MENU TITLE: Survey of Youth Gang Programs Series: OJJDP Published: Draft 5/90 219 pages 450,211 bytes ------------------------------ Some figures, charts, forms, and tables are not included in this ASCII plain-text file. To view this document in its entirety, download the Adobe Acrobat graphic file available from this Web site or order a print copy from NCJRS at 800-851-3420. ------------------------------ Survey of Youth Gang Problems and Programs in 45 Cities and 6 Sites Irving Spergel, G. David Curry, Ron Chance, Ruth Ross with Amy Clark, Alba Alexander, Candace Kane, Phillis Garth, Dawn Isis, Michael Hyatt, Roberto Caldero, Paul Hamanda, Pamela Rodriguez, Janet Curry, and Sandra Oh National Youth Gang Suppression and Intervention Program School of Social Service Administration University of Chicago Distributed By: Juvenile Justice Clearinghouse Box 6000 Rockville, MD 20850 1-800-638-8736 ------------------------------ ABSTRACT This draft report presents the results of a national survey of 254 law enforcement, judicial system, education, and social welfare professionals who deal with the youth gang problem in 45 cities and 6 institutional sites (mainly correctional institutions). The cities and sites were located in four regions of the country: Northeast, Midwest, South, and the West. The goal of this survey was to achieve a better understanding of the nature of the organized response to the youth-gang problem by a range of criminal-justice and community-based agencies. The survey included a written questionnaire, telephone survey, and site visits. It was conducted by a team of researchers at the University of Chicago and sponsored by the Office of Juvenile Justice and Delinquency Development, U.S. Department of Justice. The report includes analysis of responses to questions concerning the definition of youth gang and related terms, the nature and causes of the gang phenomenon, trends in gang growth and activity, and the relative effectiveness of various suppression and intervention strategies. The report contains 6 charts and 90 tables of text. Additional analysis, with 10 charts and 52 tables, is provided in two appendixes. ------------------------------ TABLE OF CONTENTS EXECUTIVE SUMMARY CHAPTER 1 INTRODUCTION METHOD AND SCOPE SUMMARY CHAPTER 2 DEFINITIONS OF GANG, GANG MEMBER, AND GANG INCIDENT THE GANG DIFFERENCES IN GANG DEFINITION BY AREA AND RESPONDENT SUMMARY: GANG DEFINITION THE GANG MEMBER DIFFERENCES IN GANG MEMBER IDENTIFICATION SUMMARY: GANG MEMBER DEFINITION GANG-RELATED INCIDENTS DIFFERENCES IN DEFINING GANG INCIDENT BY AREA SUMMARY: GANG INCIDENT DEFINITION SUMMARY CHAPTER 3 GANG-CHARACTERISTICS SIZE OF THE PROBLEM GANG SIZE RACE, ETHNICITY, AND GANGS GANG ORGANIZATIONAL PARAMETERS LEVEL OF GANG CRIMINALITY ADULTS AND GANGS DRUGS AND GANGS GANG PATTERNS: INTERRELATIONSHIP FOR STRUCTURAL AND BEHAVIORAL CHARACTERISTICS CRIMINAL ORGANIZATION, DRUG TRAFFICKING, AND THE SPREAD OF THE YOUTH GANG PROBLEM SUMMARY CHAPTER 4 RESPONSE TO THE PROBLEM INDIVIDUAL ORGANIZATIONAL CHARACTERISTICS SIZE OF AGENCY SPECIFIC POLICIES TRAINING ORGANIZATIONAL SPECIALIZATION ADVISORY AND INTERAGENCY STRUCTURES PATTERNS OF AGENCY RESPONSE SPECIAL POLICIES ORGANIZATION INFLUENCE AND COORDINATION SPECIALIZED TRAINING ORGANIZATIONAL SPECIALIZATION RELATIONSHIPS OF AGENCY RESPONSE TO GANG PROBLEM POLICIES ADVISORY STRUCTURES AND TASK FORCE RESPONSES AVAILABILITY OF TRAINING ORGANIZATIONAL SPECIALIZATION SUMMARY CHAPTER 5 STRATEGIES AND CAUSES COMMUNITY ORGANIZATION SOCIAL INTERVENTION OPPORTUNITIES PROVISION SUPPRESSION ORGANIZATIONAL DEVELOPMENT AND CHANGE MDS STRATEGY DIFFERENCES PERCEIVED CAUSES OF THE GANG PROBLEM MDS CAUSAL PERCEPTIONS RELATIONSHIPS BETWEEN PERCEIVED CAUSE AND STRATEGY NETWORK CHARACTERISTICS OF AREAS AND RESPONDENTS SUMMARY CHAPTER 6 EFFECTIVENESS OF INTERVENTION CHANGES IN THE LEVEL OF THE PROBLEM PERCEIVED IMPROVEMENT AND THE NATURE OF THE PROBLEM POLICY AND ORGANIZATIONAL CHANGES UNIT EFFECTIVENESS INTERAGENCY EFFECTIVENESS SELECTING PROMISING APPROACHES DISCRIMINATING BETWEEN CHRONIC AND EMERGING CITIES FINDING EFFECTIVE STRATEGIES BUILDING CAUSAL REGRESSION MODELS SUMMARY REFERENCES APPENDIX A An MDS Approach to Gang Definition by Area An MDS Approach to Gang Definition by Respondent Category An MDS Approach to Membership Identification by Area An MDS Approach to Gang Member Definition by Respondent Category An MDS Approach to Gang Incident by Site An MDS Approach to Gang Incident by Respondent Category MDS Strategy Differences Across Areas and Respondent Categories Differences among Sites in Strategy Differences Among Respondent Categories in Strategy Comparison of Causal Perceptions Across Sites Comparison of Causal Perceptions Across Respondent Categories Network Measures APPENDIX B LIST OF FIGURES Figure 6-1 General Effectiveness Score and Community Organizational Strategy Figure 6-2 General Effectiveness Score and Social Intervention Strategy Figure 6-3 General Effectiveness Score and Opportunities Provision Figure 6-4 General Effectiveness Score and Suppression Strategy Figure 6-5 General Effectiveness Score and Organizational Development Strategy TEXT TABLES Table 1-3. Respondents in Analysis by Geographic Area (n = 45) and Selected Institutions (n = 6) Table 1-4. Respondent Categories Table 1-5. Type of Gang Problem by Area (City, County, Site) Table 2-1. Groupings of Dichotomous Variables Derived from Content Analysis of Definition of a Gang (Item III-1) Table 2-3. Rotated Factor Matrix from GLS Factor Analysis of Tetrachoric Correlation Coefficients of Gang Definition Characteristics Table 2-4. Average Degree of Consensus on Gang Definitional Criteria by Area Table 2-5. Average Degree of Consensus on Gang Definitional Criteria by Respondent Category Table 2-6. Average Degree of Consensus for 42 Areas and 13 Respondent Categories by Definitional Criteria Table 2-13. Methods of Identifying Gang Members Table 2-15. Rotated Factor Matrix from GLS Factor Analysis of Tetrachoric Correlation Coefficients of Gang Incident Definition Characteristics Table 2-16. Average Degree of Consensus on Membership Criteria by Area Table 2-17. Average Degree of Consensus on Member Identification Methods by Respondent Category Table 2-18. Average Degree of Consensus for 42 Areas and 13 Respondent Categories by Member Identification Criteria Table 2-25. Definitions of a Gang-Related Incident Table 2-27. Rotated Factor Matrix from GLS Factor Analysis of Tetrachoric Correlation Coefficients of Gang Incident Definition Characteristics Table 2-28. Average Degree of Consensus on Incident Criteria by Area Table 2-29. Average Degree of Consensus on Incident Identification Methods by Respondent Category Table 2-30. Average Degree of Consensus for 42 Area and 13 Respondent Categories by Definitional Criteria Table 3-1. Number of Gangs and Gang Members in 1987 Table 3-2. City and County Populations (1984) Table 3-3. Correlations: City/County Population Size, Numbers of Gangs, and Numbers of Gang Members Table 3-20. Pearson Correlations for Selected Gang Problem Characteristics Table 3-21. Pearson Correlations for Selected Gang Problem Characteristics Table 3-22. Pearson Correlations for Selected Gang Problem Characteristics Table 3-23. Point Biserial Correlations of Gangs of Same Name in Different Neighborhoods and Selected Measures of the Gang Problem Table 3-24. Point Biserial Correlations of Gangs Affiliated with Adult Criminal Organizations and Selected Measures of the Gang Problem Table 3-25. Point Biserial Correlations of Drugs a Primary Purpose of the Gang and Selected Measures of the Gang Problem Table 3-26. Point Biserial Correlations of Gang Members Involved in the Importation of Drugs and Selected Measures of the Gang Problem Table 3-27. Gangs with the Same Name are Found in Different Areas and Respondents Reporting Gang Affiliation with Adult Criminal Organizations Table 3-28. Gangs with the Same Name are Found in Different Areas and Drugs are a Primary Purpose of the Gang Table 3-29. Gangs with the Same Name are Found in Different Areas and Gang Members are Involved in the Importation of Drugs Table 3-30. Respondents Reporting Gang Affiliation with Adult Criminal Organizations and Drugs are a Primary Purpose of the Gang Table 3-31. Respondents Reporting Gang Affiliation with Adult Criminal Organizations and Gang Members are Involved in the Importation of Drugs Table 3-32. Drugs are a Primary Purpose of the Gang and Gang Members are Involved in the Importation of Drugs Table 4-10. Tetrachoric Correlation Coefficients between Measures of the Policy Aspects of Gang Program Responses Table 4-11. Organizations Who Have Special Policies and Procedures for Dealing with Gangs and Organizations whose Special Gang Policies and Procedures are in Writing Table 4-12. Tetrachoric Correlation Coefficients between Measures of Communication and Coordination Aspects of Gang Program Responses Table 4-13. Tetrachoric Correlation Coefficients between Measures of Policy and Measure of Communication and Coordination in Gang Program Responses Table 4-14. Tetrachoric Correlation Coefficients between the Availability of Special Gang Training for Personnel and Other Gang Program Responses Table 4-15. Tetrachoric Correlation Coefficients between Organizational Specialization and Other Gang Program Responses Table 4-16. Point Biserial Correlations of Having a Special Gang Policy and Selected Measures of the Gang Problem Table 4-17. Point Biserial Correlations of Having a Special Gang Policy in Writing and Selected Measures of the Gang Problem Table 4-18. Point Biserial Correlations of Influencing Gang Legislation and Selected Measures of the Gang Problem Table 4-19. Tetrachoric Correlation Coefficients between Measures of Policy Responses and Selected Measures of the Nature of the Gang Problem Table 4-20. Point Biserial Correlations of Having an Internal Advisory Structure and Selected Measures of the Gang Problem Table 4-21. Point Biserial Correlations of Having an External