that if the size of the target area was not reduced, they would no longer serve as technical advisors. This ended the debate, and Hermoso Park, determined to have the highest level of violence, was selected as the target area. As seen in Exhibit 1, the Hermoso Park neighborhood is an area bounded by Broadway Road to the north, Vineyard Road to the south, 16th Street to the west, and 24th Street to the east, making it approximately 1.5 by 1 mile in size. Named for the city park located at the heart of the area on the northeast corner of Southern Road and 20 th Street, Hermoso Park is a predominantly residential area of older homes. 20 Exhibit 1: TRUCE Target Area 21 PROCESS EVALUATION PROCESS EVALUATION METHODOLOGY A major component of any comprehensive evaluation is an examination of the process of programmatic implementation. Measuring the content, dosage, and procedures of specific program interventions and activities is essential in order to be assured that the intervention has been fully carried out - that is, that the neighborhood received the intended treatment. Without the certainty that activities have been fully implemented, it cannot be determined whether any success or failure of the program can be attributed to the programmatic activities. Describing and measuring an intervention confirms whether programmatic activities are being administered as intended, in a systematic and standardized manner. Unstandardized administration can result in an intervention having differing effects for different groups, making replication very difficult. It was therefore important to measure the extent to which TRUCE clients were receiving standardized “dosages.” In conducting the process evaluation of the Phoenix TRUCE project, we relied on data pertaining to the implementation process and specific intervention activities. The evaluation was organized around the five core components of the Chicago CeaseFire model. By organizing it in this way, we were able to examine how well each component was implemented and determine whether the objectives explicitly outlined in the original project proposal were met. The six objectives of the Phoenix TRUCE project were as follows: Objective 1: Provide outreach and intervention services to high-risk individuals. Objective 2: Mobilize the community to build a base of support for CeaseFire initiatives. 22 Objective 3: Educate the community to facilitate behavior change and promote nonviolence. Objective 4: Use police data on homicides and shootings to develop immediate responses to violence. Objective 5: Coordinate with St. Joseph Hospital’s SMARTR program to provide outreach services to patients with violence-related injuries. Objective 6: Coordinate with collaborative partners and advisory board members to ensure efficient and effective implementation. Exhibit 2 below clarifies how the proposed Phoenix TRUCE project objectives correspond with the five core components of the Chicago CeaseFire model. Exhibit 2: Relationship of Chicago CeaseFire Core Components and Phoenix TRUCE Objectives Chicago CeaseFire Model Core Components Phoenix TRUCE Objectives 1. Community mobilization Objective 2; Objective 6 2. Youth outreach and intervention Objective 1; Objective 5 3. Faith-based leader involvement Objective 6 4. Public education Objective 3 5. Criminal justice participation Objective 4 The data used to evaluate the project’s implementation were collected from a number of sources, including interviews of TRUCE staff and other stakeholders, a randomized in-person household survey, and the Chicago Project for Violence Prevention Evaluation Database. First, as part of the process evaluation, semi-structured interviews were conducted regarding the implementation process and programmatic activities. In all, 12 semi-structured interviews were conducted with TRUCE staff members and project stakeholders. Interviews took place between 23 June 2011 and October 2011 and ranged from 45 to 60 minutes long.. Where applicable, interviews focused on such issues as:  What is the problem with violence in the community? Is it a problem?  Is there an area, place, or people who are at the root of the problem?  Are there other, more important problems?  What responses, if any, have been tried in the past? Were they successful?  What have you learned thus far about the problem and the intended TRUCE response?  What short-term successes have you seen? Failures?  What’s working? What isn’t working?  What intervention strategies were selected?  What rules or guidelines were established to select outreach workers and violence interrupters?  What initial training was provided? What about ongoing training?  What activities were first implemented?  Did things go as planned? If not, what obstacles emerged and how were they handled?  What was the dosage (e.g., number and type of mentoring contacts, services needed, and services provided) provided to clients?  If there were gaps, what were they and how were they addressed? Interviews were focused on the above key issues, but were conducted as an open dialogue. They were recorded with the participant’s consent and transcribed. The qualitative analysis file of the interviews was 150 single-spaced typed pages of text in MS Word. In November 2011 a randomized in-person household survey was conducted, in English and Spanish, within the TRUCE target area. Questions were asked about respondents’ perceptions of crime and safety in the neighborhood, their experiences with shootings, and the extent of their knowledge of the TRUCE program. We used U.S. Postal Service data to 24 determine the number and location of households in the TRUCE target area. These data indicated that the TRUCE target area was comprised of 3,220 residential addresses. The data were geo-coded and we randomly selected 165 (about 5%) households to be approached for an in-person survey. A total of 123 households were approached; however, many residents (n= 49) were not home or did not answer the door, and some (n=8) refused to participate in the survey. The final response rate was 53.6% (n=67). (See Appendix A for a copy of the household survey.) Data were extracted from the Chicago Project for Violence Prevention Evaluation Database Evaluation. The database is an online system developed for use by Chicago CeaseFire model sites (such as Phoenix) for use in recording and evaluating important components of CeaseFire sites. The system integrates recording and reporting tools for participant intake, shootings, violence interrupter logs, conflict mediations, and community activities. This detailed level of data for intervention activities provided a quantitative base for evaluating the core components. Last, it should be noted that ASU personnel regularly attended project functions and activities. For example, ASU personnel attended the Phoenix TRUCE Partners’ Meetings, at which attendees served as the steering committee for the project. Partner’s meetings were generally held monthly. Meetings began in September 2009 following notification of the award, and have been ongoing ever since. In addition to the Partner’s Meetings, ASU attended all of the planning and other meetings of the Advisory Board, and attended several community garden events, three community forum events, one of the peace marches, one community barbeque, and the mural painting event. Additionally, multiple tours of the target community, comparison areas, and other neighborhoods of the South Phoenix region were conducted. 25 PROCESS EVALUATION FINDINGS Core Component 1: Community Mobilization The first core component of the CeaseFire model is Community Mobilization. Objectives 2 and 6 of the Phoenix TRUCE project address this component. Objective 2 is to “mobilize the community to build a base of support for TRUCE initiatives” and Objective 6 is to “coordinate with collaborative partners and the advisory board to ensure efficient and effective implementation.” Activating a broad range of community members and stakeholders is an important part of the program and an essential element in changing the norms of a community. Community Events The Chicago CeaseFire model calls for significant integration of the program into the targeted community. The Community Mobilization component relies on a variety of activities to market the project and its purpose, disseminate public education material, and coordinate the community response to violence. The primary focus of the community mobilization effort is to enlist the support of local businesses, social service agencies and organizations, members of the faith-based organizations serving the neighborhood, and community members at large. The support of these groups helps create a safer community, through improved community efficacy and informal social control. Part of the mobilization effort should be to build coalitions and cooperation where they do not exist, and to strengthen those relationships where they do. Mobilizing the community should invigorate and motivate the community to be supportive of the violence interrupters and outreach workers and their combined efforts. Improvements to broad-based community support, efficacy, and social control is intended to yield normative changes in attitudes and behaviors favorable to violence in general, and gun 26 violence in particular. The most basic approach to fostering pro-social community-level relationships is comprised of the organization of community-level events and activities and the efforts of individual violence interrupters and outreach workers. The TRUCE Project Manager and the Outreach Supervisor organized most of the community activities, but violence interrupters and outreach workers contributed to the coordination of the events. As seen below in Exhibit 3, during the first four months of programmatic activity, TRUCE conducted between three and six community events each month. During October 2010, TRUCE conducted 15 community events. Between October 2010 and August 2011, TRUCE held between 12 and 18 community activities each month. From September 2011 through December 2011, there were 14 community events held. During the 19 months included in the analysis, a total 188 community activities were recorded by TRUCE staff. 27 Exhibit 3: Number of Community Events by Month 20 18 16 14 12 10 8 6 4 2 0 About half of the community activities organized (n=98, 52.1%) revolved around community gardens (Exhibit 4). Most community gardening events and activities took place in a lot adjacent to the Tanner Gardens Apartments; others were held at Southminster Presbyterian Church, which also serves the target neighborhood. In addition to community-wide gardening, TRUCE sponsored special events such as inter-generational gardening days, youth gardening days, and client gardening days. The community gardens provided a venue for community members to gather in safety and cooperation. In addition, garden produce could be sold at local farmers’ markets, allowing participants, including youths, to earn a small profit from their work. TRUCE gardening events often included a barbeque and sometimes music. One of the TRUCE 28 violence interrupters was the regular barbecue chef, and he would often take this opportunity to talk with visitors about TRUCE and nonviolent alternatives to conflict resolution. One of the stakeholders summed up what the community garden meant to the community this way: …the garden is a neutral area, no matter what gang you’re in or where you live, you can come and eat and pick and work, or whatever. Establishing a location regarded as neutral by gangs created not only a space where participants who were gang-involved could be engaged in the community without fear of violence, but also a place where the community could feel safe from warring gang factions. Canvassing the neighborhood, talking to residents about TRUCE and its mission, was also a common community activity (n=29, 15.4%). Canvasses served a dual purpose. They provided an opportunity to spread the word about TRUCE to the community and to distribute public education materials, and they also provided a means of connecting with members of the community who might know potential clients or be potential clients themselves. There was no record of whether or not any clients were recruited through this means of initial contact, but given the anecdotal descriptions of how violence interrupters and outreach workers went about recruiting, it is likely this was one means of doing so. At a minimum, it expanded the community’s knowledge of TRUCE and the TRUCE staff. Peace marches were an important activity for Chicago CeaseFire sites that implemented community mobilization. The peace marches were intended to demonstrate a community-wide outcry against violence; they were organized shortly following a shooting. . Phoenix TRUCE staged two peace marches during the implementation period. One was organized within a week of a shooting, and was planned and executed with close consultation with the stakeholder group, particularly the police. Peace marches addressed a number of core components; 29 specifically, the November 2010 peace march was organized by Arizonans for Gun Safety and included numerous public education components. Along with the peace march, a pre-event press conference was held, door hangers were placed on every door in the target area, and posters and flyers were distributed throughout the community. Additionally, the march included speakers, one of whom was the lead violence interrupter for TRUCE. More than 220 community members and faith-based leaders were in attendance, along with nine TRUCE staff and nine police officers. Hundreds of public education materials and TRUCE logo takeaways (n=357) were distributed. TRUCE also engaged in one other peace march following a shooting. This one included more than two dozen community members and youth, all of whom afterward held a candlelight vigil at the site of the shooting. Similarly, TRUCE organized another candlelight vigil without a march following another shooting incident. (See Exhibit 4.) Throughout the implementation period, TRUCE held 15 (8%) community barbeques, primarily midnight barbeques. The barbeques were an entertaining community gathering that offered at-risk youth a safe alternative to being out on the streets late at night.. The barbeques were generally well-attended, typically with 50-100 attendees, and never had a reported incident of violence. Other entertainment events were also held (n=6, 3.2%), including baseball games and a hip-hop summit. These events, like the midnight barbeques, sought to provide a safe alternative to being on the streets at night. (SeeExhibit 4.) Community association and education events were conducted as separate events (n=9, 4.8%). These included meetings with groups of community members and other gatherings suitable for educating the community about TRUCE and its mission. There were also other types of community based service oriented events (n=10, 5.3%), including an AIDS Walk, community 30 center discussion groups, an event providing Christmas gifts to families in need, a Women’s Domestic Violence Health Fair, neighborhood planning meetings, and youth events. Events like these were instrumental in gaining the support of additional community partners. Engaging in these community-building activities demonstrated that TRUCE was making an effort to expand the program’s presence and its reach in coalition building. (See Exhibit 4.) Community activities also focused attention directly on program participants through discussion groups, group activities, and outings (n=19, 10.1%). These events and activities were used as a mechanism for connecting clients with the community more pro-socially, but also to integrate the clients with one another, effectively as a form of support group. By fostering a sense of belonging among the clients, as they struggled, they knew that others with similar backgrounds were also trying to avoid violence. The public education component of CeaseFire programs is an important part of the project’s implementation in both measurable and immeasurable ways. TRUCE disseminated public education materials were in variety of ways at community events. The model calls for items such as buttons, flyers, posters, bracelets, dog-tags, and other swag to be regularly distributed at community events, and TRUCE did a good job of implementing this task. During the 188 community activities documented, nearly 6,000 public education items were disseminated. One event that we would like to draw particular attention to was the TRUCE Mural Project.1 This project satisfied at least three of the core components: public education, community mobilization, and faith-based leader involvement. The building that was painted 1 The building used for the TRUCE Mural Project was donated by the Progressive Church of God and Christ and Pastor Victor Rushing. 31 was owned by a local church, and community residents helped to design and paint the mural. In all, more than 100 residents contributed to the design process and attended the Community Paint Day. At the event, community members, TRUCE outreach workers, and faith-based leaders spoke to residents from the target area to promote the anti-violence campaign. Overall, in terms of the frequency and breadth of contact with the residents of the target neighborhood, TRUCE did a good job of reaching out to residents and community members. There was less emphasis, however, on community mobilization efforts that focused on businesses, faith-based organizations serving the community, and other potential community stakeholders. This issue is discussed further below. Exhibit 4: Community Events and Activities by Type Event Type n % Community Gardens 98 52.1 Canvassing 29 15.4 Barbeques / Social 21 11.2 Client-focused Activities 19 10.1 Community Service Events 10 5.3 Community Association Meetings 9 4.8 Peace Marches / Vigils 4 2.1 Total 188 100.0 Planning meetings The Phoenix TRUCE project held monthly stakeholder meetings. These were attended by CPLC and the three partners who were directly involved in the project – Arizonans for Gun Safety, St. Joseph’s Hospital, and the Phoenix Police Department. ASU also attended these meetings as part of its role in monitoring and evaluation. Although these stakeholders met routinely, there was not a strong, community-wide, advisory board. The Advisory Board was intended to play a major role in identifying and 32 leveraging different partners toward the project’s goals and objectives. The lack of an active board created a challenge for community mobilization. Advisory Board meetings were held monthly for only five months, between August 2010 and January 2011. TRUCE stakeholder interviews revealed that the absence of the Advisory Board had hindered the project, and they provided some explanations for why the Board had not played a stronger role. First, not all of the relevant stakeholders were aware that advisory board meetings were being held. Second, stakeholders explained that the board was never fully functional. For example, one stakeholder indicated that developing the advisory board “fell through [the] cracks.” The stakeholder told us, “[The Advisory Board] has been a challenge for us, exactly deciding, you know, some of the individuals that we want to come to the table... we just don’t have it where we want it.” Another stakeholder put it more bluntly: …we need people on that Advisory Board who can help us with the issues of sustainability. We are supposed to be involving the faith-based community, so we need people on that Advisory Board who are part of the faith-based community. We have a hospital component. We should have people who are on there who …represent hospitals. We had one Advisory Council meeting, people that I had given the names of, specific names that I thought would be good from my perspective….