Advisory Structure and Selected Measures of the Gang Problem Table 4-22. Point Biserial Correlations of an Interagency Task Force and Selected Measures of the Gang Problem Table 4-23. Tetrachoric correlation Coefficients Between Measures of Communication and Coordination Responses and Selected Measures of the Nature of the Gang Problem Table 4-24. Point Biserial Correlations of Having Special Training for Agency Staff Available and Selected Measures of the Gang Problem Table 4-25. Tetrachoric Correlation Coefficients Between Organizational Specialization and Selected Measures of the Nature of the Gang Problem Table 5-1. Distribution of All Strategy Rankings Table 5-2. Primary Strategy by Respondent Type Table 5-3. Number of Responses by Respondent Category Table 5-4. Variety of Strategies Used by Respondents Table 5-10. Perceived Causes of the Gang Problem (Law Enforcement Only) Table 5-11. Perceived Causes of the Gang Problem (Not Including Law Enforcement) Table 5-16. Comparison of Strategy and Cause Responses: Number (%) Table 5-17. Tetrachoric Correlation Coefficients between Perceived Primary Cause and Primary Strategy Table 6-1. How Has the Gang Situation Changed Since 1980? Table 6-2. Gang Situation has Improved Since 1980 Table 6-3. Means for Selected Measures of the Gang Problem and Perceived Change in the Problem Since 1980 Table 6-4. Frequencies and Percents for Respondents Reporting Improvement in the Serious Gang Problem Since 1980 and Tetrachoric Correlations for Selected Measures of the Nature of the Gang Problem Table 6-5. Tetrachoric Correlation Coefficients between Improvement in the Gang Problem Since 1980 and Measures of the Policy Aspects of Gang Program Responses Table 6-6. Tetrachoric Correlation Coefficients between Improvement in the Gang Problem Since 1980 and Measures of the Coordination Aspects of Gang Program Responses Table 6-7. Evaluation of Own Agency's Effectiveness (1987) Table 6-8. Point Biserial Correlations of Agency Effectiveness Rating and Selected Measures of the Gang Problem Table 6-9. Polychoric Correlation Coefficients between Agency Effectiveness Rating and Selected Measures of the Nature of the Gang Problem Table 6-10. Polychoric Correlation Coefficients between Agency Effectiveness Rating and Selected Measures of Response to the Gang Problem Table 6-11. Evaluation of Inter-Agency Task Force or Community-Wide Organizations (1987) Table 6-12. Point Biserial Correlations of Community Level Program Effectiveness Rating and Selected Measures of the Gang Problem Table 6-13. Polychoric Correlation Coefficients between Community Level Program Effectiveness Rating and Selected Measures of the Nature of the Gang Problem Table 6-14. Polychoric Correlation Coefficients between Interagency Community Level Program Effectiveness Rating and Selected Measures of Agency Response to the Gang Problem Table 6-15. Ratings of Agency Effectiveness in 1987 and Perceived Improvement in the Gang Problem Since 1980 Table 6-16. Ratings of Community Level Program Effectiveness in 1987 and Perceived Improvement in the Gang Problem Since 1980 Table 6-17. Rating of Agency Effectiveness and Rating of Community Level Effectiveness in 1987 Table 6-18. Selected Combinations of Responses on Improvement Since 1980, Rating of Agency Effectiveness in 1987, and Rating of Community Level Program Effectiveness in 1987 Table 6-19. Ranking of Areas by Average General Effectiveness Scores Table 6-20. Ranking of Areas By Improvement in Gang Situation Since 1980 Table 6-21. Differences between Chronic versus Emerging Gang Problem Cities on Selected Variables Table 6-22. Discriminant Analysis Results on Chronic vs. Emerging Cities for Differences in the Ethnicity of Gang Members and 1984 Population Table 6-23. Discriminant Analysis Results on Chronic vs. Emerging Cities for Proportions of Respondents Exhibiting Three Primary Strategies Table 6-24. Analysis of Covariance Results for Score with Proportion of Respondents with Primary Strategy Community Organization as a Covariate for Chronic and Emerging Cities Table 6-25. Analysis of Covariance Results for Score with Proportion of Respondents with Primary Strategy Social Intervention as a Covariate for Chronic and Emerging Cities Table 6-26. Analysis of Covariance Results for Score with Proportion of Respondents with Primary Strategy Opportunities Provision as a Covariate for Chronic and Emerging Cities Table 6-27. Analysis of Covariance Results for Score with Proportion of Respondents with Primary Strategy Suppression as a Covariate for Chronic and Emerging Cities Table 6-28. Analysis of Covariance Results for Score with Proportion of Respondents with Primary Strategy Organizational Development as a Covariate for Chronic and Emerging Cities Table 6-30. Regression Model for General Effectiveness Measure for Chronic Cities (n=9) Table 6-31. Regression Model for General Effectiveness Measure for Chronic Cities, (n=18), with Addition of External Advisory Group Proportion Table 6-32. Regression Model for General Effectiveness Measure for Emerging Cities (n=9) APPENDIX TABLES AND FIGURES Appendix Table 1-1. Reasons for Non-Inclusion of Cities in this Study (N-55) Appendix Table 1-2. Questionnaire Appendix Table 2-2. Eigenvalues from Generalized Least Squares Factor Analysis of Gang Definition Variables Appendix Table 2-7. Correlations between MDS Dimensions for Site Dissimilarity and Gang Definitional Criteria Appendix Figure 2-1. Multi-dimensional Scaling Results for Differences in the Definition of Gangs by Geographic Site Appendix Table 2-9. Correlations between MDS Dimensions for Respondent Category Dissimilarity and Gang Definitional Criteria Appendix Figure 2-2. Multi-dimensional Scaling Results for Dissimilarity in the Definition of Gangs by Respondent Category Appendix Table 2-10. Eigenvalues from Generalized Least Squares Factor Analysis of Gang Member Definition Variables Appendix Table 2-15 Correlations between MDS Dimensions for Area Dissimilarity and Gang Definitional Criteria Appendix Figure 2-3. Multi-dimensional Scaling Results for Differences in the Identifying Gang Members by Geographic Site Appendix Table 2-16. Correlations between MDS Dimensions for Respondent Category Dissimilarity and Gang Member Identification Criteria Appendix Figure 2-4. Multi-dimensional Scaling Results for Dissimilarity in Gang Member Identification Methods by Respondent Category Appendix Table 2-18. Eigenvalues from Generalized Least Squares Factor Analysis of Gang Incident Definition Variables Appendix Table 2-23. Correlations between MDS Dimensions for Site Dissimilarity and Gang Definitional Criteria Appendix Figure 2-5. Multi-dimensional Scaling Results for Differences in the Identifying Gang Members by Geographic Site Appendix Table 2-24. Correlations between MDS Dimensions for Respondent Category Dissimilarity and Incident Definitional Criteria Appendix Figure 2-6. Multi-dimensional Scaling Results for Dissimilarity in Gang Incident Identification Methods by Respondent Category Appendix Table 3-4. Estimates of the Average Size of the Total Gang Appendix Table 3-5. Estimates of the Percent of Gang Members Who Are Black (1987) Appendix Table 3-6. Estimates of the Percent of Gang Members Who Are Hispanic (1987) Appendix Table 3-7. Estimates of the Percent of Hispanic Gang Members Who Are Mexican (1987) Appendix Table 3-8. Estimates of the Percent of Hispanic Gang Members Who Are Puerto Rican (1987) Appendix Table 3-9. Estimates of the Percent of Hispanic Gang Members Who Are Other Hispanic (1987) Appendix Table 3-10. Estimates of the Percent of Gang Members Who Are White (1987) Appendix Table 3-11. Estimates of the Percent of Gang Members Who Are Asian (1987) Appendix Table 3-12. Gangs with the Same Name are Found in Different Areas Appendix Table 3-13. Estimates of the Percent of Total Part 1 or Index Offenses Attributed to Youth Gangs in 1987 Appendix Table 3-14. Estimates of the Percent of Gang Offenders with Prior Police Records (1987) Appendix Table 3-15. Estimates of the Percent of Gang-Related Incidents or Cases Involving Adults in 1987 Appendix Table 3-16. The More Serious Part of the Gang Problem Appendix Table 3-17. Respondents Reporting Gang Affiliation with Adult Criminal Organizations Appendix Table 3-18. Drugs are a Primary Purpose of the Gang Appendix Table 3-19. Gang Members Involved in the Importation of Drugs Appendix Table 4-1. Total Number of Staff in Respondent's Agency Appendix Table 4-2. Organizations with Special Gang Policies and Procedures Appendix Table 4-3. Organizations whose Special Gang Policies and Procedures are in Writing Appendix Table 4-4. Organization that have Influenced Legislation Related to Gangs Appendix Table 4-5. Organizations that Provide Special Training on Gangs for Personnel Appendix Table 4-6. Agencies with Organizational Specialization to Deal with the Gang Problem Appendix Table 4-7. Organizations that have a Special Internal Advisory Structure Appendix Table 4-8. Organizations that have a Special External Advisory Structure Appendix Table 4-9. Respondents who Report Coordination Efforts by Inter-Agency Task Forces or Community-Wide Organizations in 1987 Appendix Table 5-5. Proportion of Respondents Exhibiting Each Strategy (Regardless of Rank) by Area Appendix Table 5-6. Proportion of Respondents in Each Site Exhibiting Strategy as Primary Strategy Appendix Table 5-7. Pearson Correlations of Proportion of Respondents Exhibiting Primary Strategy and MDS Dimension for 42 Areas Appendix Figure 5-1. MDS Plot of 42 Sites on Dimensions Produced from MDS of Euclidean Distances Generated From Proportion of Respondents Exhibiting Primary Strategy Appendix Table 5-8. Proportion of Respondents for Each Respondent Category Exhibiting Strategy as Primary Strategy Appendix Table 5-9. Pearson Correlations of Average Proportion of Respondents Exhibiting Primary Strategy and MDS Dimension for 13 Respondent Categories Appendix Figure 5-2. MDS Plot of 13 Respondent Categories on Dimensions Produced from MDS of Euclidean Distances Generated From Proportion of Respondents Exhibiting Primary Strategy Appendix Table 5-12. Primary Perceived Cause -- Proportion by Area Appendix Table 5-13. Pearson Correlations of Site Average Proportion of Respondents Ranking Cause Perception First and MDS Dimension for 42 Sites Appendix Figure 5-3. MDS Plot of 42 Sites on Dimensions Produced from MDS of Euclidean Distances Generated From Average Proportion of Respondents Ranking Cause Perception First Appendix Table 5-14. Most Important Cause -- Proportion of Respondents per Category Appendix Table 5-15. Pearson Correlations of Site Average Proportion of Respondents Ranking Cause Perception First and MDS Dimension for 13 Respondent Categories Appendix Table 5-18. Three Example Sociometric Networks (Rows are respondents indicating agencies with whom they are in most contact. Each column represents the corresponding row agency as being indicated as in contact with the row agency. The diagonal is arbitrarily filled with zeroes.) Appendix Table 5-19. Network Density Measures by Site Appendix Table 5-20. Means of Network Analysis Prominence Measures by Site Appendix Table 5-21. Means of Network Analysis Prominence Measures by Respondent Category Table B-1. Normalized (PROBIT) Scores for Three Gang Program Effectiveness Measures Table B-2. Eigenvalues Produced by Principal Components Analysis of Three Normalized Measures of Effectiveness Table B-3. Coefficients for First Principal Component ------------------------------ EXECUTIVE SUMMARY Introduction A questionnaire and telephone survey of 254 respondents in 45 cities and 6 institutional sites, mainly correctional schools, was conducted to assess the youth gang problem and especially to discover organized and "promising" programs for dealing with it. The cities and sites were located in four regions of the country: Northeast, Midwest, South, and the West. The largest number of cities from the same state in the study, 15, were from California. The next largest number was from Illinois with five cities. The study was concerned with analysis of the gang problem and the organized response to it in two types of cities: chronic gang problem cities, which often had a long history of gang problems, and emerging gang problem cities, often smaller cities that recognized and began to deal with the gang problem only since 1980. A wide range of respondents with direct operational and some policy experience in dealing with the gang problem was selected to participate in the survey. The 254 respondents were classified into 10 major categories and 6 subcategories of professional or agency representatives. Major categories included law enforcement (52), prosecution (26), judges (14), probation (32), corrections (8), parole (11), and schools (35). Subcategories included academic (26) and security staff (9) for schools; and youth service (46), grassroots (5), youth and family treatment (8), and comprehensive crisis intervention programs (3) for community organizations. The response rate for respondents in the survey was high. Seventy and one half percent (70.5 percent) of all questionnaires sent out were returned. Questionnaires were sent to cities that seemed to meet the selection criteria of the study, based on initial information: the presence and recognition by agencies, especially by the police, of a youth gang problem; the presence of a youth gang program; the existence of a program for at least one year, whether presently functioning or in the recent past; an articulated set of program goals; evidence of more than a simplistic unitary response; and some capacity to describe the impact of the program. ------------------------------ Definitions The importance of a common definition was stressed for understanding the scope and seriousness of the gang problem and for developing effective policies and programs of intervention. The survey respondents were asked to define three key components of the youth gang problem in their jurisdiction: the gang, gang member, and gang incident. A variety of research methods and statistical procedures were used to analyze the data collected, including content analysis, factor analysis, and multi-dimensional scaling. We found a diversity of meanings that serve as a constraint in interpreting the results of the research. The most frequently cited elements of a definition for gang were certain group or organizational characteristics, symbols, and a range of specific and general criminal activities, particularly violence, drug use, and drug sales. The most frequent elements used to define a gang member were symbols or symbolic behavior, self-admission, identification by others, especially the police, and association with gang members. The three most frequently cited ways of identifying a gang incident were the involvement of a gang member in the activity; gang function (i.e., gang-related interest and motivation); and modus operandi (i.e., behavior distinctively associated with gang activity, such as drive-by shootings). In general there was more consensus on definitions based on city or location than by discipline or agency affiliation of the respondent. In other words, the fact that a respondent was located in a particular city was more important than his professional or agency affiliation in determining the way he defined a gang, a gang member, or gang incident. To some extent these differences could be accounted for by whether the city was a chronic or emerging gang problem city. We found, for example, that respondents from chronic gang cities emphasized the general or specific criminal or antisocial nature of the gang, while those in the emerging problem cities focused on symbolic or organizational aspects in their definitions of a gang and gang incident. Respondents from chronic gang cities tended to use multiple factors in identifying gang members or gang incidents; the respondents in emerging cities usually emphasized one factor. Across areas, it was not always clear that criminal justice or non-criminal justice agencies were in greater agreement as to which component of the definition should be emphasized. Criminal justice system respondents generally tended to emphasize specific types of crimes while non- justice system respondents relied more on symbolic behaviors to identify youth as gang members. ------------------------------ Gang Characteristics and Behaviors Despite the existence of a great deal of variability in definition, there were clear patterns in assessment of the scope and severity of the problem. Gang problems and programs seemed to be concentrated in certain states and cities, mainly California and the Los Angeles area and Illinois, including Chicago. There were highly significant correlations between numbers of gangs and numbers of gang members, but not quite as strong with size of city. Some large cities did not report large numbers of gangs and gang members, and relatively smaller cities had disproportionately large and serious gang problems. Blacks (54.6 percent), mainly African-Americans, and Hispanics (32.6 percent), mainly Mexican- Americans, were the major racial/ethnic groups in the gang populations reported by law enforcement. Two-thirds of the respondents perceived gangs as groups affiliated across neighborhoods, cities, or states; they also stated that 75 percent of gang youth had prior police records; and that 22.7 percent of total index crimes in a jurisdiction were committed by gang youth. The gang problem was no longer viewed as only juvenile; adults were involved in 45.6 percent of the youth gang-related incidents. About three-quarters of the respondents stated that youth gangs or their individual members or cliques were involved with adult criminal organizations. A majority of respondents believed that a primary purpose of youth gangs is selling drugs mainly at the street level and that certain youth gang members are involved in transporting drugs across jurisdictions. In regard to more organized forms of gang criminality, there is a negative or inverse relationship between the percentage of Hispanic gang members and involvement with adult criminal organization. The relationship between drug selling as a primary purpose of the gang is characteristic of black but not Hispanic gangs. When drug distribution is a primary purpose of the youth gang, a higher percent of index crime in the community is attributed to gangs. Non-local (that is, gangs with the same name in different parts of one city, State, or county) are much more likely to be affiliated with adult criminal organizations, and such gangs are highly likely to be engaged in both street level and higher level drug trafficking, e.g., transporting drugs from place to place. In general, higher-level drug trafficking is not related to a variety of more general characteristics of gangs and gang behavior. The complexity of the youth gang's relationship to drug distribution, however, belies any easy classification of gangs into distinct categories of traditional and criminally organized youth gangs. It is likely that the availability of significant drug selling or trafficking opportunities has more to do with the development of a serious criminal youth gang problem than that the presence of youth gangs has a significant influence on the general drug problem. Response to the Problem A variety of criminal-justice and community-based organizations currently respond to the gang problem. Nevertheless, the law enforcement response is still the dominant one. It also differs somewhat from the response of other agencies. Police departments tend to be larger organizations. An explicit and formal approach tends to characterize law enforcement programs, including specially organized gang units, written policies, and special training. However, police department are less likely to have interdepartmental or external advisory gang structures, but do participate in community-wide coordinating or task-force efforts. We find