[and CPLC] never sent out an invitation to them. Stakeholders acknowledged that there were a number of consequences of not having an active advisory board. They explained that the advisory board plays a key role in connecting with community agencies, and it “is an integral part of what needs to happen to make it successful.” If TRUCE were to have had a strong advisory board, “I think they would have heard 33 very early on where the need for change was, whether or not they acted on it, but at least the option would have been there.” Stakeholders stated that instead of developing partnerships through the Advisory Board, CPLC relied on the provision of services through its own organization. For example, one stakeholder told us: [TRUCE’s] biggest partner right now internally is our workforce development … they are doing a lot of the referrals for our temp agency that we have internally, and then they are also providing more of a counseling type with some of our clients. So there are a lot of internal referrals going on. Core Component 2: Youth Outreach and Intervention The second core component was to “Provide outreach and intervention services to high risk individuals.” The following sections describe our findings related to conflict mediation, client recruitments, and enrollment and client contacts and risk reduction. Conflict mediation and violence interrupters Conflict mediations are an important part of the CeaseFire model and have been in use since the start of TRUCE’s programmatic activities. The violence interrupters, and to a lesser degree outreach workers, mediate potentially violent conflicts in the community. The Chicago CeaseFire model describes violence interrupters as the primary liaisons of the program, who intervene during potentially violent events using conflict mediation techniques. Their mediation efforts seek to alter trends in violence by intervening at key points in the continuum of violence escalation. Violence interrupters operate in the targeted TRUCE community as liaisons in much the same manner dictated by the Chicago CeaseFire model. The individuals selected as 34 interrupters for TRUCE adhered to the principle of the model –they were people with legitimacy in the sense of “street credibility.” Individuals being considered for the position were interviewed on a variety of criteria, including their legitimacy with the targeted high-risk population. Among some of the descriptions of these qualifications for various violence interrupters were the following:  “[I]s a long time resident of the target area…and knows key players in the community.”  “[H]as lived the street life and is no longer involved.”  “Is well-known and has the street credibility needed to resolve conflicts before they escalate into gun violence.”  “[H]as the networks necessary to gather information about what groups are having conflicts.”  “[W]ell known in the Latino enclave of the community.” As such, personal experience in what was described as “street life,” and knowing key people in the community and being known by them, was required to be selected as a violence interrupter. This requirement was consistent with the guidelines and expectations of the CeaseFire model. Experience in the “street life,” for the purposes of the model, depended principally on having a history of criminality and gang affiliation. These prior affiliations and current knowledge positioned the violence interrupters in the community with sufficient legitimacy and credibility to garner the respect and attention of those at risk for involvement in gun violence. The violence interrupters mediated conflicts using the language and logic of the streets. The model emphasizes the importance of reducing gun violence, while being less concerned 35 with other crimes and concerns. Violence interrupters communicated the importance of reducing gun violence in a variety of ways, such as telling drug dealers that “shootings are bad for business” (Skogan et al., 2008), that they draw the attention of the police, increase the likelihood of retaliatory violence, and scare away customers. By focusing on reducing gun violence specifically, and not being concerned with other crimes (e.g., drug dealing), the interrupters can communicate within the limits of the logic and in the language of the streets. This was an important distinction in the CeaseFire model, and one that Phoenix TRUCE implemented well. As discussed above, violence interrupters needed to be familiar with the culture of the street and to communicate in a manner understood by the target population, in order to establish their credibility. Skogan and colleagues (2008) noted that the violence interrupters in Chicago sometimes had difficulty separating their former lives of violence from their present. One of the Chicago violence interrupters recalled an incident in which he had thoughts of violence stemming from a minor traffic situation. A car was sitting and blocking the road ahead of him, and he grew angry. “I hoped the person would get out, so I could bust his head” (pg. 55). The Phoenix violence interrupters walked a similar narrow line between past and present. One TRUCE stakeholder recounted a time when he reacted strongly during an interaction with a gang leader who was dealing drugs. The gang leader was angry with the TRUCE people because they “were taking away my soldiers”and he threatened, “Oh, I’m gonna go get my Uzi.” The violence interrupter described his immediate reaction: 36 Oh, he got checked real quick. I’m from the streets, man, I’m not sugar coat’n nothing… I’m a beat a mother… [I told the gang leader], “I done did a lot of violent stuff. Probably more violent stuff than you and did more prison time. He went on to describe how the interaction ended: “Even though they may have threatened me, at the end of the night, he was shaking our hand, gett’n our number.” The role of the violence interrupter demands that one maintain a difficult balance. It requires individuals with a history of crime and violence, yet who are removed enough from that time to value a life and community without the violence. The importance of credibility on the streets was summarized succinctly by one of the TRUCE stakeholders: “If you not real with this job, you really in the wrong place, because somebody going to hurt you.” Violence interrupters and outreach workers become aware of conflicts in a variety of ways, then make contact with the parties involved in an effort to help negotiate a peace and reduce the likelihood of shootings. As seen in Exhibit 5, the Phoenix TRUCE project engaged in 58 conflict mediations from June 2010 through December 2011. In 42 of the 58 (72.4%) conflicts mediated, TRUCE workers became aware of the conflicts through personal contacts with one or more of the involved parties, or through a personal contact who was aware of the conflict and asked for TRUCE staff to become involved. The remaining 16 (27.6%) mediations came to the attention of TRUCE staff through direct street work. These mediations took place when TRUCE staff were in the community when the conflict and the opportunity to mediate arose. 37 Exhibit 5: Method of Awareness of Conflict Personal Contact Street Police Hospital Other Total % 72.4 27.6 0.0 0.0 0.0 100.0 N 42 16 0 0 0 58 As shown in Exhibit 6, most mediations were conducted in small groups or pairs of mediators (n=24, 41.4%). These mediations typically included at least one violence interrupter (n=51, 87.9%), but occasionally outreach workers conducted mediations on their own (n=7, 12.1%). Exhibit 6: Position of the Mediator Violence Interrupter and Outreach Worker Violence Interrupter Only Outreach Worker Only Other % N 41.4 24 46.5 27 12.1 7 0.0 0 Total 100.0 58 Most typically, conflicts involved either 3 or 4 disputants (n=31, 53.4%), with an average of 3.9 disputants. One case involved approximately 80 disputants from rival gangs. For the purposes of analysis, this case was treated as an outlier, and the value of the number of disputants was truncated to the next highest value, which was 10. Prior to intervention, there was a high degree of risk for an escalation of violence and for the conflict to lead to shooting. The mediation report indicated that the intervention was successful, with members from the respective gangs “shaking hands” at the conclusion. 38 Mediations were coded with one of three outcomes: resolved, resolved temporarily, or conflict ongoing. Mediations were coded as “resolved” when further conflict between the parties was considered very unlikely. Ten (17.2%) of the 58 conflicts mediated were deemed “resolved.” Mediations were coded as “temporarily resolved” when the existing conflict between the parties was resolved for the moment, but future conflicts between the parties were possible. Almost two-thirds (n=37, 63.8%) of the conflicts were considered temporarily resolved. Mediations coded as “conflict ongoing” were considered still to present some risk for violence, although none of the mediations considered these conflicts imminent. Ten (17.2%) conflicts were coded as ongoing. (See Exhibit 7.) Exhibit 7: Status of Mediation Outcome Conflict Resolved Conflict Temporarily Resolved Ongoing Conflict Outcome Unknown % 17.2 63.8 17.2 1.7 Total 100.0 N 10 37 10 1 58 Exhibit 8 shows that gang issues were the most frequent source of conflict (n=27, 46.6%). This was similar to a finding for Pittsburg’s One Vision One Life (Wilson et al., 2011), which reported 41.3% of all conflicts mediated stemmed from inter-gang disputes. Second to gang-related conflicts were those arising from personal interactions or relationships (n=19, 32.8%), followed by domestic violence related conflicts (n=10, 17.2%). Personal arguments/disputes were also the second most common reason for conflicts in Pittsburg (25%), although domestic disputes were less frequent (8.7%). Conflicts related to narcotics were rare in Phoenix (n=1, 1.7%), and although not frequent, were more common in Pittsburg (9.6%). 39 Exhibit 8: Primary Reason for Conflict Gang Personal Altercation Domestic Violence Narcotics Robbery Child Abuse Unknown Other Total % 46.6 32.8 17.2 1.7 1.7 0.0 0.0 0.0 100.0 N 27 19 10 1 1 0 0 0 58 Consistent with the CeaseFire model, Phoenix TRUCE targeted the conflicts and disputants who had the highest likelihood of escalating to gun violence according to the seven criteria outlined in the above section. As seen in Exhibit 9, the majority of individuals involved in the conflicts mediated by violence interrupters were either gang-involved (n=41, 70.7%), had a history of violence (n=37, 63.8%), or were 16-25 years old (n=53, 91.4%). Twenty-two (37.9%) of the conflicts involved individuals who had all three risk factors (i.e., they were gang-involved youth with a history of violence). This suggests that the Phoenix TRUCE violence interrupters adhered to the CeaseFire model for the implementation and use of conflict mediations by focusing efforts on those members of the community meeting criteria for a high-risk gun violence involvement.2 2 Detailed information on the risk level of disputants was not reported/available for other CeaseFire replication sites to conduct a comparison of the performance of Phoenix TRUCE relative to other replication sites. 40 Exhibit 9: Risk Factors of Disputants in Conflicts Mediated (N=58) Gang Related History of Violence High Risk Street Activity Recent Shooting Age 16-25 Recently Released from Prison Involved with Weapons % 70.7 63.8 44.8 10.3 91.4 25.9 36.2 N 41 37 26 6 53 15 21 Client recruitment and enrollment The violence interrupters are an important and distinguishing element of the CeaseFire programmatic model. However, outreach workers who identified and recruited high-risk individuals for TRUCE, and then followed up with customized services, were critical to sustaining any violence reductions achieved. Outreach workers strove to alter the community’s cultural norms related to violence - to gun violence, in particular. TRUCE outreach workers followed the CeaseFire model in selecting clients for the case management and risk reduction components of the program. The model calls for potential clients to be identified as high-risk for involvement in gun violence, either as victims or perpetrators, using a rubric of seven at-risk characteristics: 1. Young – aged 16 to 25 years old 2. Prior history of offending 3. Gang involvement 4. Key role or position in the gang 5. Involvement in high-risk street activity 6. Recently released from prison 7. Recent victim (past 90 days) of a shooting Not all of the criteria were well-defined. TRUCE staff had no clear guidance, for instance, in determining how “recently” a person had to have been released from prison to qualify as 41 “high-risk.” When asked, TRUCE outreach workers generally responded that they considered “recent” to mean a release within the past year. Being a “recent” victim of a shooting was better defined by the model and well-understood by TRUCE staff: Individuals who had been the victim of a shooting within the past 90 days were considered to be recent victims for the purposes of risk assessment, whether injured or not, and whether or not they had been the intended target. The final and most ambiguous criterion was involvement in high-risk street activity. The Chicago CeaseFire evaluation identified that approximately 90% of their nearly 600 clients were involved in street-level drug markets (Skogan et al., 2008). In Chicago, involvement in high-risk street activity functionally served as a proxy for involvement in street-level drug markets, although the model allows for variation. The risk criterion also could include strong-arm robberies, threats and intimidation in furtherance of a gang, and other activities associated with violence. CeaseFire implementation in Baltimore (Safe Streets) and Pittsburg (One Vision One Life) (see Webster et al., 2009; Wilson et al., 2011) had more variation in the kinds of activity used to qualify than did the Chicago sites. Outreach workers used their own existing networks in the community to begin the process of identifying and recruiting potential program clients. The model describes the selection process as a courting ritual. Outreach workers and violence interrupters came into TRUCE with pre-existing knowledge of the community and its people. Initially, TRUCE workers contacted potential clients already known to them.. These individuals were assessed using the model’s selection criteria for their suitability and willingness to participate in TRUCE; some became clients. 42 Details on how many individuals were contacted during this initial phase, and how many of those were recruited and enrolled as clients, were not recorded, although it seems apparent from interviews with staff and the review of the client management database that these efforts did occur. These individuals, their family or friends, and other people with whom outreach workers had an existing relationship often identified other potential clients in a snowball strategy. The One Vision One Life (Pittsburg) implementation of the Chicago CeaseFire model recorded the sources of clients referrals (e.g., school official, community member, church), but this type of referral and recruitment information was not collected for analysis by TRUCE. Consistent with the Chicago CeaseFire model, clients were recruited for Phoenix TRUCE only if they were assessed as having four of the seven selection criteria listed above, thus placing them in the “high-risk” category. At the time of this analysis, a total of 118 clients had been fully enrolled, all of whom met at least four of the selection criteria. As seen in Exhibit 10, of 118 clients, 81 (68.6%) met four of the seven criteria, another 27.1% (n=32) met five criteria, 3.4% (n=4) met six criteria, and one client met all seven high-risk criteria (0.8%). It should be noted that one client was not fully assessed because the client dropped out of the project soon after agreeing to participate. That client was excluded from analyses. Exhibit 10: Identified Clients for At-Risk Gun Violence Assessed High-Risk Clients Total % 100.0 100.0 N 118 118 68.6 27.1 3.4 0.8 100.0 81 32 4 1 118 Assessed High Risk Criteria Factors Four Five Six Seven Total 43 The most common of the seven risk factors among clients, as seen in Exhibit 11, was being involved in a gang, with 89% (n=105) of clients identified as such. Gang membership was closely followed by a prior criminal history (n=100, 84.7%) and involvement in high-risk street activity (n=100, 84.7%). More than two-thirds (n=79, 66.9%) of clients were between 16 and 25 years old, and most held a key position in the gang (n=67, 56.8%). Thirty-three clients (28%) had been recently released from prison. Not surprisingly, being a recent (past 90 days) victim of a shooting was the least frequent criterion present, although this described more than a quarter (n=31, 26.3%) of the clients. Thus, the typical TRUCE participant was someone between the ages of 16 and 25, gang-involved, with a history of violent crimes, and currently involved in high-risk street activity (n=43, 36.1%). Exhibit 11: Participant Risk Factors % Assessed High-Risk Clients Total Gang Involvement Prior History of Offending Involvement in High-Risk Street Activity Youth - Aged 16-25 Years Old Key Role or Position in the Gang Recently Released from Prison Recent Victim (Past 90 Days) of a Shooting N 100.0 118 100.0 118 89.0 105 84.7 100 84.7 100 66.9 79 56.8 67 28.0 33 26.3 31 Other characteristics of TRUCE clients were collected as part of program intake, as well. The seven selection criteria combined with other relevant factors helped outreach workers develop a Risk Reduction Plan tailored to each client’s needs. As seen in Exhibit 12, most clients 44 were male (n=86, 72.9%). The majority of clients were Black/African-American (n=93, 78.8%), followed by Hispanic/Latino (n=20, 16.9%), Native American (n=3, 2.5%), and white (n=2, 1.7%). Forty-seven clients (39.8%) were determined to have completed at least high school or its equivalency. While the educational attainment for 17 clients (14.4%) was recorded as “To Be Determined” in the database, 44.9% of all clients had less than a high school education (n=53). Twenty-five clients (21.2%) were unemployed at the time of intake, and another 33.9% (n=40) were employed only part-time. Only six clients (5.1%) reported full-time employment. Among all clients (n=118), 70.3% (n=83) were actively looking for employment at the time of intake, which included those employed part-time (essentially seeking full-time employment). A substantial proportion of clients (n=50, 42.4%) were currently on probation or parole at the time of intake; 15 (12.7%) were currently on parole and 35 (29.7%) were currently on probation. No clients were currently on both probation and parole at the time of intake. Another 9.3% (n=11) of clients had previously served and completed the terms of probation, and 5.9% (n=7) had completed parole. 45 Exhibit 12: Client Characteristics (n=118) % N Male Female 72.9 27.1 86 32 African American Latino Native American Caucasian 78.8 16.9 2.5 1.7 93 20 3 2 Completed Grammar School Completed High School/GED Completed Trade School TBD Don't Know 44.9 34.7 5.1 14.4 0.8 53 41 6 17 1 Unemployed Part-time employment Full-time employment Actively looking for employment Formerly Incarcerated Yes No Probation On Probation Completed Terms of Probation Not on Probation TBD Don't Know Parole On Parole Completed Terms of Parole Not on Parole TBD Don't Know 21.2 33.9 5.1 70.3 25 40 6 83 65.3 34.7 77 41 29.7 9.3 53.4 5.9 1.7 35 11 63 7 2 12.7 5.9 75.4 3.4 2.5 15 7 89 4 3 Gender Race Education Level Employment 46 Client contacts and risk reduction The fundamental purpose of appropriately identifying, recruiting, and enrolling high-risk individuals into the program is to change the norms and behaviors of those most likely to engage in gun violence. The client selection process is intended to develop a caseload for outreach workers comprised of individuals in the target community at the highest risk for either being the victim or the perpetrator in a shooting. The outreach workers then focus on changeable risk factors and other characteristics (e.g., employment, education, life skills) that will generate a lasting change in the individual. Through one-by-one changes, the program is designed to attain sustained reductions in gun violence by incrementally changing the cultural norms of the target community favorable toward using shootings as a means for conflict resolution into alternative resolution strategies. The outreach workers are to manage clients through this portion of the programmatic intervention by cooperatively developing a Risk Reduction Plan tailored to each individual participant. Once developed, the outreach worker is to make frequent contact with their clients, checking their progress, adjusting the plan as necessary, and referring the clients to needed services. The Phoenix TRUCE program implemented the CeaseFire model very well regarding the routine of frequent contact and interaction with their clients. However, we found that the implementation of the cooperative development of a Risk Reduction Plan and the use of continuous progress updates was not fully implemented. Risk Reduction Plan The Risk Reduction Plan is an important tool for outreach workers and project supervisors to use in order to track the specific course and progress of each client in 47 transitioning from high risk for gun violence to a reduced risk level. Developing the cooperative Risk Reduction Plan is to be conducted soon after a client’s intake. As seen in Exhibit 13, of the 118 clients, 43 (36.4%) had a documented Risk Reduction Plan. Exhibit 13: Clients with Risk Reduction Plans % N Does the client have a risk-reduction plan? Yes 36.4 43 No 63.6 75 Total 100.0 118 Of those plans that were completed, most were completed within a reasonable amount of time; however, a few risk reduction plans took months to complete. For example, Exhibit 14 shows that 5 of the 43 clients (11.6%) had a Risk Reduction Plan documented 123 days or more (range=123-255 days) after the client’s intake date. 48 Exhibit 14: Days between Intake Date and Risk Reduction Plan % N 0 1 2 3 4 5 6 7 11 12 19 22 25 33 123 151 185 255 34.9 16.3 9.3 4.7 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3 4.7 2.3 15 7 4 2 1 1 1 1 1 1 1 1 1 1 1 1 2 1 Total of Clients with Risk Reduction Plans 100.0 43 Days After reviewing the status of the client management database in May 2011, evaluators from ASU conducted a training session for outreach workers and violence interrupters on data entry procedures and data quality concerns. During this training session, TRUCE staff indicated that they will sometimes process potential clients with whom they are working, but these individuals have not fully committed to the program; thus, they have been processed as clients, and outreach workers are working with them as clients, but they have not developed a cooperative Risk Reduction Plan. Although this explanation helps describe the handful of 49 anomalies in delays of a plan development, it does not sufficiently explain the overwhelming absence of a documented Risk Reduction Plan for the majority of clients. During this same training session, TRUCE staff said that many of the clients who did not have a documented Risk Reduction Plan were still being managed as though they did, although a clearly defined plan was not documented into the system. For example, one outreach worker indicated not recalling ever having been trained to develop and document such a plan, and as such, none of the outreach worker’s clients had a documented Risk Reduction Plan prior to June 2011. It should also be noted that detailed information about specific services (i.e., anger management, substance abuse programs, etc.) was not available for evaluation. This report summarizes the activities of the TRUCE outreach and violence interrupter workers, and does not evaluate specific services to which clients were referred and received. Exhibit 15 shows the results of the analysis of the 43 documented Risk Reduction Plans, which indicated that 53.5% (n=23) of clients were identified as needing help to resolve conflicts and problems through non-violent means. Some form of life skills training was an assessed need for all of the clients with a Risk Reduction Plan. Anger management was identified as a need for 27.9% (n=12) of clients. Family issues and mentoring combined as indicators of a more generalized need for life skills training, with 32.6% (n=14) and 86.0% (n=37) of clients needing the training, respectively. No clients were identified as requiring mental health treatment, and 37.2% (n=16) of clients needed assistance with reentry issues. 50 Exhibit 15: Risk Reduction Plan Assessment Needs of Life Skills Training % N Life Skill Training Yes 36.4 43 No 0.0 0 N/A* 63.6 75 Total Number of Clients 100.0 118 Type of Service (n=43) Mentoring Resolving Conflicts/Problems without Violence Re-entry Issues Family Issues Anger Management Mental Health * Clients do not have a Risk Reduction Plan 86.0 53.5 37.2 32.6 27.9 0.0 37 23 16 14 12 0 Exhibit 16 shows that employment needs were also a common feature of the Risk Reduction Plans, with 39 (33.1%) clients having employment assistance identified as part of their plans. Of those clients, 56.1% (n=23) needed assistance with job readiness. Job readiness focused on issues related to appropriate clothing for job seeking and job keeping, acquiring appropriate legal documents and identification for proof of employment eligibility, transportation needs (e.g., teaching the client about how to use public transportation), and similar basic skills to assist them in getting ready to get and keep a job. Job preparedness was even more important to the clients, with 73.2% (n=30) of clients needing assistance in this area. Job preparedness is defined as learning how to locate job openings and complete employment applications, preparing a resume, and practice successful interviewing techniques. Finally, 51.2% (n=21) of clients were identified as needing a vocational or job training program. 51 Exhibit 16: Risk Reduction Plan Assessment Needs of Assistance with Employment % N Assistance with Employment Yes 33.1 39 No 3.4 4 N/A* 65.3 75 Total Number of Clients 100.0 118 Type of Service (n=39) Job Preparation Job Readiness Enrolling in a job training program * Clients do not have a Risk Reduction Plan 73.2 56.1 51.2 30 23 21 As noted previously in Exhibit 12, educational attainment was low among the caseload, with 44.9% (n=53 of 118) of clients identified as having less than a high school or equivalent education as their highest attainment. The Risk Reduction Plans reflected the need for education among the clients, although dominated by post-high school education. One client (2.3%) was identified as needing pre-high school (GED preparation) or English as a Second Language (ESL). Eleven clients (25.6%) needed assistance with getting their GED. Five clients (11.6%) wanted assistance preparing for college or university. The most common educational assistance need was for alternative or trade school educational attainment, with 37.2% (n=16) of clients wanting/needing assistance in this area. Two clients reported a need for assistance in two educational categories. (See Exhibit 17.) 52 Exhibit 17: Risk Reduction Plan Assessment Needs of Assistance with Enrolling in an Educational Program % N Yes 26.3 31 No 10.2 12 Assistance with Enrolling in an Educational Program N/A* 63.9 75 Total Number of Clients 100.0 118 Type of Service (n=31) Alternative School/Trade School GED/High School College/University GED Tutoring/ESL 37.2 25.6 11.6 2.3 16 11 5 1 * Clients do not have a Risk Reduction Plan Substance abuse problems were relatively rare among the clients, in terms of their identified needs on their Risk Reduction Plans. The most common type of substance abuse assistance needed was detoxification (25.6%, n=11). Almost 12% (n=5) of plans called for enrolling the participant in a substance abuse treatment program, and the same number of clients (n=5, 11.6%) were identified as needing to be enrolled in a substance abuse self-help program (e.g., Alcoholics Anonymous, Narcotics Anonymous). 53 Exhibit 18: Risk Reduction Plan Assessment Needs of Assistance with Substance Abuse % N Assistance with Enrolling with Substance Abuse Yes 13.6 16 No 22.9 27 N/A* 63.6 75 Total Number of Clients 100.0 118 Type of Service (n=16) Detox Enrolling in a substance abuse program Enrolling in a self-help group (AA, NA, etc) * Clients do not have a Risk Reduction Plan 25.6 11.6 11.6 11 5 5 Related to the Risk Reduction Plan is the concept of client “graduation.” The primary purpose of the plan is to have a planned course of action addressing particular criminogenic needs in an effort to reduce an individual’s relative severity of risk to be involved in gun violence. When clients reach a stage where they believe they have sufficiently addressed their needs, and the outreach worker believes that their overall risk level has been substantially reduced, the individual is removed as an active client. In total, of the 118 cases, 71 (60.2%) were closed with identified reasons, another 14 (11.9%) were closed without any indicated reason, and the remaining 33 (28%) were open and active at the time of analysis. Analysis of program data on the closed cases, as shown in Exhibit 19, indicated that 27 (31.8%) clients’ cases were closed because they were no longer considered high-risk for involvement in gun violence. The most common reason given for closing a case was due to the client moving outside the area (n=38, 44.7%). Fifteen clients’ (17.6%) cases were closed because they were no longer in contact with TRUCE personnel. Although rare, 8.2% (n=7) of clients’cases were closed as active cases because the client was convicted and given a long prison sentence. Two (2.4%) 54 cases were closed because the client had died. There were an additional 18.8% (n=16) clients whose cases were closed for “other” unidentified reasons. Exhibit 19: Participant Reasons for Closure % N Reason for Closure Identified Yes 60.2 71 No 11.9 14 N/A* 28.0 33 Total 100.0 118 Reason** (n=85) Moved Outside Area No Longer Qualifies as High-Risk No Contact (For more than 1 Quarter) Long Term Prison Sentence Deceased Other (unspecified) * Active/open cases ** Reasons are not mutually exclusive categories. 44.7 31.8 17.6 8.2 2.4 18.8 38 27 15 7 2 16 Client Contacts Analysis indicated that outreach workers established and maintained routine, frequent contact with the clients in their caseloads. The 118 documented active and closed clients were contacted by outreach workers a total of 5,414 times. Clients were active from 1 to 17 months, at an average of 4.3 months. Outreach workers averaged a total of 45.9 contacts per client, although the range was substantial, from 0 (n=6) to 240 (n=1). The client with 240 contacts was also the longest serving client at 17 months. The six clients who were recorded as having no contacts post-intake were all closed cases. Two cases were closed within the same month as intake because the clients moved away. A third case was closed within the same month as intake, although no reason for closing the case was recorded. Another two clients were closed 55 after TRUCE staff were unable to make contact with them immediately following intake. The last of these six cases had a recorded intake date in January 2010, had a documented office visit to develop a Risk Reduction Plan in June 2010, and no further contact after that time. There is insufficient documentation on this case to make any determination or offer any explanation. Exhibit 20 shows the number of case notes (n=5,414) per participant (n=118). Exhibit 20: Number of Case Notes per Participant (N=5,414) # Case Notes % 0-25 26-50 51-75 76-100 101-125 126-150 151-175 176-200 200+ 8.2 14.1 12.8 24.0 17.2 2.5 2.9 10.1 8.3 Cumulative % 8.2 22.3 35.1 59.1 76.3 78.8 81.7 91.8 100.1 Outreach workers contacted clients in a variety of ways. The 5,414 contacts across 112 participants (note: 6 clients did not have any recorded contacts after intake, as described above) were recorded as case notes. The case notes followed a standardized set of checklists that focus on common activities that are part of a typical CeaseFire client interaction and a review of the client’s Risk Reduction Plan. The case note form collects information about the date, length, and type of each contact, and details about the client’s recent involvement with violence, employment status, education, substance abuse, probation and parole status, and criminal justice system involvement. The majority of contacts, as seen in Exhibit 21, were 56 relatively brief, with 64.3% of all contacts estimated as lasting 60 minutes or less. The overall average length of a contact was from about an hour to an hour and 15 minutes. Exhibit 21: Time Spent with Clients N % Cum. % Time 30 Minutes or Less 1,781 32.9 32.9 30 - 60 Minutes 1697 31.4 64.3 60-90 Minutes 722 13.4 77.7 90 - 120 Minutes 510 9.4 87.1 120 - 150 Minutes 542 8.4 95.5 150 - 180 Minutes 178 3.3 98.8 180 - 210 Minutes 38 0.7 99.5 210 Minutes or More 30 0.6 100.1 Total 5,408 * 6 contacts did not have a recorded length of contact The most common type of contact was made by phone (n=2,031, 37.7%), closely followed by home visits (n=1,392, 25.9%). More than 10% of contacts (n=548, 10.2%) were made on the street (or other public areas in the target community), and 6.6% of contacts (n=357) were office visits. Another 19.6% (n=1,053) of contacts were recorded as “other” forms of contact, and specific details were not available. Thirty-three contacts (0.6%) did not include information on the type of contact. Evaluations in Chicago (Skogan et al., 2008), Baltimore (Webster et al., 2009), and Pittsburg (Wilson et al., 2011) have emphasized the importance of face-to-face contacts, particularly those conducted in the community. Phoenix TRUCE personnel made face-to-face contacts with clients in the community 42.4% (n=2,297) of the time. All clients had had at least one face-to-face contact since their intake date, with 84.7% 57 (n=100) of clients having five or more face-to-face contacts. The average number of total faceto-face contacts per client was 39.1 (n=5,414)3. Exhibit 22: Type of Contact with Participant Type of Contact Phone Home Visit Street Office Other % N 37.7 25.9 10.2 6.6 19.6 Total 100.0 2,031 1,392 548 357 1,053 5,381 The content of the interaction between outreach worker and client is also very important. Broadly, outreach workers discuss issues related to violence, education, employment, substance abuse, and criminal justice concerns. Case notes record each contact and predefined checklists, provided by the Chicago Project for Violence Prevention for use by CeaseFire and CeaseFire replication sites, delineate a common framework for use across an individual’s specific contacts, clients, and programmatic sites. The first discussion area is violence. Outreach workers ascertain and record whether the client has perpetrated violence in the past 30 days, or made threats of violence. Next, outreach workers record whether they discussed using non-violent means to settle disputes, if they mediated a conflict for the client during the contact, and if they referred the client to an anger management or life skills training course. 3 Reports from Chicago (Skogan et al., 2008), Baltimore (Webster et al., 2009), and Pittsburg (Wilson et al., 2011) do not provide the details on the comparative distribution of face-to-face (i.e., home visits and on the street in the community) contacts to all client contacts (phone, office visits, or other locations), so we are unable to make a direct comparison to other sites implementing the Chicago CeaseFire model. 58 Exhibit 23 shows that among all contacts, 11.8% (n=633) recorded that the client had been violent recently, and another 2.5% (n=135) had voiced threats recently. Discussions about using non-violent alternatives to resolve conflicts was a common theme, with 49% (n=2,621) of all contacts discussing the issue. Conflict mediations were rare, with only 1% (n=51) of all contacts including a conflict mediation. Referrals for anger management or life skills courses were almost nonexistent, with only one such reported referral. This finding conflicts with the fact that 27.9% (n=12) of clients with a Risk Reduction Plan were identified as needing anger management services and 32.6% (n=14) needed life skills training related to family issues and violence. Exhibit 23: Case Notes Reporting Violence (n=5,349) Participant has been violent recently (within the past 30 days) Participants voices threats of violence Discussed the use of non-violent means to solve problems/conflicts/issues/ etc. Mediated conflict with participant Referred the participant to the following anger management/life skills program % N 11.8 633 2.5 135 49.0 2,621 1.0 51 <.01 1 Analyzing the contact case notes revealed the importance of the employment issue for participants. Exhibit 24 indicates that during contacts with the outreach workers, clients reported being employed full-time just 12.9% (n=688) of the time, and unemployed during 7.8% (n=416) of the time. Another 47.5% of contacts (n=2,525) found clients employed at least parttime. During more than half of all contacts (52.4%; n=2,787), clients were actively looking for 59 work. This last figure indicates the condition of under-employment among clients. Case notes also included whether the client’s employment needs were discussed in general and whether the outreach worker assisted with job readiness (e.g., clothing, identification, transportation), job preparation (e.g., completing applications, preparing a resume, interviewing strategies), or a job referral. TRUCE outreach workers indicated that they discussed the client’s employment needs 15.8% (n=839) of the time. Assistance with job preparation was given 8% (n=428) of the time, and job readiness assistance was given in only 1% (n=52) of contacts with clients. Job referrals were similarly low, with 1.7% (n=89) of contacts resulting in job referrals. Exhibit 24: Case Notes Reporting Employment (n=5,318) % N Unemployed Employed Part-Time Employed Full-Time Actively looking for employment 7.8 47.5 12.9 52.4 416 2,525 688 2,787 Discussed the participant's employment needs Assisted with job readiness (clothing, state ID's, etc.) Assisted with job preparation (resume, applications, etc.) Referred to employment 15.8 1.0 8.0 1.7 839 52 428 89 Exhibit 25 presents our findings related to case notes on education. About one-fourth (n=1,284, 24.7%) of all contacts included at least some discussion on the subject. Referrals to specific educational programs were very low, with just 54 instances (1%) of an educational referral being made during the contact. Exhibit 25: Case Notes Reporting Education (n=5,188) Interesting in enrolling in school Discussed education issues Referred the participant to education institution/program % N 33.6 24.7 1.0 1,744 1,284 54 60 Exhibit 26 indicates that outreach workers discussed the need for the client to comply with the terms of their probation during 39.9% (n=559) of contacts among clients on probation (n=1,400), and reported 13 instances (0.9%) of having met with the client’s probation officer. In 214 contacts (30.8%), outreach workers discussed the importance of complying with the terms of parole among those clients on parole (n=695), and reported having met with the client’s parole officer on 21 instances (3%). Additionally, outreach workers reported having visited the client while in jail on seven occasions (0.1%) and attending court with them on nine occasions (0.2%). Exhibit 26: Case Notes Reporting Probation and/or Parole % N On Probation Discussed the need to comply with terms of Probation Met with the participant's Probation Officer 26.6 1,400 39.9 559 0.9 13 On Parole Discussed the need to comply with terms of Parole Met with the participant's Parole Officer 13.1 30.8 3.0 695 214 21 Hospital Contribution to Youth Outreach and Intervention The original plan for TRUCE included an objective explicitly defining the role of a hospital partner in the efforts toward youth outreach and intervention. The objective called for St. Joseph’s Hospital and Medical Center and its SMARTR program to provide outreach services to patients with violence-related injuries, who might also be potential clients for TRUCE outreach services. An employee of the hospital was to be cross-trained as an outreach worker for TRUCE, thus integrating and leveraging the combined resources. Implementation of this objective was 61 significantly delayed