among agencies generally that special policies and structures directed to the gang problem are reasonably well interrelated. For example, organizations with special units tend to have special policies and training. There appears to be a negative relationship, however, between the existence of a special gang unit or programs and the presence of external advisory group structures. This might be because an organization that has made a special commitment to dealing with the gang problem may not want some outside group closely examining or advising on what it is doing. In general, there appears to be a greater presence of special program policies, structures, and activities for dealing with the gang problem when the problem worsens. Agencies are especially active or reactive when there is evidence of non- local gangs or a spread of gangs across communities and evidence of affiliation of youth gangs with adult criminal organizations. The response of agencies, particularly through special policies, structures, and activities, appears to be largely reactive. In other words, the response is associated with an increasing problem. There is one noteworthy exception, however: a significant relationship exists between an external advisory structure and a reduction of the gang problem, i.e., lower numbers of gangs, gang members, smaller gang size, lower percent of gang incidents involving adults, and lower percent of gang members in the community with prior police records. This set of relationships indicates that an external advisory agency somehow contributes to a reduction of the gang problem. However, an external advisory structure is not associated with a change in the more serious or criminal aspects of the gang problem. Neither of the other coordinative structures (i.e., internal advisory group or interagency or community-wide task force) has this dramatic effect. ------------------------------ Strategies of Intervention Five strategies of intervention were identified as the basis for agency programs dealing with the gang problem. They included community organization (e.g., community mobilization and networking); social intervention, focusing on individual behavioral and value change; opportunities provision, with special focus on improved basic education, training, and job opportunities for youth; suppression, which emphasizes arrest, incarceration, and close monitoring and supervision of gang members; and organizational development and change or the creation of special organizational units and procedures. These strategies were usually employed in various combinations. However, a classification of primary strategies indicated that suppression (44.0 percent) was most frequently employed, followed by social intervention (31.5 percent), organizational development (10.9 percent), community organization (8.9 percent), and opportunities provision (4.8 percent). Prosecutors and judges were most highly committed to the use of a suppression strategy. Respondents from social agencies and grass-roots organizations were most committed to the use of social intervention strategies. Respondents from chronic gang problem cities emphasized a broad range of approaches, combining community organization and suppression with social intervention strategies. Respondents in emerging gang problem cities were divided in their approaches: some focused primarily on community organization and organizational development, while others focused on suppression. The respondents in the survey were also asked to indicate and rate what they considered to be causes of the gang problem. A content analysis of the many types of responses suggested four major classifications: system/structural, comprising poverty and unemployment, as well as opportunities to sell drugs; institutional, especially failures of basic institutions such as family and schools to do an adequate job with children and youth; individual level, including psychological problems, substance abuse, needs for status and protection by gang youth; and, organization response defects, such as the inability of police, courts, and social agency programs to deal effectively with the problem. Respondents in the various professional or agency categories selected highly similar patterns and rankings of causes. The most frequently selected category of causes was system/structural, with institutional failures second, agency response defects third, and individual youth problems, fourth. Analyses of the relation between strategies of intervention and causes of the youth gang problem suggested a certain degree of mismatch. While there was great emphasis on system/structural level causes, the related strategy of opportunities provision was insufficiently selected. While there was limited emphasis on individual level youth psychological or personal problems as a cause of gang problems, there seemed to be a disproportionately heavy emphasis on social intervention, especially in reference to various levels of counseling or crisis intervention. At the individual level of analysis respondents believed that system level problems of unemployment, poverty, and drug dealing would not be solved through a primary social intervention approach. It was not clear to which set of causes, a strategy of suppression was most closely related, except as a reaction to failures of existing institutions and programs. ------------------------------ Effectiveness of Approaches We attempted, based on cross-sectional survey data, to determine whether different strategies, policies, structures, and procedures lead to a perceived reduction in gang crime. First, we asked respondents whether the gang situation had improved or deteriorated in the past seven or eight years and whether agency and interagency efforts had been effective. A large majority of respondents believed that the gang situation had deteriorated, although law enforcement were less pessimistic than non-law enforcement respondents. Only 23.1 percent of the police and 10.4 percent of non-police respondents believed that there had been an improvement in the gang situation in their communities since 1980. In only 17 of our 45 cities was there evidence of any level of improvement in the gang situation. Those perceptions were associated with estimates of significantly fewer numbers of gangs, gang members, size of gangs, and a decline in the percent of total index crime attributed to youth gangs. Serious gang crime, including drug selling, was also reported as significantly lower. There was no evidence that improvement was more or less likely to occur in large or small, chronic or an emerging gang problem city. Moreover, no special policy, organizational or procedural development was associated with any of the characteristics of improved gang situation, with the exception of the presence of an external advisory group to a program. On the other hand, respondent ratings of how effective their agency or local interagency or task force efforts had been were far higher than ratings of an improved gang situation. More than 40 percent of all respondents saw their agencies as very effective in dealing with youth gangs. While the rating of agency or interagency effectiveness was positively related to a variety of special policies, structures, and procedures, it was generally not related to the various measures of gang or gang behavior, suggesting little direct relationship between special agency efforts and a change or lowering of the gang problem. There was evidence of a somewhat stronger effect by interagency task forces than individual agency efforts towards a reduction of the gang problem in the community, however. Since the three ratings (situation improved, agency effectiveness, and interagency effectiveness) were significantly intercorrelated, a general effectiveness score was constructed and used as a basis for ranking cities on whether the gang problems had been successfully addressed, at least based on respondent perceptions or reports. These rankings became a basis for the selection of cities and institutions for field visits and further inquiry as to which programs might be promising and could serve as models for other cities and institutions. At this point, our study shifted from focus on individual-level respondent data to respondent aggregated or averaged data on a city-wide basis. We were particularly interested in whether approaches to dealing with the problem might be more effective in one type of city than another. First, we had to make sure we had classified our cities reasonably well. Discriminant analysis enabled us to determine systematically how chronic and emerging gang problem cities differed. Chronic problem cities were larger and characterized by greater proportions of Hispanic gang members. Emerging gang problem cities were more likely to be smaller and have higher proportions of black gang members. Respondents in the smaller cities were also more closely interconnected in terms of networks of relationship. This classification pattern was correct for 86.5 percent of the cities. The discriminant analysis procedure also permitted us to classify with considerable accuracy primary strategies across agencies by type of city. Programs in emerging cities were more likely to exhibit community organization as a primary strategy. Programs in chronic problem cities were more likely to be characterized by social intervention and opportunity provision as primary strategies. The discriminant function procedure permitted us to correctly classify cities on the dimension of primary strategies in two thirds of the cases. Analysis of covariance, furthermore, suggested that community organization and opportunities provision were better strategies in achieving higher scores on general effectiveness than suppression, social intervention, or organization development and change. Community organization was a significantly effective strategy in emerging gang problem cities. Opportunities provision was a significantly effective strategy