and burdened with obstacles. Chief among these was finding an appropriate person who could meet the programmatic qualifications for TRUCE (i.e., being a credible messenger for at-risk youth) and who could still meet the criteria for employment eligibility with the hospital. This obstacle, combined with other less intrusive bureaucratic ones, caused a year-long delay in hiring the hospital liaison outreach worker. Once implemented, the partnership and communication between the hospital’s SMARTR program and TRUCE has been routinized and productive. At the time of analysis for this report, data were not available to measure client and service referrals from the hospital directly to TRUCE clients or potential clients. Core Component 3: Faith-Based Leader Involvement The third core component of the CeaseFire model is Faith-Based Leadership Involvement. This component is included in the first part of the sixth objective, which states, “coordinate with collaborative partners and advisory board to ensure efficient and effective implementation.” Interviews indicated that a coordinated and collaborative relationship between TRUCE and the faith-based community was not strong. While TRUCE staff believed that there was faith-based involvement in the TRUCE project, those in the faith based community thought that there could have been a stronger relationship between the two. Stakeholders involved with the project explained that faith-based organizations were involved in TRUCE because they allowed the project to use their space for meetings, and, therefore, were aware that the project existed. Likewise, interviews with TRUCE staff revealed that they felt that the faith community was involved because they attended such events as the peace march, candle light vigils, and the TRUCE Mural Project. However, one TRUCE project 62 employee indicated, “…the faith-based side is something we need to develop a little bit more, and get some more interaction. We do have some individuals that come out like for our shooting responses, and you know, offer that side, that spiritual side for individuals who might be having issues with that. That’s still something we want to work on a little bit more next year.” Some of the faith-based leaders, however, did not feel that they were included in the implementation of TRUCE. One stakeholder indicated that they, in fact, heard about the TRUCE project from law enforcement in the area and they had never been approached by TRUCE outreach staff. This was a year after all the staff had been hired and about two years after funding. Likewise, another stakeholder stated: [The faith community] still aren’t seeing any type of referrals happening, but at least the connection and the awareness that they need to be connected has happened…[TRUCE] should be manned and managed and invested in by the community at large, because that is who it is supposed to serve. Core Component 4: Public Education The Public Education core component corresponds with the third objective of the Phoenix TRUCE project, which states, “Educate the community to facilitate behavior change and promote nonviolence.” This component has been led by Arizonans for Gun Safety (AzGS), which is a nonprofit organization dedicated to reducing gun deaths and injuries in Arizona communities. AzGS promotes practical, community-based solutions through partnerships with stakeholder agencies in health care, law enforcement, education, faith-based, youth, and community services. The organization engages in programs and activities to educate youth and 63 the public on gun violence prevention, promote gun safety and violence prevention messages in at-risk communities, and advocate for sensible gun-related policies. Data from stakeholders indicated that a number of different educational materials were distributed as part of the projects implementation. Many of the educational materials were based on the Phoenix TRUCE project’s brand of “TRUCE” and marketing the slogan “Start the Peace. Stop the Violence” to communicate an anti-violence message. With eleven different forms of media and over 11,000 individual items, the community was saturated with educational materials. The table below provides information on the types and amounts of educational material distributed. Exhibit 27: Public Education Materials Distributed During the Project Product T-shirts Hats Dog Tags Wrist bands Brochures Billboards Door hangers Postcards Posters for rally Flyers Videos Total Quantity 1,100 50 1,150 1,500 1,500 2 4,000 1,000 520 500 2 11,324 Date April, 2010 and August, 2010 September, 2010 October, 2010 October, 2010 Throughout the project November, 2010 November, 2010 November/December 2010 November, 2010 and October, 2011 November, 2010 and October, 2011 November, 2011 We examined awareness of the TRUCE project through a randomized in-person household survey. Exhibit 28 shows that 12.5% of residents reported that they knew about the 64 TRUCE Program. Furthermore, in Exhibit 29, we found that only 33.3% of respondents who knew about TRUCE understood the purpose of the TRUCE Program. Exhibit 28: "I know about the TRUCE Program" 100 90 87.5 80 70 60 50 40 30 20 12.5 10 0 No Yes Exhibit 29: "I understand the purpose of the TRUCE Program" 80 70 66.7 60 50 40 33.3 30 20 10 0 No Yes However, when we focused the analysis on those whom the program targeted, the data showed that a fairly substantial portion of the community was aware of the project. Exhibit 30 65 shows that more than 25% of respondents who knew a shooting victim were knowledgeable about TRUCE. However, only about 7% of those who did not know someone who had been shot knew about TRUCE. Exhibit 30: Know about TRUCE by proximity to violence 100 93.3 90 80 73.7 Percent 70 60 50 40 26.3 30 20 6.7 10 0 Do NOT know a shooting victim Have NOT heard of TRUCe Know a shooting victim Heard of TRUCE TRUCE has attempted to reach community residents in a variety of ways. Survey respondents were able to select all the ways they had heard of TRUCE. As shown in Exhibit 31, about 30% of residents who knew of TRUCE stated they had seen a TRUCE poster and 20% heard about TRUCE at a community barbecue. About 10% of residents who had heard of TRUCE indicated they had heard about it from a violence interrupter, a flyer, TRUCE clothing, or some other community event. 66 Exhibit 31: "How have you herd about TRUCE?" 100 90 80 70 60 50 40 30 20 10 0 30.0 20.0 10.0 0.0 10.0 10.0 0.0 10.0 0.0 Core Component 5: Criminal Justice Participation The criminal justice component of the Chicago CeaseFire model is an important part of the collaboration upon which the model depends. Along with faith-based organizations, service agencies, community coalitions, schools, and businesses, the police play a crucial role in the implementation and execution of the model’s intended programmatic activities. Police are the most visible and active agency in the criminal justice component, although prosecutors may also contribute. As partners in the CeaseFire model, criminal justice agencies are generally expected to provide strict enforcement options in targeted communities, open information sharing regarding crimes, particularly gun related offenses, provide safety for public events, and generally support cooperative efforts with programmatic activities. Criminal justice partners for the Phoenix TRUCE site were almost exclusively limited to the Phoenix Police Department. While the police were supportive, there were limitations in the full implementation of the 67 collaboration as the model intended.4 Specifically, stakeholders noted two issues: one was related to the relationship between CPLC and the PPD and the other was related to information sharing. Relationship between CPLC and the PPD The Phoenix Police Department has served as a strong partner in the strategic planning of the Phoenix TRUCE project from the outset, and has continued as a strong partner throughout the project’s tenure. The community targeted by TRUCE is located in the South Mountain precinct of the Phoenix Police Department. The PPD has been a regular member of the stakeholder group, attending most meetings over the entire project period. The PPD was also an active partner in the planning of multiple public events sponsored by Phoenix TRUCE. In this capacity, the police have provided traffic control and support, and safety at public events (e.g., public parks and streets). For example, in preparation for a peace march (planned on short notice), the PPD was involved in the planning of the event from the outset. The PPD assisted with the route selection in and around the targeted neighborhood by providing guidance related to important traffic control and safety concerns. The PPD coordinated the arrangements to have police support the marchers by scheduling motorcycle officers to assist with traffic control, particularly along the busiest streets bordering the neighborhood where the safety of pedestrians was of chief concern. The police further provided extra presence at 4 The CeaseFire model in part calls for the use of criminal justice agencies in the public education materials. Public education materials may inform the public about the formal criminal justice sanctions, including, but not necessarily limited to aggressive police enforcement and arrest of serious violent offenders, rigorous prosecution of serious and/or repeat violent offenders, and sentencing enhancements for gun crimes. The evaluation of Chicago CeaseFire conducted by Skogan and colleagues (2005) noted that this was not implemented, and it may well have not been expected in Phoenix. As part of the public education campaign, post cards were disseminated with information on gun crime sanctions; however, this was limited and not a major part of the program. 68 the public park where the marchers ended the route with a large event, including speakers, barbeque, and fair-like booths and vendors. Sites in Chicago used police in the same manner. However, with the above said, interviews with TRUCE stakeholders shed light on the difficulties of establishing rapport between the police and TRUCE. Two major themes emerged. The first major theme was related to the necessity of TRUCE workers to clearly function independently from the police. The CeaseFire model emphasizes the importance for Violence Interrupters and Outreach Workers to have a certain amount of distance and independence from the police. Just as it is important that police share information about events that have an increased likelihood to result in retaliatory violence, particularly shootings, the community must be able to trust that the violence interrupters and outreach workers will not report them to police. This issue of trust is important for the violence interrupters and outreach workers to be able to work with the population. One stakeholder explained what they thought the interaction between street-level project staff and the police ought to be: “We’ve had meetings with some of the officers that, even supervise patrol officers, so that they know, they know who we are, they know our logo, they know why we’re out there.” TRUCE personnel working directly with the clients stated that the police should have a little knowledge of and respect for the project, but should keep a healthy distance. A number of stakeholders voiced concern about the ability to do their job and being able to work with the project’s target population if they were perceived as working too closely with the police. For example, one stakeholder explained, We try not to [work too much with the police], cause when I first started, cool cause we’ll work with them, but because of our population…you know, they hate the police, and if the police come to our office and 69 anything they feel like “you know what, you all ain’t fo real. You all gonna tell the po po on me.” No, we’re not. The second major issue expressed by CPLC staff was related to their distrust of the police. Distrust in the police was expressed very viscerally by one stakeholder: “These people drive around and basically bully and harass people on a regular basis,” and the police “have a job to do, [but] the police of South Phoenix, are assholes.” This sentiment of distrust served as one indicator of the police’s diminished legitimacy in the community. Stakeholders explained that members of the community generally do not trust the motivations or intentions of the police, and do not believe they will be treated fairly by the police officers serving their community. These findings are somewhat similar to those of Skogan and colleagues (2005) who reported that communities selected for CeaseFire replication sites (i.e., TRUCE) are chosen because they have disproportionately high rates of gun violence. Areas and neighborhoods with high rates of shootings also exhibit high rates of assaults, robberies, and street-level drug sales. The TRUCE target community of Hermoso Park is no different. Communities such as these are also typically economically disadvantaged, and have a history of crime, poverty, and poor relations with the police (Walker and Katz, 2011). Information sharing An important activity linked to the partnership between the intervention team and the police is information sharing. Specifically, it is crucial for violence interrupters and outreach workers to be aware of shootings and other events that may lead to shootings (i.e., conflicts and retaliations) as soon as possible. The violence interrupters and outreach workers have close ties to the community and often become aware of conflicts very contemporaneously, allowing 70 for more effective conflict interventions and mediations. The police also have a lot of knowledge about events as they occur. The CeaseFire model calls for the police to provide information about shootings and other violent crimes in the community, including information about the people, places, circumstances, and motivations of the events. This aspect of the program has been somewhat inconsistent, and has not been as strong as it could have been. While the Phoenix TRUCE program has received information from police throughout the life of the program, information sharing was not always routine or quick. This might be because the TRUCE project did not have a standardized reporting mechanism for information to be dispatched from the police to CPLC. As evidenced in the conflict mediation reports, of the 41 conflict mediations conducted by TRUCE since programmatic implementation, none of the conflicts were brought to the attention of violence interrupters and/or outreach workers through police contacts. The TRUCE personnel expressed value in acquiring the information, and what was shared was well received and considered useful. Additionally, police have continued to regularly attend stakeholder meetings and provided liaisons to the project, so a commitment to the partnership exists, yet it is not clear why a more routinized process of information sharing has not been developed. Other Impediments to Implementation Stakeholders discussed three issues that that they believed had impeded the implementation of the TRUCE project. These seemed more the consequences of underlying difficulties rather than direct obstructions. First, some stakeholders were frustrated with CPLC’s ability to respond to the needs of the outreach workers. Second, the relationship between the stakeholders has, at times, been less than ideal. Third, the concept of community and the 71 selection of the target area have raised some interesting theoretical issues surrounding the ability to replicate the Chicago CeaseFire model. Each of these is discussed below in more detail. Several stakeholders stated that they were not satisfied with CPLC’s ability to provide timely support for the project. From within CPLC, outreach workers and violence interrupters expressed frustration with the levels of bureaucracy required to get minor tasks accomplished. A clear example of this was when new business cards were printed, but it took months for the payment to be made to the printer and for the cards to be released to the outreach workers. One employee stated, “We’ve been asking [CPLC] for business cards so I can just pass these cards out to different people in the neighborhood so they can get help, you know what I mean, been waiting on those for the longest time.” Another worker expressed the need for bus passes for clients to get to work. They explained that the outreach workers go to great lengths to help clients get jobs, but then the clients need assistance getting to their jobs while they are waiting for their first paycheck. Expressing his discontent with bureaucratic processes, one employee explained: … we need bus passes, I buy them out of my own money. Because if I got you a job, you just starting, you don’t have money to buy a two week bus pass. And we ask for them, and ask for them and we don’t get them, I’ve been asking for over a year, so I started buying them myself... And [CPLC]’s like “ok, we’re gonna get it, we’re gonna get the bus passes,” we have no business cards, haven’t had business cards in over a year, you know, can we get business cards? Ok, they’re coming, they have to cut the check. You know, ‘cause if I meet you out, you know, “Do you have a business card?” Uh, no, but here’s this and I’ll write my name and number on it, you know what I mean, you just don’t feel professional, when you don’t have what you need to work with, and I mean, 72 that’s just a personal thing… we’re not asking a lot, you know what I mean, just a bus ticket, and not everybody, just ten bus tickets a month for like a week, and they have jobs… finally I just went up and bought ’em… I pay for it out of my own pocket, because they need to go to work, and if we can’t get them to work, what’s the purpose? Being responsive to the needs of the frontline workers is important. Relatively, their needs are not great, but meeting those needs is essential in helping them do their jobs and to connect with the clients, who are the primary focus of the project. The second issue that some stakeholders expressed was that of not being satisfied with ASU’s responsiveness in providing updated reports about the project that could, in their view, have helped focus the project. Many of the stakeholders expected ASU to provide immediate feedback on the progress of the TRUCE project, particularly about the clients and the crimes in the target area. At the root of the problem was a misunderstanding of responsibilities by some stakeholders who thought that ASU would have immediate access to police data; it did not. This misunderstanding was, in part, the fault of ASU. A couple of unfunded requests were made by stakeholders, and ASU staff agreed to fulfill those requests. The requests were not completed for one of two reasons. First, ASU needed the appropriate data to complete some of the requests, and often, stakeholders were not forthcoming with the necessary data to complete the request. Or second, on at least one occasion the request was not completed because it was forgotten. Our data indicated that the lack of communication and understanding that existed, at times, was the consequence of not having a strong advisory board. Without a strong strategic direction in the TRUCE project, many stakeholders were left to create their own idea of the 73 strategic direction of the project, and many times these ideas were not compatible with one another. Additionally, a strong advisory board would have increased the accountability of all those involved in the project and would have clarified the direction of the project. Finally, the third issue that impeded the implementation of the Phoenix TRUCE project was the selection of the target area. The selection of the target area raised a number of issues about the concept of community and the selection of an appropriate target area for the CeaseFire model generally. As the TRUCE project progressed, the issue about the appropriateness of the target area became a source of contention between the stakeholders in Phoenix and those who were providing technical assistance from Chicago. Originally, the grant proposal identified four areas in South Phoenix that would be targeted by the CeaseFire model; however, the technical advisors from Chicago required that only one area be selected. At the heart of the issue was the question: by what factors are sites to identify and select an appropriate target area? Related, how does a city’s concept of community play a role in the implementation of the CeaseFire model? Many of the Phoenix stakeholders felt the target area was too small and that in order to better address the violence problem in the area it should be expanded. One stakeholder expressed, Out here in Phoenix, South Phoenix, you can take a bus for a long time and not worry about anything or walk blocks and not worry about anything, there is gangs yes, there’re different gangs, different sets but at least, it’s not so congested. It’s not congested and that’s one of my biggest, I always debate when it come to this, it’s like, it’s not the same as [Chicago]. 