in chronic gang problem cities. Our final step in the search for promising approaches was to construct causal models using multiple regression analyses. First, using the variable of improved gang situation as our dependent or outcome measure (this is probably the most valid of the three component measures), we found in a probit regression analysis that the interaction of strategies of community organization and opportunities provision is the single strongest predictor. It accounts for 40.2 percent of the variation in our dependent variable, perceived improvement in the gang situation. The second significant predictor is the proportion of local respondents involved with each other in a city dealing with the problem. Together, these two predictors or independent variables account for almost 60 percent of the variance. We are unable, however, using this procedure, to find variables or factors that predict success in the emerging gang problem cities. We turned next to use of the general effectiveness score as the dependent variable for measure of success. For the chronic gang problem cities, we achieved an extremely potent set of predictors. The two primary strategies of intervention separately -- community organization and opportunities provision -- and community consensus on definition of gang incident account for 69 percent of the variance. The fourth variable that enters the regression equation is the proportion of agencies with an external advisory group. Together these four variables account for or can predict 82 percent of the variance in the general effectiveness score in chronic gang problem cities. The model for predicting general effectiveness in dealing with the gang problem in emerging gang problem cities is not so elaborate. Only community organization as a primary strategy contributes to an explanation of 31 percent of the variance in the outcome variable. This multiple regression model did not tell us much more than we already knew from our analysis of covariance, but a final set of analyses across all cities indicates that the primary strategies of community organization and opportunities provision along with maximum participation by key community actors is predictive of successful efforts at reducing the gang problem. Because of the very high levels of statistical prediction obtained using "perceived improvement" as a dependent variable or component of a general effectiveness measure, we carried out a special validity check. We obtained information on five empirical indicators of change in the gang problem between 1980 and 1987: numbers of gangs, gang members, gang-related homicides, gang-related assaults, and number of gang-related narcotic incidents. Statistical tests indicated a significant correlation between perceptions and empirical indicators, leading us to conclude that our "perceived improvement" measure was valid. Our report concludes with the recommendation that future policy and research test strategies that emphasize opportunities provision, particularly improved educational, training, and job opportunities for gang members and gang prone youth; and broad community organization or mobilization of key agency efforts, including development of common definitions of the problem and the possible use of external advisory groups to aid with the creation of agency gang programs. We observed that strategies of suppression and social intervention were present in to all of our cities and essential for dealing with the youth gang problem, but that success was more likely when community organization and opportunities provision strategies were also present and emphasized. ------------------------------ CHAPTER 1 INTRODUCTION The purpose of the survey of 254 respondents in 45 cities and 6 institutional sites, mainly correctional schools, was to discover promising approaches to dealing with the youth gang problem. Our key purpose in this research was to develop ideas for policy and program organization to be further tested and refined later under experimental or quasi-experimental conditions. In order to develop empirically based conceptions of promising programs, we collected data on youth gang activity and especially the response to it from respondents in criminal justice agencies, i.e., police, sheriff, court, probation, state's attorneys, parole, corrections, and other relevant justice units (including research and planning departments), as well as from a variety of community-based public and private agencies, such as schools, youth gang intervention projects, youth service organizations, family treatment agencies, neighborhood groups, churches, and others. We sought information on key aspects relevant to policy and program development: definitions of terms, such as gang, gang member, and gang incident, characteristics of the gang problem, the respondent organization's goals, activities, structures, strategies, as well as perceptions of change in the problem and effectiveness of intervention efforts. Our survey was partially guided by certain ideas about response to the youth gang problem developed in the course of our literature review. Historically, as well as contemporaneously, it seemed to us that the policies and programs of agencies and community groups exemplified various strategies or basic approaches and could be classified into four or five categories or combinations of them: local community mobilization, social intervention, provision of social opportunities, and suppression. A fifth modifying strategy was organizational development and change. Indicators of these strategies evolved and were further classified in the course of our survey. Our report utilizes two units of analysis: professional or agency respondent, with special emphasis on the information and views of law enforcement, and type of area -- chronic gang problem or emerging gang problem area. Our discussion alternates between category of respondent, such as law enforcement or non-law enforcement, or more specifically any of 13 professional orientations (e.g., corrections, school, etc.), and type of area (e.g., by city). In respect to certain problem characteristics such as number of gangs and gang members in a locality, police estimates were usually considered the most reliable, since the police are the primary collectors of this kind of information. Our area classification, in fact, includes not only cities and sites or institutions, but counties and subareas of cities. Often we refer to these subunits collectively as areas or localities. We are concerned with cities mainly, however. For empirical and policy, and also theoretical purposes, we define chronic and emerging gang problem areas as follows. A chronic youth gang problem area is one where the youth gang problem has been recognized as more or less serious for a long period, at least prior to 1980, is still present at the end of 1987, and where there has been an organized response to it, especially by the police or a criminal justice unit. An emerging youth gang problem area is one where the youth gang problem has been recognized as more or less serious since 1980, is present in 1987, and where there has been an organized response to it, especially by the police or criminal justice unit. By youth gang we are referring to a delinquent and/or criminal group containing young people mainly between the ages of 12 and 25 characterized by certain symbolic and/or communal as well as at times organized criminal purposes. We note that in this report we do not deal with certain other ways of framing our analysis, also possibly important for the development of policy, practice, and theoretical considerations. These alternate classifications or frameworks for more detailed analysis include size of cities and certain characteristics of gangs such as age, gender, and race/ethnicity of members; local or non-localized gangs, i.e., affiliated across jurisdictions and criminally organized versus traditional turf and not primarily criminally organized. These comparisons and the issues they pose are considered to some extent, in our present report but are not a primary basis for analyzing data. We have tried to keep technical discussion to a minimum in the body of the report. Supplementary material, often very important tables and figures, as well as procedural details, are provided in the Appendices. Even these are kept to a minimum. Additional explanatory materials are available from the authors. A variety of descriptive and exploratory inferential statistics have been used including percentages, means, medians, tri-means, correlations, tests of means, multivariate discriminant and regression analyses, multi- dimensional scaling and factor analysis. Nevertheless, our analysis must be regarded as essentially exploratory. In the remainder of this introductory chapter, we deal with the method and scope of study. In the second chapter on definitions, we try to discover what our respondents, in their own terms mean by the central terms--gang, gang member, and gang incident. This is an essential backdrop for the third chapter in which we examine the scope and perceived nature of the youth gang problem in the respondents' jurisdictions or service areas. In Chapter Four, we inquire into the extent the respondents have formulated policies and procedures and established structures for dealing with the problem, e.g., whether specialized training and special gang units or advisory groups exist. Chapter Five analyzes the underlying structure of approaches to the problem in terms of strategies of intervention, perceived causes, and, briefly, communication or network structures of the respondents, particularly on a city or site basis. Finally, in Chapter Six, we examine perceptions of a reduction in the gang problem and effectiveness of agency and interagency efforts and their relation to certain characteristics of the problem. We are particularly interested in exploring what works and does not work in chronic and emerging gang problem