74 On the other hand, stakeholders from Chicago expressed the need to have a target area that is geographically small enough to walk, but has a high concentration of gun incidents. One technical advisor from Chicago stated: [CeaseFire is] predicated on having a large enough concentration, you know, of a problem in order to impact that problem. And the model that we use, you have to get to know the people who are most closely associated with the problem and have some sort of effect on them and for those people that you really are looking to change, or help change over the long term, that requires, we think, fairly intensive contact. You know, three face-to-face a month, three home visits a month, daily phone contact or frequent phone contact. You know, it is pretty intensive and it becomes more difficult to do that when you are chasing all over the place. So from a practical point of view, we think it is difficult to do the model [if the area is too large]. The concept of a self-defined community that is walkable and has the concentration of incidents equal to communities in Chicago does not exist in Phoenix. Space looks different, and the density of residents and gun crime is much lower. In the end, given the requirements of the Chicago technical advisors, most of those involved in the Phoenix TRUCE project questioned whether the CeaseFire model could be adapted to cities that look and feel so different than Chicago. As one Chicago technical advisor stated, “…it may be that this is, either the target area wasn’t the right target area, or it could be that maybe this model isn’t appropriate for this city. That’s possible too.” Some adjustments might need to be made to accommodate cities with different geographical space, from the size of the target area to the time of day and week that outreach staff routinely work. One Phoenix stakeholder summarized this by saying, “Do I feel that we can make it like CeaseFire? Yes, we can. Do I feel like, that we can make it with the 75 same concept they have? No, I don’t. No, I don’t. We gotta, I feel that Phoenix, it’s a complete different city [than Chicago].” 76 IMPACT EVALUATION IMPACT EVALUATION METHODOLOGY The impact evaluation relied on a quasi-experimental methodological strategy known as a pre-test/post-test nonequivalent control group design. This methodology relies on a target area where the intervention takes place and a comparison area, which is purposely matched as closely possible to the targeted area. Data were collected from target and comparison areas before and after program implementation. We identified comparison areas by constructing a disadvantage index score for all census blocks in Phoenix. This allowed us to identify areas of the city that were most comparable to the targeted area. After we scored each block group, census, police, and hospital data were used to identify five contiguous block groups (the size of the target area) that best matched our target area. As a result, we identified three comparison areas. Exhibit 32 presents information on the characteristics of each area, and Exhibit 33 shows the location of the comparison areas in the city. The comparison areas were matched well with the target area with respect to such neighborhood characteristics as population, number of households, vacant housing, household income, police incident data for violent crime, and ER visits for violence. However, there was one important difference between the target and comparison areas. We were unable to identify any other comparison area with a similar ethnic composition as our target area. About 40% of the residents in the target area were African-American compared to 10, 6, and 4% of those in the comparison areas. 77 Exhibit 32: Characteristics of Target and Comparison Areas Area Target Area A Total Population* Total Households Total Housing Units Vacant Housing Units % of Units Vacant %White % Black % Hispanic Median Household Income 11,771 3,220 3,322 102 3 22 40 54 $34,426 11,749 3,303 3,581 278 8 45 4 65 $38,447 11,294 2,862 2,986 124 4 57 10 82 $35,490 12,519 3,018 3,172 154 5 49 6 81 $33,961 551 781 790 537 7 17 13 4 Total Violent Incident Reports 5/09 thru 5/10 Total Violent Emergency Room Visits (ER) 5/09 thru 5/10 *Based on ESRI 2008 Census Estimates Area B Area C 78 Exhibit 33: Target and Comparison Areas for the TRUCE Project We measured the impact of the project using two sources of data. The first source of data used to measure impact was incident report data from the Phoenix Police Department (PPD). The PPD provided their Computer Aided Dispatch (CAD) and Record Management System (RMS) data.5 These included all police incident reports from January 1, 2007 through December 31, 2011 (60 months) for the target and three comparison areas. Our analyses focused on three incident types aggregated monthly: shootings, assaults, and all violent incident reports. 5 An earlier confidential draft interim report of this evaluation used a misnomer of “calls-for-service” when a more accurate description of the police data used for the analysis in the report would be incident data. 79 The second source of data used to measure the impact of the project was firearm injury data provided through the Arizona State Trauma Registry (ASTR). These data were used to examine changes over time in the number and rate of firearm related injuries and deaths, and served as an alternate measure of crime. The data were coded through “E-Codes,” which categorize injuries. For example, different codes provide information about whether the injury was accidental or intentional, as well as the type and circumstance. Similar to the police incident report data, variables were constructed to measure the number of shootings, assaults, and all forms of violent, intentional injury. These data were in turn aggregated for monthly totals. For detailed maps of the target area and the comparison areas, see Appendix C. ANALYTIC STRATEGY We first assessed the impact of the TRUCE project on shootings, assaults, and overall levels of violence6 through police and hospital data. We examined change by comparing the police data 41 months pre-implementation to the 19 months post-implementation and hospital admissions 6 months pre-implementation to the 6 months post-implementation. The Arizona State Trauma Registry (ASTR) data is compiled and available for confidential external dissemination biennially. Given this availability restriction and the timing necessary to conduct analyses for the present study, the hospital data (ASTR) were only available for six months of the post implementation period. We limited the pre-implementation hospital data to six months to match the post-implementation restriction. Police data consisted of dispatched calls 6 Homicides were not examined separately from other forms of violence due to the small number of homicides that occurred in the target area (n=4) and each of the three comparison areas (n=3, 2 & 2) during the analysis period. 80 for service (CFS), officer-initiated events, and callbacks, all of which were derived from the Phoenix Police Department’s CAD/RMS system. Incidents recorded by the police were coded into one of three designated categories (i.e., assaults, shootings, all violent crime) or were removed from the analysis. Incidents described as “shootings” or “shots fired” were compiled for our shooting measure. Our assault measure was created by aggregating the following incident types: assault, attempted assault, aggravated assault, attempted aggravated assault, cutting or stabbing, attempted cutting or stabbing, subject threatening, and fighting. The measure for all violent incidents included the incidents described above as well as the following: purse snatch, attempted purse snatch, strong armed robbery, attempted strong armed robbery, armed robbery, attempted armed robbery, subject with a gun, misuse of weapon, and homicide. Second, we use a number of time-series models that allowed us to assess the effects of different forms of “dosage” on the outcome, while controlling for the trends in the comparison areas. Specifically, we used pooled cross-sectional time-series analysis. This type of analysis, for example, allows for the fact that the number of police incident reports in one month is related to the number of calls in the previous month (temporal autocorrelation). Additionally, the model allows for the fact that the target area and comparison areas might all be affected in the same way in a given month. Across all models, we control for changes in crime that might be seasonal (by month). This longitudinal, multivariate technique allowed us to isolate the program effects as best possible. For more information on this method of analysis, please see Podesta (2000). 81 TRENDS IN POLICE INCIDENT REPORTS AND EMERGENCY DEPARTMENT DATA We first examined the impact of the TRUCE project on violence and shootings in the target area by assessing the within-area trends in the outcome data. This section examines whether there were any significant changes in the average number of police incident reports and hospital visits before and after the implementation of the TRUCE project. The exhibits below display the outcome data from the target area and the three comparison areas. The time points are 6-month averages to smooth out the lines, and the vertical dashed line indicates the point when TRUCE was implemented. Exhibit 34 shows the trend of the number of police incident reports that were made for shootings in the target area and the three comparison areas. The graph indicates that all four areas experienced a downward trend from June 2007 to about December 2009. This trend does, however, appear to have changed and all four areas have seen increases since June 2010, although during the second half of 2011 the trend leveled-off in the target area and declined in one of the comparison areas (area C). The target area had the highest number of police incident reports for shootings. In order to test whether or not there was a significant change between the 41 months prior to implementation and the 19 months since, we conducted a t-test for difference of means for each area. Exhibit 37 below further displays these results. There were no significant differences in the number of police incident reports for shootings between the 41 months pre-implementation and the 19 months post-implementation. This was true for the target area and the three comparison areas. 82 Exhibit 34: Monthly Incident Reports for Shootings 14.0 Six-Month Avarage Calls 12.0 10.0 8.0 6.0 4.0 2.0 0.0 Jun-07 Dec-07 Jun-08 Target Dec-08 Jun-09 Area A Dec-09 Area B Jun-10 Dec-10 Jun-11 Dec-11 Area C Exhibit 35 displays the number of police incident reports (6-month averages) related to assaults in the target area and the three comparison areas. Exhibit 37 (further below) shows that while there was a significant increase in assaults in comparison Area A, there were no significant changes for the other three areas. 83 Exhibit 35: Monthly Incident Reports for Assault 80.0 Six-Month Avarage Calls 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 Jun-07 Dec-07 Jun-08 Target Dec-08 Jun-09 Area A Dec-09 Area B Jun-10 Dec-10 Jun-11 Dec-11 Area C Exhibit 36 displays the trends in all violence-related police incident reports. In all four areas, the trend started to increase about the time TRUCE was implemented. Exhibit 37 shows that there were no significant differences between the mean number of violence-related police incident reports from the 41 months pre-implementation to the 19 months postimplementation. Both the target area and comparison area B experienced decreases in violence-related police incident reports, although neither was statistically significant. 84 Exhibit 36: Monthly Incident Reports for Violence 80.0 Six-Month Avarage Calls 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 Jun-07 Dec-07 Jun-08 Target Dec-08 Jun-09 Dec-09 Area A Jun-10 Area B Area A Area B Area C Assaults Target area Area A Area B Area C All violent Target area Area A Area B Area C Jun-11 Dec-11 Area C Exhibit 37: Change in Police Incident Reports Pre-Post Truce Implementation Pre-TRUCE Post-TRUCE Mean SD Mean SD Shooting Target area Dec-10 Change t-test sig 4.00 3.26 4.68 2.32 (2.24) (2.75) (2.85) (1.97) 5.95 4.00 4.79 2.63 (4.22) (2.24) (2.39) (1.54) 1.95 0.74 0.11 0.31 0.084 0.370 0.903 0.585 38.84 59.47 57.79 40.63 (8.03) (8.97) (8.28) (5.03) 36.11 67.16 56.16 42.47 (6.91) (13.67) (10.27) (8.02) -2.73 7.69 -1.63 1.84 0.268 0.048 0.593 0.402 46.95 70.79 70.11 46.05 (9.27) (10.32) (11.93) (5.94) 45.68 77.84 66.16 47.63 (8.89) (15.21) (11.83) (8.76) -1.27 7.05 -3.95 1.58 0.671 0.103 0.313 0.520 Note: Pre-TRUCE refers to the 19 month mean prior to implementation, the Post-TRUCE value is the mean of the 19 months after implementation. 85 Overall, there was only one significant change in police incident reports within area for any of the four areas (target area and three comparison areas), and that was an increase in assault incident reports in area A. Given the lack of meaningful within-area significant differences, we did not compare between areas. Next, we examined whether or not there were any significant changes within the four areas with respect to the hospital trauma data. Exhibit 38 shows the trend (6-month averages) in the number of hospital trauma incidents that were caused by shootings. Even with the 6month averages, there is a great deal of variation between periods. This is largely because the number of incidents is low within each area. All the areas had less than one shooting incident per month that resulted in a hospital trauma incident. Exhibit 41 shows the mean differences in the number of incidents between the six months pre-implementation and the six months postimplementation. While there were no significant differences in the number of shooting incidents, comparison area C experienced a marginally significant (p=.078) increase in shooting incidents. 86 Exhibit 38: Number of Monthly Shootings Reported in Emergency Rooms Six-Month Avarage Injuries 1.2 1.0 0.8 0.6 0.4 0.2 0.0 Jun-07 Dec-07 Jun-08 Target Dec-08 Area A Jun-09 Area B Dec-09 Area C Jun-10 Dec-10 87 Exhibit 39 shows the trend in hospital trauma incidents related to assaults in the target area and the three comparison areas. The graph displays the 6-month smoothed average. The target area experienced a decrease in assaults; however, the average from pre-implementation to post-implementation was not significantly different. Comparison area C experienced a significant increase in the number of assaults, but comparison areas A and B were unchanged. Exhibit 39: Number of Monthly Assaults Reported in Emergency Rooms Six-Month Avarage Injuries 2.5 2.0 1.5 1.0 0.5 0.0 Jun-07 Dec-07 Target Jun-08 Dec-08 Area A Jun-09 Area B Dec-09 Area C Jun-10 Dec-10 88 Exhibit 40 displays the monthly averages of the total number of violent injuries based on hospital trauma data for the target area and the three comparison areas. The target area was the only area that experienced a noticeable decrease in violent injuries. While the finding is substantively important, the mean differences between the six months pre-implementation and six months post-implementation is non-significant (see Exhibit 41). The only area to see significant differences in the total number of violent injures was comparison area C, which went from a monthly average of 0.14 incidents to almost 2.5. There was no significant difference between comparison areas A and B between the pre-post implementation periods. Exhibit 40: Number of Monthly Violent Incidents Reported in Emergency Rooms Six-Month Avarage Injuries 3.0 2.5 2.0 1.5 1.0 0.5 0.0 Jun-07 Dec-07 Target Jun-08 Dec-08 Area A Jun-09 Area B Dec-09 Area C Jun-10 Dec-10 89 Exhibit 41: Change in Hospital Admissions Pre-Post Truce Implementation Pre-TRUCE Post-TRUCE Mean SD Mean SD Shooting Target area Area A Area B Area C Assaults Target area Area A Area B Area C All violent Target area Area A Area B Area C Change t-test sig 0.14 0.14 0.57 0.00 (0.38) (0.38) (0.53) (0.00) 0.29 0.29 0.29 0.43 (0.49) (0.49) (0.49) (0.53) 0.14 0.14 -0.29 0.43 0.552 0.552 0.317 0.078 0.57 1.14 0.71 0.14 (0.79) (1.07) (1.50) (0.38) 0.43 0.86 0.57 2.00 (0.53) (0.90) (0.53) (1.41) -0.14 -0.29 -0.14 1.86 0.698 0.598 0.816 0.013 0.71 1.29 1.29 0.14 (0.95) (1.38) (1.38) (0.38) 0.71 1.14 0.86 2.43 (0.76) (1.07) (0.69) (1.27) 0.00 -0.14 -0.43 2.29 1.000 0.832 0.477 0.003 Note: Pre-TRUCE refers to the 7 month mean prior to implementation, the Post-TRUCE value is the mean of the 7 months after implementation. TIME-SERIES ANALYSIS We controlled for trends in the data, both within the target and comparison areas, through time-series analysis. Different models were run to assess the overall effect of the TRUCE program and to assess the impact of program dosage, attention to risk reduction, and client characteristics. Time-series analysis helps us assess which activities might have the greatest impact on overall violence, shootings, and assaults. Since we only have six months of available hospital post-implementation data and the incident counts were low, the time-series analysis was limited to the police incident report data. The analyses presented below are for the programmatic effects only. Each effect corresponds with a unique cross-sectional time-series model, all of which can be found in 90 Appendix D. The full models control for the levels in the comparison areas, the trends in the data over time, and any seasonal effects that could be related to higher police incident reports during specific months of the year. Exhibit 42 displays the effects related to the dosage of TRUCE implementation by month. The first effect, Truce Implementation, is simply a dichotomous variable for the target area post-implementation. This tells us that the implementation of the TRUCE project was related to a significant increase in shootings, and a significant decrease in assaults. Specifically, the program implementation corresponded to an increase of 3.2 shootings and a decrease of more than 16 assaults per month. The reduction in our overall measure of violence was largely driven by the assault category. We should point out that there were two months in the police incident report data (November and December 2010) that were abnormally high in terms of shootings; however, even after removing these two months from the analysis, the direction and the significance of the findings remained the same. In terms of specific dosage levels, we found that the number of active clients per month and number of contacts with clients were significantly related to increases in shootings. One should note, however, that these increases were small, the largest being an increase of 0.07 shooting incident reports per month. Differently, for every conflict that was mediated by TRUCE staff, we found a decrease of almost two assaults (1.9) and a decrease of almost three violent crimes (2.7) overall. Additionally, each conflict mediation had a specific number of people involved, and the number of people involved mattered. Specifically, for each additional person who was involved in a mediated conflict there was about a fifth of a call reduction in assault (0.17) and a fifth of a call reduction for all violent (0.20) incident reports per month. 91 Exhibit 42: Summary of program dosage effects by police incident report type Predictor Shooting Coef. SE Assault Coef. TRUCE Implementation 3.201 * (1.304) -16.436 Number of clients 0.074 * (0.037) Number of contacts 0.007 * Number of home contacts 0.023 Number of street contacts 0.039 Hours spent with clients SE (5.133) -16.575 -0.275 (0.160) -0.205 (0.184) (0.004) -0.023 (0.014) 0.023 (0.017) -(0.012) -0.073 (0.054) -0.068 (0.063) (0.029) -0.159 (0.094) -0.202 (0.112) 0.001 (0.001) -0.005 (0.003) -0.005 (0.003) Conflicts mediated 0.275 (0.268) -1.901 ** (0.721) -2.66 ** (0.846) Number of people in conflicts 0.028 (0.032) -0.170 * (0.074) -0.200 * (0.089) * p<.05; **p<.01 ** All Violent Coef. SE ** (6.362) 92 Exhibit 43 shows the summary of effects of client selection on violent crime. The analysis indicated that a number of the client characteristics were related to slight, but significant, increases in shootings. Specifically, whether they were female, increases in the number of core gang member clients, clients with a prior criminal history, and high risk clients were all significantly related to increases in shootings. Differently, there were four client characteristics that were significantly related to reductions in assaults. An increase of one active gang member client reduced assaults by 3.3 per month. An increase in one client with recent violent victimization experience reduced assaults by one per month, having an additional client in the target age range corresponded to a reduction of more than 2.3 assaults per month. Finally, an increase of one client per month who had been recently released from prison was related to a reduction of 5.7 assaults per month. 93 Exhibit 43: Summary of client selection effects by police incident report type Shooting Predictor Coef. Male clients 0.093 Female clients 0.290 Gang involved clients 0.588 Core gang member clients 0.130 Clients with prior criminal history 0.090 High risk clients 0.089 Recent victims of violence SE Assault Coef. All Violent SE Coef. SE (0.051) -0.336 (0.211) -0.253 (0.245) (0.127) -1.036 (0.554) -0.777 (0.639) (0.357) -3.329 (0.865) -3.795 * (0.064) -0.388 (0.273) -0.270 (0.314) * (0.045) -0.351 (0.191) -0.258 (0.222) * (0.044) -0.323 (0.188) -0.244 (0.217) 0.187 (0.114) -0.973 * (0.474) -0.964 (0.543) Clients aged 16-25 0.291 (0.303) -2.334 ** (0.849) -3.396 ** (0.994) Clients with recently incarcerated 0.863 (0.784) -5.654 ** (1.907) -5.999 ** (2.332) * p<.05; **p<.01 * ** ** (1.042) 94 Exhibit 44 shows the effects of the variation in risk reduction activities on the number of shootings, assaults, and overall violent incidents. Again, a number of activities were significantly related to increases in shootings. Two of the activities, unemployed clients (0.38) and referred to education program (0.55), were significantly related to an increase of about one half of a shooting per month. Interestingly, for every additional client that was referred to anger management, there was a reduction of more than four shootings per month. This effect was not significant largely because there were so few referrals to anger management. Discussing problem solving with non-violent means (0.01), discussing employment needs (0.05), discussing educational needs (0.04), and assisting with job preparation (0.10) were all significantly associated with increases in shooting incident reports, although none were substantively influential. The only risk reduction activity that significantly reduced the number of assaults was discussing educational issues with clients. 95 Exhibit 44: Summary of risk reduction effects by police incident report type Shooting Predictor Coef. SE Assault Coef. All Violent SE Coef. SE Unemployed clients 0.382 ** (0.149) -0.637 (0.628) -0.214 (0.727) Discussed problem solving 0.013 * (0.007) -0.042 (0.024) -0.044 (0.029) Mediated client conflict 0.004 (0.184) -0.641 (0.464) -0.714 (0.552) (3.039) 2.503 (7.269) -0.118 (8.965) (0.019) -0.086 (0.064) -0.057 (0.077) (0.220) 0.011 (0.678) 0.293 (0.805) (0.043) -0.205 (0.130) -0.104 (0.157) 0.046 (0.024) -0.116 (0.074) -0.124 (0.090) -0.118 (0.899) 0.052 (2.222) -0.874 (2.664) (0.048) -0.127 (0.541) 0.175 Referred clients to anger management -4.237 Discussed employment 0.047 Assisted with job readiness 0.341 Assisted with job preparation 0.097 Discussed substance abuse Referred to substance abuse treatment * * Discussed educational issues 0.036 ** (0.014) -0.131 Referred to education program 0.546 ** (0.197) -0.376 * p<.05; **p<.01 ** * (0.058) (0.646) 96 CONCLUSIONS AND RECOMMENDATIONS In the present study we examined the impact of the Phoenix TRUCE project on neighborhood levels of violence. The Phoenix TRUCE Project was modeled after the Chicago CeaseFire program, and as such adopted a public health approach in responding to violence in the community. First, we examined the fidelity of program implementation and analyzed data from a number of sources including interviews of TRUCE staff and other stakeholders, a randomized household survey, and the Chicago Project for Violence Prevention Evaluation Database. Second, we examined the impact of the project through a pre-test/post-test nonequivalent control group design. Specifically, we examined data from the targeted neighborhood and three comparison neighborhoods before and after program implementation. Below, we summarize and discuss our findings and provide some recommendations to be considered for adoption by the Phoenix TRUCE Project in the future. PROCESS FINDINGS Process measures indicated that the Phoenix TRUCE project was actively involved in increasing community awareness about the costs of violence to enlist broad-based community support for changing attitudes and behaviors related to violence. More than 188 community activities and events were held over the project period. The activities and events included community gardens, canvassing the neighborhood, holding barbeques, and other client-focused activities. Furthermore, more than 11,000 public education materials were distributed throughout the project. For example, 4,000 door hangers, 1,500 wrist bands, and 1,100 t-shirts were distributed. One post-test randomized in-person household survey indicated that a 97 sizeable portion of the community was aware of TRUCE. About 13% of neighborhood residents knew of TRUCE, and about 25% of those in the neighborhood who knew someone who was shot knew about TRUCE. However, one major problem related to the implementation of Project TRUCE was a lack of community mobilization. While some stakeholders met frequently, at least once a month, a community-wide advisory board was never established. This resulted in lack of strategic direction for the program and reduced the program’s capacity to mobilize larger institutions within the community. Similarly, interviews indicated that coordination and collaboration between TRUCE and the faith-based community was largely absent from the project. This necessarily resulted in fewer opportunities for clients to access jobs and services, and a reduced the capacity of TRUCE to present its message of anti-violence. The use of violence interrupters and conflict mediation was consistent with the Chicago CeaseFire model. Violence interrupters distinguish the CeaseFire model from other communitybased crime prevention efforts. TRUCE selected interrupters based on important characteristics of personal experience and knowledge in a manner consistent with the interrupters used in Chicago, and as expected by the model. Nearly three-quarters (n=42, 72.4%) of all conflicts mediated came to the attention of TRUCE through personal contacts and connections in the community known to individual violence interrupters. This finding demonstrates the requisite knowledge of and connection to the community. Additionally, violence interrupters reported that 47 of the 58 (90%) conflicts mediated resulted in at least a temporary resolution to the conflict, indicating the sufficient legitimacy of the violence interrupters to fulfill their role. 98 Finally, the findings indicate that there was substantial program fidelity in that TRUCE workers selected the appropriate high-risk individuals on whom to focus their attention. As previously stated, disputants were typically gang-involved (n=41, 70.7%), historically violent (n=37, 63.8%), and young (n=53, 91.4%), three criteria the model uses to define individuals as high-risk for gun violence involvement. Other sites (see Skogan et al., 2008; Webster et al., 2009; Wilson et al., 2011) did not report exact percentages of risk factors associated with conflict mediation disputants, so direct comparison cannot be made here. The Chicago evaluation did discuss the nature of the outreach work and conflict mediations as involving those who were young, violent gang members; however, it did not report specific frequencies (Skogan et al., 2008). Limitations to these findings are not dissimilar from those found in the Skogan et al. (2008) evaluation of the Chicago CeaseFire model. The Phoenix TRUCE violence interrupters used the conflict mediation database developed for the model. From the beginning, TRUCE staff were trained to use the conflict mediation forms to record their activities. However, from interviews and reviews of the database, evaluators found that there may have been some of the same shortfalls with the quality of the data – that is, not all conflicts and mediations were recorded, and some of those recorded may not have been very likely to result in a shooting. TRUCE used the conflict mediation forms from the onset of implementation (which was different than in Chicago) and should have contributed to a better overall quality to the data. In any case, these limitations do not indicate any substantive shortfall in the implementation of the violence interrupters and the use of conflict mediations as intended by the programmatic model. 99 The recruitment and enrollment of clients for the Phoenix TRUCE project was consistent with the guidelines and expectations of the Chicago CeaseFire model. All of the clients (n=118) enrolled were assessed as being at high-risk for involvement in gun violence. Also consistent with the model, clients were typically between the ages of 16 and 25, gang-involved, had a history of involvement in violent crimes, and were currently involved in high-risk street activity. Further, consistent with the CeaseFire model’s intervention plan through diversion to services for life skills, education, and employment, TRUCE clients were substantially under-educated and under-employed. A review of the client management database with respect to the quantity and quality of case management indicated the Phoenix TRUCE project was implemented with a great deal of fidelity during the project. There was an average of more than 300 contacts per month between outreach workers and clients, of which on average 42.4% (mean=127.6, n=2,297) were conducted face-to-face in the client’s home or on the street in their community. On average outreach workers made 45.9 contacts per client. During the contacts, outreach workers reported having discussed resolving conflicts without violence in nearly half of all contacts (49%, n=2,621). Outreach workers also identified that employment needs were common among clients. For example, during more than half of all contacts (52.4%, n=2,787), clients reported that they were actively looking for work, whether or not they were currently employed. The challenge of unemployment and under-employment is an important concern for TRUCE participants. Issues related to both education and probation/parole compliance was less commonly discussed, as was substance abuse issues 100 (which was also a rarely identified need). In summary, both the frequency and general tenor of the contacts demonstrated a good dosage of client outreach effort. Outreach workers were expected to use the Risk Reduction Plan to track a client’s progress toward reducing their high-risk classification. This allows for clients to “graduate” from the program and for new clients to be enrolled. This is an important part of the project for both the diffusion of benefits and overall impact on changing norms. The implementation of this component was not consistent with the model and was only partially implemented. During the refresher training conducted in May 2011, trainers impressed upon TRUCE personnel the importance of every client having a Risk Reduction Plan in place in a reasonable time frame after intake, and that an explanation would be expected if there were an undue delay. This modification would help the integrity of the implementation of the Phoenix TRUCE site to be in compliance with the CeaseFire model regarding client management and risk reduction efforts. Of the 33 (28%, n=118) open clients at the time of analysis, 51.5% (n=17) of them had a documented Risk Reduction Plan. The 16 active clients without a documented Risk Reduction Plan were ongoing for an average of more than nine months, ranging between 108 and 517 days from the time of their intake until December 31, 2011, the end date for the analysis of this report. In the end, only 43 (36.4%) of the clients had a documented Risk Reduction Plan. Criminal justice participation in TRUCE was largely limited to local police. The police shared information with TRUCE staff about neighborhood violence, provided security during events, attended partner meetings, and were generally supportive of requests by partners. As expected, outreach workers and violence interrupters kept a healthy distance from the police in public. This was in part because they were concerned about their street image. Specifically, 101 they were concerned that their clients would not trust them if they knew that they talked to the police. Regardless, the police fulfilled all expectations asked of them, and most importantly they often shared information with TRUCE, which allowed TRUCE to target clients for services. IMPACT FINDINGS We examined the impact of the TRUCE project on shootings, assaults, and overall levels of violence through police and hospital data. We examined change by comparing the police data 41 months pre-implementation to the 19 months post-implementation and hospital admissions six months pre-implementation to the six months post-implementation. We further examined the data through pooled cross-sectional time-series analysis, which allowed us to examine the impact of different services on the outcome, while at the same time controlling for the trends in the comparison areas. T-test showed that there was no significant change in police incident reports for shootings, assaults, and overall levels of violence from the pre-TRUCE period to the post-TRUCE period in the target and three comparison areas. Similarly, when we examined hospital admissions pre-post TRUCE implementation, with the exception of comparison area C, we did not find a significant change in hospital admissions for shootings, assaults, and overall levels of violence. In comparison area C, we found that assault-related hospital admissions significantly increased by about 1.9 admissions per month and hospital admissions for overall levels of violence significantly increased by about 2.3 admissions per month. Different results were identified after controlling for trends in the three comparison areas through time series analysis. Time-series analysis indicated that program implementation corresponded to a decrease of more than 16 assaults on average per month, controlling for the 102 comparison areas and the trends in the data. The time-series analysis indicated that the more conflicts mediated and the more people involved in mediated conflicts, the greater the decreases in assaults. Additionally, time-series analysis indicated that the project is related to decreases in assaults when caseload clients are gang-involved, aged 16-25, recent victims of violence, and recently released from prison. Our analysis, however, also showed that program implementation was associated with an increase in shootings. Specifically, time-series analysis indicated that program implementation corresponded to an increase of 3.2 shootings, on average, per month, controlling for the comparison areas and the trends in the data. The significant increase in shootings was related to an increase in the number of clients and the total number of contacts (both home and on the street). We also found that a number of the client selection effects were related to the slight but significant increase in shootings in the time-series analysis, including clients who were gang members, clients with prior criminal history, clients who were involved with high-risk street activity, and female clients. LIMITATIONS Prior to the interpretation of the findings being complete, at least five limitations to the evaluation should be noted to help contextualize the results. First, the Phoenix TRUCE project was implemented in an African American community by a primarily Hispanic serving organization. While this was not discussed as a problem by program partners or field workers (and as such might not have been a problem), we nevertheless wondered whether this might have had an impact on CPLC’s organizational capacity to more fully mobilize the community (i.e., advisory board). Second, the evaluation uncovered several problems associated with 103 replicating the Chicago Ceasefire project in Phoenix—two communities that use geographic space very differently. Chicago protocols and technical advisors called for the program to be implemented in a geographically small and dense (both with respect to population and gun violence) neighborhood. The type of neighborhood described by the Chicago technical advisors simply does not exist in Phoenix. This led many to question whether, if such density is required for program implementation, the program can be replicated with fidelity in many western cities, including Phoenix. Third, and related, we were unable to examine the impact of the Phoenix TRUCE project on homicides in general or retaliatory homicides in particular because of the low volume of homicides in the target area. This issue necessarily required us to restrict our impact evaluation to shootings and assaults. Fourth, while the target area was predominately inhabited by Hispanic residents, it also included a large proportion of African American residents. In fact, we were unable to find any other community in the county with as large an African American community as that found in the target area. As a consequence, while our comparison areas were well matched in many ways, they were not well matched with respect to the ethnicity of residents. Last, we would have liked to have had the opportunity to examine the hospital data over a longer period of time, as we did the police data. While the findings were similar across police and hospital data (at least for the 12-month period both data were available), hospital data has the potential to serve as an important independent source of information that might not otherwise be available. 