cities. Again we observe that our data have not been fully analyzed. The variables, classifications, and materials of the various chapters still need to be systematically interrelated. We hope to remedy these shortcomings and others that we become aware of in the near future. The present report, nevertheless, should be viewed as substantive. However, it should not be used as a basis for describing what the current nature and scope of the youth gang problem is nationally, nor what communities are doing generally about the problem. The findings are relevant only to those agencies, cities, and sites investigated which appeared to have reasonably well organized approaches to a significant youth gang problem, presently or in the recent past. ------------------------------ METHOD AND SCOPE Various sources were used to identify initial informants, including knowledgeable individuals, contacts with criminal/juvenile justice planning agencies in all states, representatives of agencies and programs known to be involved in organized programs to deal with gangs. We also advertised for participants in the survey through national agency, professional journal, and media sources. Telephone calls were then made to these identified potential respondents to discover whether gang-related problems existed, however defined, and the presence of an organized response. The aim of the initial screening process was to establish a fairly wide net for the inclusion of cities, agencies, and programs dealing with the problem of youth gangs. In this initial screening phase, hundreds of different agency and community organization representatives were contacted. Approximately 100 cities were screened by the end of 1988, 55 of which were excluded from our study either because they had no recognized gang activity or no response which specifically addressed youth gangs according to our screening criteria (see Appendix Table 1-1). A total of 360 questionnaires were distributed (see Appendix Table 1-2 for copy of the Questionnaire). The selection criteria included: presence and recognition of a youth gang problem, the presence of a youth gang program; the existence of a program for at least a year, whether at present or in the recent past; an articulated set of program goals; evidence of more than a simplistic unitary response to the problem, such as either only police arrests or youth gang recreation programs; and some capacity to describe the impact of the program. A total of 254 questionnaires and follow-up interviews were received and preliminary review suggested that most of these criteria were met. When in doubt we included an area even with a partially sustained program or with only a unitary approach. There were four (4) such cities. Our response rate (minimally) was 70.5 percent. The non-responses, generally speaking, included approximately four (4) agencies that we knew had eligible programs, but who for a variety of idiosyncratic reasons, explicitly refused to participate. Eleven (11) persons did not complete questionnaires because, although well informed generally (e.g., academics), they did not have specific program related information. We believe that the other potential respondents (91) who failed to return questionnaires did not have time to complete them or came from agencies that, in fact, did not meet the criteria of our survey. However, we identified and received questionnaires from representatives of at least one or two agencies in all eligible cities, except Wilmington, Delaware. The non-respondents were mainly in large cities such as New York, Chicago, Los Angeles, and Miami where we already had a sufficient proportion of respondents who could provide us with an adequate perspective on the citywide gang problem and program information relevant to their own agency as well as others. In essence, we have a better sample of cities than of constituent agencies. Nevertheless, we do not clearly have a sample from a known population of eligible cities or agencies. What we have in the study is a fairly large group of cities and agencies generally recognized to have youth gang problems and organized programs to address them. The cities, institutional sites, and the numbers of respondents in the different jurisdictions included in the study are listed in Table 1-3. The regions of the country represented in the survey are the northeast, midwest, south, and especially the west. The state with the largest number of cities (15), sites (3), and respondents (105) is California. The next largest gang program state represented in the study is Illinois with 5 cities and 26 respondents. Cities with the largest numbers of respondents are New York (15), Los Angeles (12), Chicago (9), and Miami (8). Several city/county jurisdictions (4) are listed with only one respondent. Table 1-3. Respondents in Analysis by Geographic Area (n = 45) and Selected Institutions (n = 6) (Number of Respondents Appear in Parentheses) NORTHEAST (21) New York (15) Philadelphia (5) Glen Mills Schools, PA (1)* MIDWEST (75) Benton Harbor (1)/Berrien County (2) Chicago (9)/Cook County (5) Cicero (2) Columbus (6)/Franklin County (4) Detroit (3)/Wayne County (1) Evanston (3) Flint (2)/Genessee County (1) Fort Wayne (3)/Allen County (1) Indianapolis (2)/Marion County (4) Madison (2)/Dane County (1) Milwaukee (7)/Milwaukee County (1) Minneapolis (2)/Hennepin County (5) Rockford (3)/Winnebago County (1) Sterling (3) Ethan Allen School for Boys, WI (1)* SOUTH (32) Atlanta (1) Fort Worth (1) Hialeah (1) Jackson (3) Louisville (2)/Jefferson County (4) Miami (8)/Dade County (7) Shreveport (3) Tallahassee (1)/Leon County (1) CALIFORNIA (105) Chino (2) East L.A. (1)/L.A. County (1)/CYA (2) El Monte (5) Inglewood (4) Long Beach (3)/LA County (1) Los Angeles City (12) Other Los Angeles County (20) Oakland (2)/Alameda County (1) Pomona (4)/LA County (1) Sacramento (3)/Sacramento County (1) San Diego (1)/San Diego County (9) San Francisco (6) San Jose (6)/Santa Clara County (1) Santa Ana (1)/Orange County (7) Stockton (5)/San Joaquin County (1) California Youth Authority [CYA] (3)* Milwaukee Paramount School Program (1)** Sunrise House (1)** OTHER WEST (21) Albuquerque (4)/Bernalillo County (2) Phoenix (2)/Maricopa County (1) Reno (3) Salt Lake City (4) Seattle (1) Tucson (2)/Pima County (1) Maclaren School, OR (1)* * denotes juvenile correctional institution. ** denotes an agency program selected because of its reported success. Table 1-4. Respondent Categories N = Category; n = Subcategories CATEGORY N %N n %n Law Enforcement 52 20.5 Prosecutors 26 10.2 Judges 14 5.5 Probation 32 12.6 Corrections 8 3.1 Parole 11 4.3 School 35 13.3 Academic 26 74.3 Security 9 25.7 Community/Service 62 24.4 Youth Service Agency 46 74.2 Youth & Family Service 8 12.9 Grass Roots 5 8.1 Comprehensive 3 4.8 Community Planning 12 4.7 Other 2 0.8 254 100.0 We classified the 254 respondents in ten major categories and six subcategories (Table 1-4): law enforcement (20.5 percent), including mainly police; prosecutors (10.2 percent); judges (5.5 percent); probation (12.6 percent); corrections (3.1 percent); parole (4.3 percent); schools (13.3 percent) subdivided into security and academic personnel,community service agencies and organizations (24.4 percent). This latter category was subdivided into youth service, youth and family service/treatment, comprehensive crisis intervention, and grass-roots groups. Two other categories were community, county or state planners (4.7 percent) and other (0.8 percent). The two largest categories were community agencies and law enforcement. All criminal justice agency personnel together comprised 56.2 percent of our respondent group. Our intent was to include the opinions and self- described practices of a variety of professionals or organizational representatives who confront the gang problem at the operational level, yet who have some responsibility for departmental or program operations and sometimes policy decisions. In other words, we selected and sent our questionnaires deliberately to the staff person who was most knowledgeable about the youth gang problem and the agency response to it. We believed foremost among the respondents in terms of such knowledge about gangs in the city as a whole, to be the police. They were the persons usually contacted first in the various cities. They were often the base or center point for the development of our set of respondents in each area. In 44 of the 45 specific city locales listed in Table 1-3, at least one police respondent is included. Finally, we observe that 21 of the cities were classified as chronic and 24 as emerging youth gang problem cities. A relatively greater proportion of our chronic gang problem cities have large or very large populations, but there are a sizeable number of small cities as well. Most of the cities classified as emerging youth gang problem cities have populations that are smaller, although some have populations over 500,000 (see Table 3-2 in Chapter 3). We were also able to classify the correctional schools and special agency programs or sites under these two rubrics. Our city or area analysis, however, will include primarily the 45 cities/counties and the California Youth Authority because it comprises a very large entity, relatively more related to a statewide jurisdiction. Schools and probation units tend to be more closely related to city and county jurisdictions. East Los Angeles is part of a county. Our respondent analysis will, of course, include all respondents whether from a city, county, sub-area, or institutional site (Table 1- 5). Table 1-5. Type of Gang Problem by Area (City, County, Site) Chronic Cities Emerging Cities (n=21) (n=24) Albuquerque, NM Atlanta, GA Chicago, IL Benton Harbor, MI Chino, CA Cicero, IL Detroit, MI Columbus, OH East LA, CA Evanston, IL El Monte, CA Flint, MI Inglewood, CA Fort Wayne, IN Long Beach, CA Fort Worth, TX LA City, CA Hialeah, FL LA County, CA Indianapolis, IN New York, NY Jackson, MS Oakland, CA Louisville, KN