104 RECOMMENDATIONS Our monitoring and evaluation of the Phoenix TRUCE project leads us to make two recommendations. First, we strongly recommend that the project enhance its standing within the community. For example, little effort has been placed on engaging the faith-based community, a key stakeholder in the area where the project is based and a core component of the Ceasefire model. Phoenix TRUCE also needs to enhance its relationship with the police. Currently there is no systematic means of transferring intelligence or information pertaining to shootings from the police to project staff. At a minimum, the police and Phoenix TRUCE should work together to create a system to share police incident reports on shootings. This information could be used to foster communication between the two organizations and should be used to identify micro-hotspots to be targeted by outreach staff. ASU provided several examples of micro-hotspot maps, which were provided to the outreach workers. Additionally, the project needs to re-establish the Advisory Board to help make strategic decisions and to leverage funding for the sustainability of the program. The creation of a strong, consistent Advisory Board should be the number one priority moving forward. Second, we recommend that Phoenix TRUCE increase communication between line staff and senior management. Executives at Chicano Por La Causa should have higher expectations of managers with respect to holding line level staff accountable. For example, we repeatedly found that client and conflict mediation files were not routinely updated. Senior management should ensure that the client and conflict mediation files are updated on at least a weekly basis. This is the main source of information needed to manage cases and to evaluate implementation. It is imperative that these data be as accurate as possible for the purpose of 105 monitoring field activity and ensuring programmatic fidelity. We also believe that managers should re-train outreach workers on the Risk Reduction Plan and establish a system where the outreach supervisor reviews the Risk Reduction Plan weekly or bi-monthly to ensure compliance. This is imperative if plans are to be carried out effectively. Executives should also set up a system through which outreach staff can make suggestions and requests that will be addressed in a timely manner. We found that simple and inexpensive requests by outreach workers were falling on deaf ears, resulting in their jobs being made more difficult. Executives should establish a communication system to ensure that the interaction between the staff and clients is the number one priority. 106 REFERENCES Boyle, D. J., Lanterman, J. L., Pascarella, J.E., and Cheng, Chia-Cherng. (2010). The impact of Newark’s Operation Ceasefire on trauma center gunshot wound admissions. Justice Research and Policy, 12(2), 105-123. Bursik, R. J., JR and Grasmick, H. G. (1993). The use of multiple indicators to estimate crime trends in American cities. Journal of Criminal Justice, 21, 509-516. Chaskin, R. (2010). Youth Gangs and Community Intervention. Columbia University Press. New York. Ceasefirechicago.org (2012). Found at Casefirechicago.org in October 2012. Fenton, J. (2012). Safe streets violence mediation program coming to West Baltimore. The Baltimore Sun. Found at http://www.baltimoresun.com/news/breaking/bs-md-cimondawn on October 4, 2012. Green-Mazerolle, L. J., Ready, J., Terrill, W., and Waring, E. (2000). Problem Oriented Policing in public housing: The Jersey City evaluation. Justice Quarterly 17, 129-158. Kane, C. (2012). Personal communication on May 3rd, 8th and 16th, 2012. Katz, C.M. and Choate, D. (2006). Diagnosing Trinidad & Tobago’s Gang Problem. Presented at the Annual Meeting of the American Society of Criminology, Los Angeles, California. Katz, C. M., Webb, V. J., and Schaefer, D.R. (2001). An assessment of the impact of quality-oflife policing on crime and disorder. Justice Quarterly, 18:4, 825-876. Katz, C. M., Wallace, D., and Hedberg, E. C. (2011). A longitudinal assessment of the impact of foreclosure on neighborhood crime. Journal of Research in Crime and Delinquency, online December 26, 2011. Kelly, J. (2010). Did ex-cons run a gang from their youth program? Youth Today. Found at http://www.youthtoday.org/view_article.cfm?article_id=3959 on October 4, 2012. Krug, E., Dahlberg, L.L., Mercy, J.A., Zwi, A.B., and Lozano, R. (2002). The world report on violence and health. The Lancet. Volume 360, Issue 9339, 5 October 2002, Pages 1083– 1088. Maguire, E. (2012). Chapter 8: Chicago’s Ceasefire and the community based violence interruption movement. Unpublished manuscript. Washington DC: American University. 107 Papachristos, A. V., Braga, A. A., and Hureau, D. M. (Not Published). Six-degrees of violent victimization: social networks and the risk of gunshot injury. Papachristos, A. V., Meares, T., and Fagan, J. (2007). Attention all felons. Journal of Empirical Legal Studies, 4, 2: 223-272. Perkins, C. (2003). National Crime Victimization Survey, 1993-2001: Special Report. Washington, DC: US Department of Justice, Office of Justice Programs, National Institute of Justice. Podesta (2000). Found at http://duke.edu/~kkk4/Fall_07/Paper/Podesta_Pooled_Time_Series_Cross_Section.pdf on December 14, 2011. Ritter, N. (2009). CeaseFire: A Public Health Approach to Reducing Shootings and Killings. NIJ Journal, 264, NCJ 228386. Sherman, L., Gartin, P., and Buerger, M. (1989). Hot spots of predatory crime: Routine activities and the criminology of place. Criminology, 27, 27-55. Sherman, L. and Rogan, D. (1995). Deterrent effects of police raids on crack houses: A randomized, controlled experiment. Justice Quarterly, Vol. 12, No. 4. Skogan, W. G., Hartnett, S. M., Bump, N., and Dubois, J. (2008). Evaluation of Ceasefire-Chicago. Washington, DC: US Department of Justice, Office of Justice Programs, National Institute of Justice. Webster, D., Vernick, J. S., and Mendel, J. (2009). Interim evaluation of Baltimore’s Safe Streets program. Center for the Prevention of Youth Violence, Johns Hopkins Bloomberg School of Public Health. Baltimore, MD. Webster, D., Whitehill, J., Vernick, J., and Parker, E. (2012a). Evaluation of the Safe Streets Program. Baltimore: Johns Hopkins School of Public Health. Webster, D. Whitehill, J., Vernick, J., and Curriero, F. (2012b). Effects of Baltimore’s Safe Streets Program on gun violence: A replication of Chicago CeaseFire Program. Journal of Urban Health, Online First DOI: 10.1007/s11524-012-9731-5. Wilson, J. M., Chermak, S., and McGarrell, E. F. (2011). Community-based violence prevention: An assessment of Pittsburgh’s One Vision One Life program. RAND Corporation: Santa Monica, CA. Wilson, J. M. and Chermak, S. (2011). Community driven violence reduction programs. Criminology & Public Policy, 10, 4: 993-1027. 108 LIST OF APPENDICES Appendix A: Randomized Household Survey Appendix B: Examples of Public Education Material Appendix C: Pictures of TRUCE Mural Project Appendix D: Detailed Maps of TRUCE Target Area and Comparison Areas Appendix E: Full Cross-Sectional Time-Series Models 96 Appendix A: Randomized Household Survey 97 TRUCE Community Survey Hello, my name is __________. I am working with Chicanos por la Causa and surveying citizens to learn more about perceptions of crime in the neighborhood. All answers are completely anonymous. Would you mind answering a few questions, it will take less than 5 minutes? ADDRESS: 1. Are you currently a resident of this household? gr e re ag is t D ha ly w ng e m ro t St So ha e ew re m ag ee So gr A is D A NO ro ee St YES ng ly 2-6. How much do you agree or disagree with the following statements... 2. I feel safe in my neighborhood. 3. Crime is a serious problem in my neighborhood 4. Violence is a serious problem in my neighborhood 5. Shootings are a serious problem in my neighborhood 6. I frequently hear gunshots in my neighborhood 7-12. Please indicate whether these statements are true for you... 7. I know about the TRUCE program 8. I understand the purpose of the TRUCE program 9. The TRUCE program has made a positive impact in my neighborhood 10. I know how to reach a TRUCE worker if I or a family member needs them 11. Do you personally know someone who has been shot or shot at? 12. Have you ever been shot or shot at? 13. How have you heard about TRUCE (Mark ALL that apply) Outreach Worker Local TRUCE rally Violence Interrupter Community barbecues TRUCE flyer Other community event Specify:_________________ TRUCE billboard I have never hear of TRUCE TRUCE t-shirt, hat, or dog tag TRUCE Poster YES NO 98 TRUCE Community Survey (Spanish) e s u a C a l r o p s o n a c i h C n o c o d n a j a b a r t y o t s E . s e e r b m o n i m , a l o H l e n e n e m i r c e d s e n o i c p e c r e p s a l e d a c r e c a s a m r e d n e t n e a r a p s a n o s r e p a o d n a t s i v e r t n e ︳︳︳︳︳︳︳︳︳ s a n u e m r a t s e t n o c a í r d o P ¿ . s a m i n ó n a e t n e m a t e l p m o c n o s s a t s e u p s e r s a l s a d o T . o i r a d n i c e v . s o t u n i m 5 e d s o n e m r a m o t a a v ? s a t n u g e r p s a t n a u c Direccion: 1. Es usted actualmente un residente de este hogar? A do er cu s a do de er uy acu M s d e r do de ue o ac r do lg ue ac lg A de s e n o i c a r a l c e d o i r a d n i c e v i m n e a o r u g e s o t n e i s e M . 2 ︵ ︶ de s e t n e i u g i s s a l n o c d e t s u á t s e o d r e u c a s e d n e o o d r e u c a e d n a t e u Q . 6 2 … o NO uy M SI 3. Crimen es un problema serio en mi vecindario 4. Violencia es un problema serio en mi vecindario 5. Balaseos son un problema serio en mi vecindario 6. Con frecuencia llego a escuchar baleseos en mi vecindario 7-12. Por favor indique si alguna de esas declaraciones es cierta para usted... 7. Yo tengo conocimiento acerca del programa de TRUCE 8.Yo entiendo el proposito del programa de TRUCE 9. El programa de TRUCE ha hecho un impacto positivo en mi vecindario 10. Yo se como contactar a un trabajador(a) de TRUCE si yo o un familiar mio lo necesita 11. Usted conoce personalmente a alguien que ha sido disparado a han disparado hacia el/ ella? 12. Le han disparado a usted o hacia usted? 13. A usted visto o escuchado algo que le haya dado conocimiento sobre TRUCE (Marque TODOS los que apliquen) Trabajador de Comunidad Marcha/ caminata local de TRUCE Interruptor de Violencia Colgador de quertas de TRUCE Volante sobre TRUCE Barbacoas de comunidad Cartones de TRUCE Otro evento de comunidad Especifique:________________ Camisetas, gorras or placas Nunca he escuchado sobre TRUCE Poster de TRUCE SI NO 99 Appendix B: Examples of Public Education Materials 100 Billboards 101 Poster 102 Postcards 103 Appendix C: Pictures of TRUCE Mural Project 104 Before After 105 Appendix D: Detailed Maps of TRUCE Target area and comparison areas 106 107 108 109 110 111 Appendix E: Full Cross-Sectional Time-Series Models 112 Predictor Definition Male Clients Number of male clients per month. Female Clients Number of female clients per month. TRUCE Implementation 1 for every month after implementation in target area. Number of Clients Number of clients being actively served per month. Number of Contacts Number of contacts with clients in a given month. Number of Home Contacts Number of face‐to‐face type of contact per month. Number of Street Contacts Number to face‐to‐face type of contact per month. Hours Spent with Clients Average number of hours spent with clients per month. Conflicts Mediated Number of conflicts mediated in a given month. Number of People in Conflicts Total number of disputants involved in a given conflict per month. Unemployed Clients Number of clients who need a job (i.e. under‐employed) in a given month. Based on case notes. Problem Solving Discussion Number of times a case worker discussed the use of non‐violent means to solve problems/conflicts/issues etc., per month. Client Conflict Mediation Number of times a case worker mediated a conflict with participant per month. Anger Management Client Referral Number of times a case worker referred participant to an anger management program or life skills program per month. Employment Discussion Number of times per month a case worker discussed employment needs. Job Readiness Assistance Number of times per month a case worker assisted with job readiness (clothing, state IDs, etc.) Job Preparation Assistance Number of times per month a case worker assisted with job preparation (resume applications, etc.) Substance Abuse Discussion Case worker discussed substance abuse with participant. Substance Abuse Treatment Referral Caseworker referred participant to a substance abuse treatment program. Educational Issues Discussion Number of times per month a case worker discussed substance abuse with participant. Educational Program Referral Number of times per month a caseworker referred participant to a substance abuse treatment program. Gang Involved Clients Number of active clients who were gang involved per month as presented during intake. Core Gang Member Clients Number of active clients who played a key role in a gang per month as presented during intake. Clients with Prior Criminal History Number of active clients who had prior criminal history (violence or weapons) per month as presented during intake. High Risk Clients Number of active clients who engaged in high risk street activity per month as presented during intake. Recent Victim of Violence Number of active clients who were a victim of a shooting (within 90 days prior to intake) per month as presented during intake. Clients Aged 16‐25 Number of active clients who were aged 16‐25 years per month as presented during intake. Recently Incarcerated Clients Number of active clients who were recently released from prison per month as presented during intake. 113 Summary of Effects CFS Shooting Predictor Coef. Male Clients Female Clients Truce Implementation Conflicts Mediated Number of People in Conflicts Number of Clients Number of Contacts Number of Home Contacts Number of Street Contacts Hours Spent with Clients Unemployed Clients Problem Solving Discussion Client Conflict Mediation Anger Management Referral Employment Discussion Job Readiness Assitance Job Preparation Assistance Substance Abuse Discussion Substance Abuse Treatment Referral Educational Issues Discussion Educational Program Referral Gang Involved Clients Core Gang Member Clients Clients with Prior Criminal History High Risk Clients Recent Victim of Violence Clients Aged 16‐25 Recently Incarcerated Clients 0.135 0.407 3.312 0.061 0.019 0.105 0.008 0.031 0.057 0.002 0.488 0.015 0.050 ‐4.986 0.038 0.444 0.102 0.031 0.026 0.033 0.495 0.135 0.183 0.129 0.126 0.304 0.111 ‐0.103 * p<.05; **p<.01 ** ** * ** ** ** * * ** * * * ** ** ** ** ** CFS Assault SE Coef. (0.044) (0.104) (1.388) (0.268) (0.031) (0.031) (0.003) (0.012) (0.026) (0.001) (0.122) (0.006) (0.175) (2.849) (0.020) (0.195) (0.052) (0.024) (0.891) (0.016) (0.191) (0.383) (0.054) (0.038) (0.037) (0.096) (0.297) (1.189) ‐0.339 ‐1.128 ‐14.218 ‐1.729 ‐0.157 ‐0.277 ‐0.021 ‐0.071 ‐0.158 ‐0.004 ‐0.813 ‐0.038 ‐0.644 4.156 ‐0.080 ‐0.164 ‐0.227 ‐0.106 0.310 ‐0.133 ‐0.382 ‐3.370 ‐0.396 ‐0.348 ‐0.325 ‐1.009 ‐1.997 ‐8.259 * * * * * ** * * ** CFS All SE Coef. SE (0.208) (0.521) (5.841) (0.771) (0.078) (0.152) (0.014) (0.053) (0.096) (0.003) (0.605) (0.025) (0.487) (7.759) (0.072) (0.694) (0.180) (0.081) (2.463) (0.057) (0.583) (0.945) (0.265) (0.185) (0.181) (0.450) (0.899) (2.917) ‐0.229 ‐0.804 ‐14.962 ‐2.947 ‐0.204 ‐0.191 ‐0.021 ‐0.064 ‐0.200 ‐0.005 ‐0.398 ‐0.042 ‐0.766 2.123 ‐0.075 0.185 ‐0.175 ‐0.136 ‐1.313 ‐0.141 ‐0.087 ‐4.514 ‐0.255 ‐0.237 ‐0.227 ‐0.908 ‐3.446 ‐10.892 (0.243) (0.606) (6.878) (0.874) (0.092) (0.178) (0.016) (0.062) (0.111) (0.003) (0.706) (0.029) (0.568) (9.341) (0.085) (0.810) (0.214) (0.946) (2.902) (0.067) (0.685) (1.096) (0.308) (0.217) (0.211) (0.524) (1.017) (3.451) * ** * * ** ** ** 114 Cross‐Sectional Time‐Series Analysis of the Effect of Male Clients on Calls‐For‐Service CFS Shooting Coef CFS All CFS Assault SE Coef SE Coef SE (0.243) Male Clients 0.135 ** (0.044) ‐0.339 (0.208) ‐0.229 Time ‐0.107 ** (0.015) ‐0.123 (0.087) ‐0.317 0.941 ‐0.961 0.330 0.180 ‐0.243 (0.800) (0.799) (0.794) (0.776) (0.700) ‐2.796 ‐4.528 7.409 4.403 7.847 (3.785) (3.727) (3.602) (3.336) (2.732) ‐3.101 ‐4.419 7.652 5.274 8.172 0.100 0.042 ‐1.261 ‐0.208 0.031 0.868 (0.753) (0.827) (0.841) (0.845) (0.845) (0.844) 2.376 1.867 3.101 2.213 ‐1.382 ‐4.539 (2.977) (3.586) (3.834) (3.944) (3.980) (3.960) 2.642 2.950 2.665 3.577 ‐1.291 ‐4.354 (0.678) 48.989 (3.455) 54.492 * (0.135) * (5.055) (4.935) (4.696) (4.253) (3.378) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 1.971 ** 216 68.32**(13) * ** ** 216 37.05**(13) (3.699) (4.595) (5.015) (5.226) (5.305) (5.280) ** (5.009) 216 28.60**(13) 115 Cross‐Sectional Time‐Series Analysis of the Effect of Female Clients on Calls‐For‐Service CFS Shooting Coef CFS All CFS Assault SE Coef * SE Coef SE (0.521) ‐0.804 (0.606) Female Clients 0.407 ** (0.104) ‐1.128 Time ‐0.112 ** (0.014) ‐0.118 (0.088) ‐0.305 0.864 ‐0.969 0.345 0.106 ‐0.319 (0.777) (0.775) (0.769) (0.754) (0.680) ‐2.702 ‐4.670 7.170 4.536 8.085 (3.832) (3.767) (3.640) (3.377) (2.763) ‐2.998 ‐4.400 7.593 5.398 8.339 0.203 0.205 ‐1.286 ‐0.236 0.022 0.789 (0.732) (0.802) (0.816) (0.819) (0.819) (0.819) 2.187 1.470 3.054 2.050 ‐1.515 ‐4.462 (3.013) (3.630) (3.880) (3.990) (4.028) (4.008) 2.490 2.679 2.650 3.519 ‐1.325 ‐4.242 (0.659) 49.297 (3.496) 54.852 * (0.137) * (5.102) (4.972) (4.729) (4.289) (3.403) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 1.842 ** 216 78.26**(13) * ** ** 216 38.25**(13) (3.727) (4.629) (5.056) (5.268) (5.351) (5.328) ** (5.056) 216 28.91**(13) 116 Cross‐Sectional Time‐Series Analysis of the Effect of TRUCE Implementation on Calls‐For‐Service CFS Shooting Coef CFS All CFS Assault SE Coef * SE Coef SE ‐(5.841) ‐14.962 * ‐(6.878) * ‐(0.134) * ‐(5.044) ‐(4.920) ‐(4.691) ‐(4.264) ‐(3.402) TRUCE Implementation 3.312 * ‐(1.388) ‐14.218 Time ‐0.103 ** ‐(0.015) ‐0.110 ‐(0.086) ‐0.280 1.182 ‐0.666 0.622 0.370 ‐0.105 ‐(0.819) ‐(0.817) ‐(0.812) ‐(0.796) ‐(0.719) ‐3.464 ‐5.565 6.390 3.488 6.936 ‐(3.783) ‐(3.721) ‐(3.603) ‐(3.351) ‐(2.763) ‐3.784 ‐5.395 6.732 4.444 7.350 0.139 0.179 ‐1.048 0.051 0.248 1.146 ‐(0.771) ‐(0.846) ‐(0.861) ‐(0.864) ‐(0.865) ‐(0.863) 2.082 1.336 2.441 1.447 ‐2.014 ‐5.341 ‐(2.982) ‐(3.588) ‐(3.834) ‐(3.940) ‐(3.978) ‐(3.954) 2.432 2.535 2.084 2.844 ‐1.855 ‐5.122 ‐(0.713) 50.268 ‐(3.488) 56.899 Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 1.919 ** 216 61.74**(13) * ** 216 40.09**(13) ‐(3.701) ‐(4.591) ‐(5.009) ‐(5.215) ‐(5.297) ‐(5.265) ** ‐(5.080) 216 32.52**(13) 117 Cross‐Sectional Time‐Series Analysis of the Effect of Conflicts Mediated on Calls‐For‐Service CFS Shooting Coef SE Coef Conflicts Mediated 0.061 (0.268) ‐1.729 Time ‐0.091 (0.015) 1.077 ‐0.838 0.504 0.199 ‐0.302 0.005 0.056 ‐1.101 ‐0.004 0.095 1.033 ** CFS All CFS Assault SE Coef (0.771) ‐2.947 ** (0.874) ‐0.133 (0.088) ‐0.307 * (0.136) (0.851) (0.847) (0.841) (0.824) (0.744) ‐2.866 ‐4.816 6.596 4.191 7.628 (3.896) (3.830) (3.705) (3.438) (2.819) ‐2.596 ‐4.252 7.282 5.318 7.930 * (5.198) (5.064) (4.819) (4.372) (3.472) (0.801) (0.878) (0.893) (0.898) (0.898) (0.894) 1.683 1.167 2.438 2.177 ‐2.404 ‐5.518 (3.078) (3.700) (3.952) (4.060) (4.107) (4.079) 1.659 2.098 2.052 3.900 ‐2.281 ‐5.086 (0.702) 49.041 (3.533) 55.487 * SE Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.502 ** 216 52.46**(13) ** ** 216 36.58**(13) (3.807) (4.725) (5.157) (5.371) (5.458) (5.422) ** (5.067) 216 37.08**(13) 118 Cross‐Sectional Time‐Series Analysis of the Effect of Number of People in Conflicts on Calls‐For‐Service CFS Shooting Coef SE Coef Number of People in Conflicts 0.019 (0.031) ‐0.157 Time ‐0.091 (0.015) 1.075 ‐0.816 0.508 0.137 ‐0.282 0.032 0.074 ‐1.102 ‐0.012 0.121 1.041 ** CFS All CFS Assault SE Coef (0.078) ‐0.204 * (0.092) ‐0.142 (0.089) ‐0.330 * (0.135) (0.845) (0.842) (0.837) (0.829) (0.738) ‐3.061 ‐5.079 6.672 5.000 7.738 (3.873) (3.805) (3.676) (3.425) (2.779) ‐3.172 ‐4.712 7.301 6.155 8.109 * (5.132) (4.999) (4.756) (4.333) (3.417) (0.795) (0.873) (0.889) (0.892) (0.893) (0.890) 2.123 1.554 2.741 1.885 ‐1.954 ‐5.264 (3.030) (3.658) (3.918) (4.033) (4.074) (4.049) 2.378 2.667 2.407 3.446 ‐1.670 ‐4.798 (0.693) 48.261 (3.531) 53.965 * SE Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.476 ** 216 52.78**(13) ** ** 216 37.23**(13) (3.744) (4.656) (5.088) (5.304) (5.388) (5.356) ** (5.017) 216 31.69**(13) 119 Cross‐Sectional Time‐Series Analysis of the Effect of Number of Clients on Calls‐For‐Service CFS Shooting Coef CFS All CFS Assault SE Coef SE Coef SE (0.178) Number of Clients 0.105 ** (0.031) ‐0.277 (0.152) ‐0.191 Time ‐0.109 ** (0.015) ‐0.120 (0.087) ‐0.312 0.915 ‐0.970 0.327 0.159 ‐0.261 (0.793) (0.791) (0.786) (0.769) (0.693) ‐2.752 ‐4.542 7.379 4.450 7.902 (3.797) (3.736) (3.610) (3.346) (2.738) ‐3.055 ‐4.389 7.658 5.312 8.211 0.127 0.080 ‐1.275 ‐0.225 0.024 0.841 (0.746) (0.818) (0.833) (0.836) (0.836) (0.835) 2.331 1.787 3.099 2.193 ‐1.398 ‐4.493 (2.985) (3.595) (3.845) (3.954) (3.991) (3.971) 2.607 2.899 2.674 3.582 ‐1.283 ‐4.303 (0.672) 49.125 (3.466) 54.661 * (0.136) * (5.069) (4.946) (4.706) (4.264) (3.385) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 1.919 ** 216 71.41**(13) * ** ** 216 37.50**(13) (3.707) (4.604) (5.027) (5.239) (5.319) (5.294) ** (5.027) 216 