Pomona, CA Madison, WI Philadelphia, PA Miami, FL Phoenix, AZ Milwaukee, WI San Diego, CA Minneapolis, MN San Francisco, CA Reno, NV San Jose, CA Rockford, IL Santa Ana, CA Sacramento, CA Stockton, CA Salt Lake City, UT Tucson, AZ Seattle, WA Shreveport, LA Sterling, IL Tallahassee, FL Chronic Sites* Emerging Sites* (n=2) (n=4) California Youth Authority Ethan Allen School Sunrise House (CA) (WI) Glen Mills Schools (PA) McClaren Schools (OR) Paramount School (CA) * In much of the analysis, the California Youth Authority is the only chronic site included as an area along with the cities. ------------------------------ SUMMARY The survey of 254 respondents in 45 cities and 6 institutional sites was conducted as part of the assessment stage of a research and development program in cooperation with the Office of Juvenile Justice and Delinquency Prevention, U.S. Department of Justice. The purpose of the survey was to collect data on youth gang problems and especially the response to them in cities with promising programs. The areas represented in the survey are located in four regions of the country: Northeast, Midwest, South, and especially, the West. The state with the largest number of cities (15) and respondents (105) in our study is California. The next most well represented State is Illinois with 5 cities and 26 respondents. Our cities are classified into chronic (21) and emerging (24) gang areas. Our respondents were classified into 10 major categories and 6 subcategories. The major categories include law enforcement (52), prosecutors (26), judges (14), probation (32), corrections (8), parole (11), schools (35), community service (62), community planning (12), and other (2). We included in our survey professionals, or organizational representatives, who confront the gang problem at the operational level, yet who had some responsibility for departmental or program operations and sometimes policy decisions. The police were probably our most knowledgeable informants. They were the persons usually contacted first to obtain preliminary information about the local gang problem and which agencies had a program response to it. They were often the base or centerpoint for the development of our set of respondents from each area. Our questionnaire return rate was quite high - 70.5 percent. Only one city was not represented that we would have wanted in the survey, based on information available to us at the time. Most of the potential respondents not completing a questionnaire were from agencies in several of the largest cities where we already had adequate representation and a good knowledge of the gang problems and agency responses to it. What we have in the study is a fairly large group of cities and agencies generally recognized to have youth gang problems and organized programs to address them. The criteria for selection of cities and sites for inclusion in the study, liberally interpreted, were the presence and recognition of a youth gang problem; the presence of a youth gang program; the existence of a program for at least one year, whether at present or in the recent past; an articulated set of program goals; evidence of more than a simplistic or unitary response; and, some capacity to describe the impact of the program. A variety of simple and complex exploratory and descriptive statistical analyses are conducted. The basic intent of the study, as indicated in the final chapter, is to develop models of what would comprise successful programs in different types of areas or localities. Discriminant and multiple regression analyses are mainly used for this purpose. The study should be viewed as substantive, although a number of important type of analyses have yet to be carried out. ------------------------------ CHAPTER 2 DEFINITIONS OF GANG, GANG MEMBER, AND GANG INCIDENT It may be extremely difficult to assess accurately the scope or nature of the youth gang problem or the value of various suppression and intervention programs because a standard or reliable definition of the problem (particularly in its component parts, gang, gang member, and gang incident) does not exist, either across jurisdictions or often within the same jurisdictions over time. Various observers have identified this problem in recent years (Miller 1975, 1982; Moore 1978; Klein and Maxson 1989; Spergel 1989). Furthermore, a variety of definitions have been used over time, sometimes by the same researcher. The problem for theory, research, and policy development is not only that different and changing definitions have existed, but that the gang problem in its structural and dynamic characteristics seems also to be changing. What we have tried to do first in the survey questions asked and in the course of the present analysis is identify how respondents or local agency and community "experts" who deal with the problem on a regular basis perceive and conceptualize its key elements. We asked each respondent to state, however he or she wished, his or her definition of a gang (Item III-1 of the Questionnaire), a gang member (Item III-3), and a gang-related incident (Item III-6; See Appendix Table 1-2). We anticipated that some of the respondents, who would not be able to provide their own definitions, would use those of police agencies. The open-ended responses were then subjected to a content analysis. The following represents an effort to systematically decompose the various definitions of the three concepts, and thereby identify core or most frequently used elements. We believe this to have been an important exercise, but not entirely successful. But it helps set some parameters and indicates that our data interpretations and conclusions have limitations. We found that the respondents in our survey do not define gang problems the same way. Thus, the meaning of the gang problem at the present time must remain imprecise and varied, at least at the definitional level. The current chapter begins to identify systematically what we mean by the key definitional terms of the gang problem. In the long run, we must develop a standard definition of these terms for use in the criminal justice system so that we will be able to (1) understand the nature and scope of the problem across jurisdictions and time; (2) devise strategies of intervention that can be assessed comparably; and (3) test whether consensus on definition is itself an important characteristic of successful intervention within a particular jurisdiction. We attempt the latter in the final chapter of this report. The analysis of the definitions provides a context for understanding the various characteristics of gangs and the gang problem discussed in Chapter 3, the organizational response to the problem in Chapters 4 and 5, and effectiveness measures or program results in Chapter 6. It also provides a basis for selection of those components of the definitions that are consistent with the purposes of criminal and juvenile justice policy. In other words, the analysis may provide greater clarity, specificity, and a basis for understanding the youth gang, youth gang member, and youth gang incident, and thereby a more informed and rational choice of youth gang policy and program. The present chapter has been an extremely complex and time consuming undertaking. We have employed a variety of statistical procedures, including factor analysis and multi-dimensional scaling, to identify the characteristics which go into the definitions of gang, gang member and incident. The discussion that follows is both more detailed and yet less sufficiently developed than we would have desired. The discussion is somewhat technical, although much of it summarizes our more technical discussions in the Appendix. ------------------------------ THE GANG The 236 open-ended responses of definitions of gang [1] were first content-analyzed into 48 empirically anchored items or concepts. These definitions included criminal and non-criminal activities, gang organizational, and symbolic behaviors-- respectively, for example, drug sales, sports activities, affiliation with other gangs, gang name or colors. We were faced with many possible ways of combining these items as dichotomous variables into fewer higher level conceptual categories for analysis. We chose the groupings (Table 2-1) that are consistent with various empirically based definitions, theoretical considerations in the literature, and simply the frequency with which the items were mentioned. The first component of the definition of gangs we created represents (1) general group characteristics. We tried to select a set of structural and behavioral characteristics that might fit the delinquent group as well as the gang (See Klein 1971; Cartwright, Tomson, Schwartz 1975). The next component or subset of structural- behavioral characteristics are those that distinguish (2) the gang from the delinquent group and we label them gang organizational characteristics (Curry and Spergel 1988; Morash 1983). This assumes the gang may have a higher degree of organization than the delinquent group (Jacobs 1977; Miller 1975, 1982; Spergel 1984). The third set of characteristics are (3) non-criminal group activities (Whyte 1943; Suttles 1968). The next set of criminal activities comprise four subcategories of dichotomous variables: (4) drug crimes, (5) violent crimes, (6) property crimes, and a more general, or difficult to classify subset of criminal activity, (7) anti-social behavior. These categories are consistent with official agency and popular classifications. The many references to gang symbols lead us to separate symbols into three subcategories, those that are more (8) uniform across the gang and collectively owned, those that are more (9) personal in nature, and other (10) general or residual references to symbols that include, for example, gang "culture." Finally, two more sets of items are included that we do not combine with other variables: the presence of (11) adult members, which may be useful in describing the