28.69**(13) 120 Cross‐Sectional Time‐Series Analysis of the Effect of Number of Contacts on Calls‐For‐Service CFS Shooting Coef CFS All CFS Assault SE Coef SE Coef SE (0.016) Number of Contacts 0.008 ** (0.003) ‐0.021 (0.014) ‐0.021 Time ‐0.103 ** (0.015) ‐0.129 (0.088) ‐0.313 0.795 ‐1.016 0.309 0.013 ‐0.322 (0.816) (0.812) (0.807) (0.791) (0.712) ‐2.523 ‐4.527 7.310 4.722 7.973 (3.851) (3.780) (3.653) (3.400) (2.770) ‐2.667 ‐4.199 7.815 5.652 8.283 0.025 ‐0.008 ‐1.315 ‐0.345 ‐0.134 0.810 (0.766) (0.840) (0.855) (0.861) (0.860) (0.857) 2.339 1.769 3.125 2.424 ‐1.237 ‐4.641 (3.019) (3.635) (3.888) (4.009) (4.041) (4.013) 2.626 2.967 2.852 3.976 ‐0.963 ‐4.187 (0.673) 48.661 (3.482) 54.363 * (0.135) * (5.124) (4.985) (4.742) (4.308) (3.406) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.208 ** 216 64.80**(13) * ** ** 216 35.28**(13) (3.730) (4.634) (5.064) (5.293) (5.368) (5.332) ** (4.993) 216 28.79**(13) 121 Cross‐Sectional Time‐Series Analysis of the Effect of Number of Home Contacts on Calls‐For‐Service CFS Shooting Coef CFS All CFS Assault SE Coef SE Coef SE (0.062) Number of Home Contacts 0.031 ** (0.012) ‐0.071 (0.053) ‐0.064 Time ‐0.102 ** (0.015) ‐0.131 (0.088) ‐0.320 0.817 ‐0.992 0.289 0.035 ‐0.303 (0.817) (0.813) (0.808) (0.791) (0.711) ‐2.545 ‐4.547 7.415 4.734 7.967 (3.834) (3.763) (3.639) (3.383) (2.753) ‐2.793 ‐4.298 7.799 5.581 8.261 0.033 ‐0.026 ‐1.360 ‐0.335 ‐0.126 0.781 (0.766) (0.841) (0.857) (0.863) (0.862) (0.860) 2.369 1.858 3.257 2.413 ‐1.178 ‐4.496 (3.000) (3.617) (3.872) (3.993) (4.025) (4.002) 2.660 3.030 2.913 3.892 ‐0.991 ‐4.147 (0.673) 48.525 (3.469) 54.102 * (0.135) * (5.110) (4.972) (4.733) (4.295) (3.395) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.230 ** 216 63.75**(13) * ** ** 216 35.53**(13) (3.717) (4.623) (5.056) (5.282) (5.358) (5.327) ** (4.975) 216 24.42**(13) 122 Cross‐Sectional Time‐Series Analysis of the Effect of Number of Street Contatcs on Calls‐For‐Service CFS Shooting Coef CFS All CFS Assault SE Coef SE Coef SE (0.111) Number of Street Contacts 0.057 * (0.026) ‐0.158 (0.096) ‐0.200 Time ‐0.100 ** (0.015) ‐0.131 (0.088) ‐0.313 0.846 ‐1.003 0.374 0.002 ‐0.378 (0.827) (0.823) (0.817) (0.804) (0.725) ‐2.616 ‐4.517 7.107 4.700 8.120 (3.882) (3.813) (3.685) (3.435) (2.807) ‐2.610 ‐4.118 7.765 5.765 8.456 ‐0.002 ‐0.010 ‐1.262 ‐0.396 ‐0.082 0.853 (0.779) (0.852) (0.867) (0.879) (0.870) (0.868) 2.240 1.645 3.036 2.746 ‐1.440 ‐4.750 (3.055) (3.674) (3.926) (4.064) (4.074) (4.049) 2.522 2.878 2.871 4.483 ‐1.052 ‐4.209 (0.678) 48.633 (3.505) 54.384 * (0.135) * (5.143) (5.007) (4.763) (4.333) (3.432) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.330 ** 216 61.30**(13) ** ** 216 34.78**(13) (3.754) (4.661) (5.091) (5.337) (5.388) (5.355) ** (4.994) 216 29.90**(13) 123 Cross‐Sectional Time‐Series Analysis of the Effect of Hours Spent with Client on Calls‐For‐Service CFS Shooting Coef CFS All CFS Assault SE Coef SE Coef SE (0.003) Hours Spent with Clients 0.002 * (0.001) ‐0.004 (0.003) ‐0.005 Time ‐0.102 ** (0.015) ‐0.130 (0.088) ‐0.313 0.786 ‐1.009 0.398 0.087 ‐0.307 (0.822) (0.816) (0.809) (0.794) (0.716) ‐2.487 ‐4.523 7.048 4.468 7.874 (3.854) (3.781) (3.646) (3.388) (2.770) ‐2.559 ‐4.187 7.590 5.429 8.192 0.032 0.013 ‐1.278 ‐0.345 ‐0.174 0.803 (0.771) (0.844) (0.859) (0.867) (0.866) (0.862) 2.236 1.654 2.929 2.376 ‐1.198 ‐4.676 (3.019) (3.635) (3.886) (4.008) (4.044) (4.013) 2.541 2.877 2.749 4.010 ‐0.805 ‐4.136 (0.676) 48.688 (3.482) 54.377 * (0.135) * (5.125) (4.982) (4.728) (4.292) (3.403) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.245 ** 216 63.49**(13) ** ** 216 35.32**(13) (3.727) (4.629) (5.058) (5.290) (5.372) (5.330) ** (4.984) 216 29.14**(13) 124 Cross‐Sectional Time‐Series Analysis of the Effect of Unemployed Clients on Calls‐For‐Service CFS Shooting Coef CFS All CFS Assault SE Coef SE Coef SE (0.706) Unemployed Clients 0.488 ** (0.122) ‐0.813 (0.605) ‐0.398 Time ‐0.113 ** (0.014) ‐0.129 (0.088) ‐0.329 0.934 ‐0.859 0.322 0.117 ‐0.240 (0.771) (0.769) (0.765) (0.749) (0.677) 3.009 ‐4.931 7.130 4.385 7.856 (3.829) (3.764) (3.640) (3.377) (2.763) ‐3.323 ‐4.730 7.459 5.238 8.184 0.205 0.264 ‐1.090 ‐0.145 0.126 0.926 (0.728) (0.798) (0.812) (0.814) (0.814) (0.812) 2.341 1.493 2.722 1.887 ‐1.693 ‐4.891 (3.012) (3.630) (3.884) (3.989) (4.028) (4.003) 2.596 2.683 2.380 3.272 ‐1.574 ‐4.668 (0.661) 48.869 (3.504) 53.932 * (0.137) * (5.072) (4.942) (4.708) (4.268) (3.385) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 1.749 ** 216 79.66**(13) * ** ** 216 35.30**(13) (3.708) (4.608) (5.033) (5.240) (5.323) (5.293) ** (5.064) 216 27.85**(13) 125 Cross‐Sectional Time‐Series Analysis of the Effect of Problem Solving Discussion on Calls‐For‐Service CFS Shooting Coef CFS All CFS Assault SE Coef SE Coef SE (0.029) Problem Solving Discussion 0.015 * (0.006) ‐0.038 (0.025) ‐0.042 Time ‐0.102 ** (0.015) ‐0.131 (0.088) ‐0.315 0.795 ‐1.030 0.309 ‐0.045 ‐0.356 (0.821) (0.817) (0.811) (0.798) (0.717) ‐2.548 ‐4.534 7.261 4.846 8.049 (3.865) (3.795) (3.666) (3.421) (2.782) ‐2.625 ‐4.150 7.851 5.832 8.369 0.012 ‐0.004 ‐1.320 ‐0.390 ‐0.155 0.804 (0.771) (0.844) (0.860) (0.869) (0.865) (0.862) 2.290 1.689 3.110 2.530 ‐1.242 ‐4.670 (3.031) (3.649) (3.903) (4.032) (4.058) (4.029) 2.579 2.903 2.890 4.152 ‐0.899 ‐4.158 (0.673) 48.583 (3.495) 54.219 ** (0.135) * (5.132) (4.995) (4.751) (4.326) (3.414) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.288 ** 216 63.61**(13) * ** ** 216 35.18**(13) (3.736) (4.642) (5.075) (5.311) (5.380) (5.342) ** (4.988) 216 29.13**(13) 126 Cross‐Sectional Time‐Series Analysis of the Effect of Client Conflict Mediation on Calls‐For‐Service CFS Shooting CFS All CFS Assault Coef SE Coef SE Coef SE Client Conflict Mediation 0.050 (0.175) ‐0.644 (0.487) ‐0.766 (0.568) Time ‐0.090 (0.015) ‐0.151 (0.087) ‐0.340 1.125 ‐0.723 0.516 0.277 ‐0.250 (0.842) (0.839) (0.833) (0.813) (0.729) ‐3.530 ‐5.084 6.626 3.918 7.685 (3.843) (3.782) (3.660) (3.399) (2.790) ‐3.778 ‐5.062 7.039 4.802 7.961 0.100 0.119 ‐1.095 0.005 0.220 1.080 (0.785) (0.866) (0.884) (0.896) (0.891) (0.885) 2.032 1.485 2.689 2.354 ‐2.189 ‐5.171 (3.041) (3.651) (3.899) (4.028) (4.049) (4.018) 2.324 2.673 2.384 3.887 ‐2.046 ‐4.898 (0.696) 47.681 (3.473) 53.405 ** ** (0.131) * (5.068) (4.946) (4.715) (4.284) (3.411) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.446 ** 216 51.71**(13) ** ** 216 34.58**(13) (3.734) (4.624) (5.039) (5.264) (5.326) (5.290) ** (4.909) 216 29.41**(13) 127 Cross‐Sectional Time‐Series Analysis of the Effect of Anger Management Referrals on Calls‐For‐Service CFS Shooting CFS All CFS Assault Coef SE Coef SE Coef SE Anger Management Referral ‐4.986 (2.849) 4.156 (7.759) 2.123 (9.341) Time ‐0.089 (0.015) ‐0.156 (0.089) ‐0.353 1.135 ‐0.610 0.516 0.364 ‐0.193 (0.819) (0.816) (0.809) (0.787) (0.699) ‐3.491 ‐5.260 6.704 3.974 7.698 (3.835) (3.767) (3.635) (3.362) (2.743) ‐3.600 ‐4.972 7.241 5.056 8.115 0.186 0.548 ‐1.195 0.033 0.319 1.124 (0.754) (0.862) (0.860) (0.864) (0.865) (0.862) 2.232 1.185 2.549 1.486 ‐2.118 ‐5.415 (2.992) (3.672) (3.877) (3.992) (4.033) (4.009) 2.575 2.511 2.299 3.029 ‐1.785 ‐4.953 (0.676) 47.402 (3.499) 52.471 ** ** (0.132) * (5.040) (4.911) (4.675) (4.238) (3.364) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.463 ** 216 55.67**(13) ** ** 216 34.38**(13) (3.685) (4.630) (5.001) (5.210) (5.293) (5.261) ** (4.905) 216 28.18**(13) 128 Cross‐Sectional Time‐Series Analysis of the Effect of Employment Discussion on Calls‐For‐Service CFS All CFS Assault CFS Shooting Coef SE Coef SE Coef SE Employment Discussion 0.038 (0.020) ‐0.080 (0.072) ‐0.075 (0.085) Time ‐0.097 (0.015) ‐0.140 (0.088) ‐0.331 1.014 ‐0.965 0.299 ‐0.022 ‐0.361 (0.828) (0.830) (0.826) (0.813) (0.724) ‐3.088 ‐4.706 7.341 4.821 8.070 (3.824) (3.768) (3.643) (3.404) (2.748) ‐3.276 ‐4.454 7.783 5.728 8.380 0.016 0.027 ‐1.200 ‐0.107 0.052 1.010 (0.778) (0.855) (0.871) (0.875) (0.874) (0.872) 2.415 1.810 2.998 1.997 ‐1.564 ‐5.061 (2.991) (3.613) (3.869) (3.986) (4.023) (3.997) 2.668 2.898 2.576 3.405 ‐1.429 ‐4.711 (0.681) 48.117 (3.486) 53.608 ** * (0.134) * (5.072) (4.958) (4.726) (4.310) (3.386) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.349 ** 216 57.99**(13) * ** ** 216 35.34**(13) (3.701) (4.603) (5.029) (5.244) (5.325) (5.291) ** (4.961) 216 28.37**(13) 129 Cross‐Sectional Time‐Series Analysis of the Effect of Job Readiness Assistance on Calls‐For‐Service Coef CFS All CFS Assault CFS Shooting SE Coef SE Coef SE (0.810) Job Readiness Assistance 0.444 * (0.195) ‐0.164 (0.694) 0.185 Time ‐0.097 ** (0.015) ‐0.150 (0.087) ‐0.352 1.001 ‐0.762 0.473 0.251 ‐0.235 (0.827) (0.824) (0.819) (0.801) (0.721) ‐3.240 ‐5.045 6.899 4.120 7.794 (3.795) (3.729) (3.603) (3.338) (2.731) ‐3.521 ‐4.886 7.303 5.126 8.181 0.120 0.110 ‐1.205 ‐0.319 ‐0.235 0.934 (0.776) (0.853) (0.870) (0.880) (0.887) (0.872) 2.256 1.552 2.746 1.810 ‐1.661 ‐5.096 (2.978) (3.591) (3.842) (3.967) (4.035) (3.968) 2.634 2.745 2.378 3.021 ‐1.829 ‐4.883 (0.681) 47.770 (3.451) 52.871 ** (0.133) * (5.049) (4.917) (4.680) (4.242) (3.368) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.333 ** 216 59.45**(13) ** ** 216 34.45**(13) (3.688) (4.583) (5.008) (5.247) (5.359) (5.273) ** (4.930) 216 27.98**(13) 130 Cross‐Sectional Time‐Series Analysis of the Effect of Job Preparation Assistance on Calls‐For‐Service CFS All CFS Assault CFS Shooting Coef SE Coef SE Coef SE Job Preparation Assistance 0.102 (0.052) ‐0.227 (0.180) ‐0.175 (0.214) Time ‐0.099 (0.015) ‐0.138 (0.087) ‐0.331 1.126 ‐0.677 0.547 0.299 ‐0.154 (0.821) (0.820) (0.814) (0.797) (0.721) ‐3.268 ‐5.146 6.852 4.044 7.570 (3.806) (3.742) (3.622) (3.363) (2.768) ‐3.525 ‐4.971 7.230 4.990 7.957 0.144 0.101 ‐1.164 ‐0.199 0.136 1.135 (0.774) (0.848) (0.864) (0.870) (0.867) (0.867) 2.219 1.556 2.955 2.323 ‐1.640 ‐5.219 (3.006) (3.613) (3.859) (3.979) (4.003) (3.984) 2.569 2.775 2.578 3.644 ‐1.560 ‐4.911 (0.694) 48.380 (3.467) 53.907 ** * (0.133) * (5.057) (4.935) (4.698) (4.265) (3.398) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.176 ** 216 59.46**(13) ** ** 216 35.33**(13) (3.711) (4.604) (5.025) (5.247) (5.311) (5.282) ** (4.982) 216 28.36**(13) 131 Cross‐Sectional Time‐Series Analysis of the Effect of Substance Abuse Discussion on Calls‐For‐Service CFS All CFS Assault CFS Shooting Coef SE Coef SE Coef SE Substance Abuse Discussion 0.031 (0.024) ‐0.106 (0.081) ‐0.136 (0.946) Time ‐0.094 (0.015) ‐0.141 (0.089) ‐0.325 1.006 ‐0.890 0.402 0.043 ‐0.325 (0.837) (0.835) (0.831) (0.822) (0.730) ‐2.936 ‐4.826 7.193 4.925 8.017 (3.849) (3.782) (3.652) (3.420) (2.757) ‐3.010 ‐4.373 7.836 6.095 8.427 0.020 0.045 ‐1.151 ‐0.013 0.111 1.030 (0.785) (0.863) (0.878) (0.881) (0.882) (0.880) 2.400 1.811 2.982 1.779 ‐1.799 ‐5.191 (3.004) (3.630) (3.889) (4.003) (4.044) (4.020) 2.684 2.938 2.599 3.255 ‐1.604 ‐4.761 (0.687) 48.119 (3.509) 53.828 ** * (0.135) * (5.115) (4.986) (4.746) (4.337) (3.403) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.409 ** 216 55 * ** ** 216 35.58**(13) (3.724) (4.634) (5.065) (5.279) (5.363) (5.331) ** (5.005) 216 29.23**(13) 132 Cross‐Sectional Time‐Series Analysis of the Effect of Substance Abuse Treatment Referrals on Calls‐For‐Service CFS All CFS Assault CFS Shooting Coef SE Coef SE Coef SE Substance Abuse Treatment Referral 0.026 (0.891) 0.310 (2.463) ‐1.313 (2.902) Time ‐0.090 (0.015) ‐0.157 (0.088) ‐0.343 1.102 ‐0.776 0.502 0.251 ‐0.282 (0.853) (0.851) (0.845) (0.830) (0.747) ‐3.408 ‐5.092 6.846 4.121 7.830 (3.811) (3.742) (3.613) (3.357) (2.751) ‐3.423 ‐4.976 7.208 4.947 7.986 0.028 0.088 ‐1.106 0.050 0.137 1.063 (0.800) (0.882) (0.899) (0.903) (0.904) (0.903) 2.269 1.564 2.710 1.627 ‐1.911 ‐5.221 (2.980) (3.599) (3.856) (3.970) (4.014) (3.998) 2.583 2.636 2.252 2.974 ‐1.873 ‐5.063 (0.710) 47.292 (3.482) 53.094 ** ** (0.134) * (5.072) (4.937) (4.698) (4.267) (3.392) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.503 ** 216 51.19**(13) ** ** 216 34.56**(13) (3.699) (4.601) (5.030) (5.244) (5.329) (5.301) ** (4.980) 216 27.92**(13) 133 Cross‐Sectional Time‐Series Analysis of the Effect of Educational Issues Discussion on Calls‐For‐Service Coef CFS All CFS Assault CFS Shooting SE Coef * SE Coef SE (0.057) ‐0.141 * (0.067) * (0.135) * (5.071) (4.961) (4.702) (4.275) (3.381) Educational Issues Discussion 0.033 * (0.016) ‐0.133 Time ‐0.099 ** (0.015) ‐0.127 (0.088) ‐0.302 1.023 ‐1.026 0.435 0.045 ‐0.347 (0.825) (0.828) (0.819) (0.805) (0.724) ‐3.025 ‐4.291 7.014 4.885 8.074 (3.830) (3.776) (3.633) (3.375) (2.748) ‐3.082 ‐3.908 7.677 5.928 8.432 0.053 0.064 ‐1.125 ‐0.043 0.088 1.032 (0.779) (0.853) (0.868) (0.871) (0.871) (8.692) 2.173 1.805 2.815 1.839 ‐1.749 ‐5.236 (2.996) (3.616) (3.874) (3.986) (4.027) (4.005) 2.398 2.836 2.357 3.217 ‐1.507 ‐4.789 (0.687) 48.944 (3.498) 54.940 Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.250 ** 216 59.99**(13) ** ** 216 39.20**(13) (3.701) (4.601) (5.028) (5.240) (5.323) (5.292) ** (4.989) 216 31.89**(13) 134 Cross‐Sectional Time‐Series Analysis of the Effect of Educational Program Referrals on Calls‐For‐Service CFS Shooting Coef CFS All CFS Assault SE Coef SE Coef SE (0.685) Educational Program Referral 0.495 ** (0.191) ‐0.382 (0.583) ‐0.087 Time ‐0.098 ** (0.014) ‐0.149 (0.087) ‐0.348 1.048 ‐0.972 0.404 0.143 ‐0.395 (0.822) (0.822) (0.816) (0.799) (0.723) ‐3.327 ‐4.835 6.904 4.133 7.923 (3.824) (3.763) (3.635) (3.371) (2.764) ‐3.495 ‐4.855 7.322 5.126 8.180 (5.039) (4.916) (4.677) (4.239) (3.369) 0.032 0.062 ‐1.138 ‐0.518 ‐0.083 0.815 (0.777) (0.850) (0.865) (0.884) (0.870) (0.869) 2.230 1.500 2.701 2.126 ‐1.713 ‐4.964 (3.009) (3.624) (3.877) (4.018) (4.026) (4.004) 2.614 2.720 2.384 3.234 ‐1.622 ‐4.794 (3.685) (4.579) (5.001) (5.252) (5.300) (5.269) (0.674) 47.757 (3.464) 52.897 ** (0.132) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.367 ** 216 62.79**(13) ** ** 216 33.80**(13) ** (4.911) 216 28.06**(13) 135 Cross‐Sectional Time‐Series Analysis of the Effect of Gang Involved Clients on Calls‐For‐Service CFS Shooting Coef SE Coef Gang Involved Clients 0.135 (0.383) ‐3.370 Time ‐0.091 (0.015) 1.092 ‐0.841 0.526 0.202 ‐0.292 0.030 0.075 ‐1.010 0.002 0.118 1.036 ** CFS All CFS Assault SE Coef (0.945) ‐4.514 ** (1.096) ‐0.130 (0.088) ‐0.311 * (0.134) (0.847) (0.846) (0.842) (0.822) (0.740) ‐3.221 ‐4.512 5.886 4.242 7.470 (3.863) (3.799) (3.679) (3.400) (2.784) ‐3.129 ‐4.056 6.514 5.393 7.805 * (5.092) (4.966) (4.728) (4.282) (3.399) (0.798) (0.876) (0.891) (0.894) (0.896) (0.892) 1.197 0.994 2.266 1.829 ‐2.573 ‐5.372 (3.040) (3.661) (3.913) (4.022) (4.070) (4.039) 1.255 1.986 1.909 3.323 ‐2.075 ‐4.751 (0.700) 49.522 (3.515) 55.741 ** SE Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.480 ** 216 52.43**(13) ** ** 216 45.10**(13) (3.729) (4.630) (5.053) (5.263) (5.351) (5.315) ** (4.973) 216 44.02**(13) 136 Cross‐Sectional Time‐Series Analysis of the Effect of Core Gang Member Clients on Calls‐For‐Service CFS Shooting Coef CFS All CFS Assault SE Coef SE Coef SE (0.308) Core Gang Member Clients 0.183 ** (0.054) ‐0.396 (0.265) ‐0.255 Time ‐0.109 ** (0.015) ‐0.125 (0.088) ‐0.320 0.911 ‐0.957 0.275 0.119 ‐0.276 (0.791) (0.789) (0.785) (0.767) (0.692) ‐2.825 ‐4.634 7.407 4.504 7.944 (3.809) (3.747) (3.627) (3.360) (2.749) ‐3.146 ‐4.491 7.654 5.338 8.235 0.097 0.084 ‐1.277 ‐0.203 0.036 0.855 (0.744) (0.816) (0.831) (0.834) (0.834) (0.833) 2.431 1.753 3.096 2.111 ‐1.452 ‐4.609 (2.996) (3.608) (3.858) (3.967) (4.004) (3.984) 2.665 2.856 2.631 3.482 ‐1.377 ‐4.432 (0.670) 48.896 (3.474) 54.305 * (0.136) * (5.073) (4.949) (4.715) (4.271) (3.388) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 1.928 ** 216 71.73**(13) * ** ** 216 36.13**(13) (3.710) (4.607) (5.031) (5.241) (5.323) (5.297) ** (5.029) 216 28.20**(13) 137 Cross‐Sectional Time‐Series Analysis of the Effect of Clients with Prior Criminal History on Calls‐For‐Service CFS Shooting Coef CFS All CFS Assault SE Coef SE Coef SE (0.217) Clients with Prior Criminal History 0.129 ** (0.038) ‐0.348 (0.185) ‐0.237 Time ‐0.110 ** (0.015) ‐0.119 (0.087) ‐0.310 0.945 ‐0.947 0.350 0.173 ‐0.264 (0.791) (0.790) (0.785) (0.768) (0.693) ‐2.820 ‐4.601 7.330 4.405 7.911 (3.794) (3.733) (3.608) (3.345) (2.738) ‐3.116 ‐4.441 7.616 5.281 8.214 0.144 0.086 ‐1.252 ‐0.220 0.038 0.878 (0.746) (0.817) (0.832) (0.835) (0.835) (0.834) 2.301 1.790 3.064 2.204 ‐1.407 ‐4.568 (2.986) (3.594) (3.843) (3.953) (3.990) (3.968) 2.586 2.896 2.642 3.575 ‐1.308 ‐4.370 (0.674) 49.219 (3.468) 54.759 * (0.136) * (5.064) (4.940) (4.702) (4.261) (3.384) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 1.881 ** 216 72.02**(13) * ** ** 216 37.77**(13) (3.706) (4.602) (5.024) (5.236) (5.315) (5.288) ** (5.037) 216 28.76**(13) 138 Cross‐Sectional Time‐Series Analysis of the Effect of High Risk Clients on Calls‐For‐Service CFS Shooting Coef CFS All CFS Assault SE Coef SE Coef SE (0.211) High Risk Clients 0.126 ** (0.037) ‐0.325 (0.181) ‐0.227 Time ‐0.109 ** (0.015) ‐0.121 (0.087) ‐0.312 0.888 ‐0.998 0.297 0.127 ‐0.280 (0.792) (0.791) (0.785) (0.768) (0.692) ‐2.734 ‐4.512 7.409 4.506 7.945 (3.803) (3.743) (3.617) (3.352) (2.742) ‐3.028 ‐4.352 7.699 5.357 8.241 0.130 0.094 ‐1.280 ‐0.254 ‐0.002 0.824 (0.745) (0.817) (0.832) (0.835) (0.835) (0.834) 2.302 1.719 3.079 2.224 ‐1.381 ‐4.510 (2.990) (3.601) (3.851) (3.962) (3.998) (3.978) 2.582 2.853 2.669 3.607 ‐1.263 ‐4.299 (3.710) (4.607) (5.032) (5.245) (5.325) (5.300) (0.669) 49.076 (3.469) 54.598 (5.023) * (0.136) ** (5.076) (4.954) (4.713) (4.269) (3.388) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 1.937 ** 216 71.91**(13) * ** ** 216 37.27**(13) 216 28.63**(13) 139 Cross‐Sectional Time‐Series Analysis of the Effect of Recent Victims of Violence on Calls‐For‐Service CFS Shooting Coef CFS All CFS Assault SE Coef * SE Coef SE (0.450) ‐0.908 (0.524) Recent Victim of Violence 0.304 ** (0.096) ‐1.009 Time ‐0.108 ** (0.015) ‐0.114 (0.087) ‐0.293 0.924 ‐0.861 0.414 0.185 ‐0.227 (0.796) (0.793) (0.789) (0.772) (0.696) ‐2.595 ‐4.740 7.157 4.421 7.782 (3.829) (3.764) (3.641) (3.381) (2.773) ‐2.861 ‐4.413 7.584 5.320 8.130 0.097 0.118 ‐1.314 ‐0.209 0.067 0.857 (0.749) (0.822) (0.837) (0.840) (0.840) (0.839) 2.370 1.592 3.299 2.232 ‐1.466 ‐4.387 (3.020) (3.633) (3.882) (3.988) (4.025) (4.005) 2.661 2.826 2.957 3.766 ‐1.217 ‐4.082 (0.677) 49.509 (3.485) 55.528 * (0.136) * (5.107) (4.975) (4.738) (4.299) (3.416) Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 1.927 ** 216 69.54**(13) * ** ** 216 38.19**(13) (3.741) (4.642) (5.070) (5.278) (5.358) (5.333) ** (5.051) 216 29.99**(13) 140 Cross‐Sectional Time‐Series Analysis of the Effect of Clients Aged 16‐25 on Calls‐For‐Service CFS Shooting Coef SE Coef Clients Aged 16‐25 0.111 (0.297) ‐1.997 Time ‐0.091 (0.015) 1.058 ‐0.852 0.499 0.188 ‐0.305 0.008 0.057 ‐1.105 ‐0.010 0.096 1.030 ** CFS All CFS Assault SE Coef (0.899) ‐3.446 ** (1.017) ‐0.132 (0.089) ‐0.303 * (0.137) (0.851) (0.847) (0.840) (0.824) (0.743) ‐2.694 ‐4.724 6.684 4.340 7.732 (3.899) (3.831) (3.703) (3.436) (2.813) ‐2.285 ‐3.998 7.456 5.582 8.103 * (5.210) (5.074) (4.825) (4.378) (3.473) (0.800) (0.877) (0.892) (0.896) (0.896) (0.893) 1.785 1.275 2.558 2.030 ‐2.365 ‐5.496 (3.072) (3.696) (3.949) (4.058) (4.106) (4.078) 1.800 2.228 2.200 3.697 ‐2.207 ‐4.984 (0.700) 49.021 (3.538) 55.448 * SE Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.489 ** 216 52.75**(13) ** ** 216 36.67**(13) (3.808) (4.729) (5.162) (5.378) (5.466) (5.430) ** (5.081) 216 37.14**(13) 141 Cross‐Sectional Time‐Series Analysis of the Effect of Recently Incarecerated Clients on Calls‐For‐Service CFS Shooting Coef SE Coef Recently Incarcerated Clients ‐0.103 (1.189) ‐8.259 Time ‐0.090 (0.015) 1.105 ‐0.744 0.496 0.261 ‐2.687 0.066 0.108 ‐1.111 0.039 0.170 1.066 ** CFS All CFS Assault SE Coef (2.917) ‐10.892 ** (3.451) ‐0.142 (0.088) ‐0.329 * (0.131) (0.844) (0.843) (0.837) (0.815) (0.730) ‐3.790 ‐4.651 6.134 4.002 7.664 (3.783) (3.711) (3.584) (3.302) (2.690) ‐3.917 ‐4.332 6.669 5.144 8.077 * (4.958) (4.832) (4.595) (4.156) (3.294) (0.788) (0.869) (0.887) (0.891) (0.891) (0.889) 1.337 1.173 2.084 1.057 ‐2.258 ‐5.664 (2.943) (3.557) (3.815) (3.930) (3.972) (3.949) 1.433 2.115 1.557 2.217 ‐1.841 ‐5.295 (0.696) 48.704 (3.461) 54.655 ** SE Month 1 2 3 4 5 6 7 8 9 10 11 12 Intercept N Wald Chi‐Square (df) * p<.05; **p<.01 2.489 ** 216 51.52**(13) ** ** 216 43.09**(13) (3.618) (4.496) (4.914) (5.122) (5.205) (5.174) ** (4.855) 216 39.00**(13)