integration of adults and youths in youth gangs, and (12) an ecological notion, gang members residing in the same area. Table 2-1. Groupings of Dichotomous Variables Derived from Content Analysis of Definition of a Gang (Item III-1). n = number or percent of respondents referring to the concept 1. General Group Characteristics (n = 186, 73.2 percent) Group Certain Number of Members Leadership Loosely Knit Regular Association Common Behavior 2. Gang Organization Characteristics (n = 120, 47.2 percent) Affiliations with Other Gangs Goals/Purpose Organized Structure Established Identity Maintain Unity Exclusion of Non-Members Recruitment Membership Rules Conflict with Other Gangs 3. Non-Criminal Activities (n = 14, 5.5 percent) Protect Selves/Community Sports Activities Recreation Some Non-violent Activities 4. Drug Crime (n = 38, 15.0 percent) Drug Use Drug Sales 5. Crimes of Violence (n = 52, 23.2 percent) Violence Drive-by Shootings Robbery Rape Shooting Heavily Armed Intimidation 6. Property Crime (n = 21, 8.3 percent) Vandalism Burglary Theft Auto Theft 7. General Criminal Activity (n = 149, 58.7 percent) General Criminal or Anti-social Activity Control Criminal Economic Territory 8. Collective Symbols (n = 173, 68.1 percent) Gang Name Graffiti Control Turf Gang Logo Colors Initiation Rituals 9. Personal Symbols (n = 53, 20.9 percent) Hand Signs Clothing Tattoos Member Nicknames 10. General Reference to Symbols (n = 88, 34.6 percent) 11. Adult Members (n = 11, 4.4 percent) 12. Reside in Same Locale (n = 35, 13.8 percent) In order to make both sense of, and reduce, the diversity that exists among our respondents in defining the term gang, we turn to statistical analysis of covariance structure techniques. Factor analysis can produce a set of linear combinations that are unrelated to each other and which (1) can provide us with a general reduction in the number of concepts with which we have to cope and (2) assist us in interpreting complex perceptions of individual respondents (Johnson and Wichern 1982) [2]. In this case, factor analysis serves, for the most part, as an illustration of just how widely diverse our respondents' definitions of the gang are. Five factors that are relatively significant emerge from the analysis as can be seen in the Eigenvalues listed in Appendix Table 2-2. These five factors account for 41.7 percent of the variation in gang definition across the 236 actors. This is not a very high degree of accountability for the variation. The five factors that emerge from this analysis are potentially interpretable (See Table 2-3). They can be identified in order of strength of meaning. Factor 1 loads heavily on specific types of criminal behavior. This factor pictures the gang as a group highly involved in drugs, violence, and property crime. A residual or general crime category, however, does not load heavily on this factor. In other words, a respondent who has a high score on this factor is concerned about gangs because of the specific, probably serious crimes its members individually or collectively commit. One respondent who scored very high on this factor was the assistant director of a planning agency that funds community service projects. She defined the gang as "a loose knit organization with young adult leadership involved in a variety of crimes including vandalism, rape, robbery, assault, theft, burglary, and drug dealing." Table 2-3. Rotated Factor Matrix from GLS Factor Analysis of Tetrachoric Correlation Coefficients of Gang Definition Characteristics Factor 1 2 3 4 5 General Group -.185 -.054 .126 .888 .102 Gang Organization -.088 .088 -.022 .167 .008 Non-Criminal .072 .102 -.153 .100 .341 Drug Crime .571 .034 -.028 -.211 .148 Violence .614 -.043 -.012 .036 -.013 Property Crime .612 .153 -.002 -.046 -.018 General Criminal -.051 -.220 .947 .142 -.176 Collective Symbols .022 .494 .047 -.115 .240 Individual Symbols .011 .384 -.159 .053 -.047 General Symbols .115 .766 -.058 .033 -.081 Adults Involved .042 .033 .014 .048 .286 Reside Same Locale -.028 -.077 -.031 -.046 .75 Factor 2 loads most heavily on all three symbolic components of the definition of gang (group, individual, and general). This factor also shows a moderately negative loading for general or other types of crimes. This factor defines the gang on the basis of its symbolic components. Respondents who score high on this factor view the gang as a highly visible social entity characterized by ritual and tradition. This symbolic, communal, yet "mildly" violent character of the gang is emphasized in the recent works of Moore (1978) and Horowitz (1983). A director of the department of pupil personnel in a public school system scores high on factor 2. His definition of a gang is "a collection of youth with initiation and membership rituals, and identification through caps, bandannas, and jackets. Some have specific turf. Some do not. The snatching of gang jewelry and assaults are common gang incidents in our schools." Factor 3 loads very heavily on general criminal behavior, including efforts to control a situation for criminal purposes, slightly positively on general group characteristics and slightly negatively on non-criminal activities and individual symbols. This factor may be viewed as a definition of a delinquent or criminal group. Respondents that have high scores on this dimension are likely to answer questions about gangs by reference to groups of youth and adults with tendencies toward criminal behavior or actual commitments to criminal careers. A parole agent who scores high on factor 3 defines the gang this way. "The gang involves itself in [a lot of] negative activity [in whatever they do] . . . " Factor 4 loads most heavily on general group characteristics. It has a modest loading on gang organization characteristics. It also has slightly positive loadings on general criminal activity as well as non-criminal activity. It has negative loadings on drug involvement and collective symbols. Individuals who score high on this factor are likely to conceptualize the gang as a collectivity or organization involved in a range of criminal and non-criminal activities, but not engaged in drug crime and collective symbols. It is the group or organizational character of the gang rather than specific behavioral activity which distinguishes this factor. Several respondents from community or youth service agencies favor this definitional emphasis. It may represent a special concern about a set of youth who are identified as running together, who can be destructive and have the potential for criminal organization. The fifth factor loads most heavily on the gang's residence in the same neighborhood or area. It has the highest loadings of the five factors on non- criminal activity and adult involvement, and next to the highest loading on collective symbols. It has a modest positive loading on drug crime. Factor 5 perhaps represents the classic view of the street corner group, not seriously delinquent or criminal. The gang is the group of fellows who grew up together, who like to have a good time partying, who drink and use drugs, and are also positively related to adults in their neighborhood. It may be correct to label this definitional configuration as the classic street corner group anchored in a particular neighborhood. This is the definitional perspective that is perhaps emphasized in the work of Whyte (1943), Suttles (1968), and Short and Strodtbeck (1965). The coordinator of a school youth services program scores high on Factor 5. She defines the gang as "a group joined according to neighborhood for companionship, support, and acceptance. The members hangout together--drinking, committing truancy . . . " The factors as they are defined are abstractions generated from the patterns of covariation among the twelve groups of definitional characteristics conceptualized above. They reflect mainly emphasis in the definition of gang. Just because an individual scores highly on one factor does not mean that he will not score moderately high on another. We also observe that 22 respondents simply referred to a definition supplied by another source or internal agency group. ------------------------------ DIFFERENCES IN GANG DEFINITION BY AREA AND RESPONDENT Since we are not entirely pleased with the strength of our factor analysis, we return to the original twelve dichotomous measures of the definition of gang. It is possible that consensus on components of the definition will give us some idea about key elements and what influences their selection. We are interested in the degree of consensus on the use of these measures first by area and second by respondent category. We are also interested in whether the locality or the respondent category is more important in contributing to consensus on definition. Is it the factor of professional or agency orientation which determines the definition of gang across localities, or does city or locality exercise the predominant influence in the distinctive definitions that result? We devised a measure of consensus such that if all members of a locality or in the respondent category use a particular component of the definition of gang, the value for the particular measure or the average of the 12 measures is 1.0; if all members do not use the component, the value is zero. The value in terms of both equal use and non-use of the measure by the members of the locality or respondent group approaches 0.5. All sites or institutions where only one respondent answered were eliminated from the site or city analysis, but not from the analysis by type of respondent. The results were further transformed and scaled into a continuous interval measure to obtain average scores by city or respondent category. We find that consensus is generally, but not significantly, higher among respondents within emerging cities than within chronic cities (Table 2-4). In part this could be accounted for by the fact that we obtained more respondents in the chronic, usually the larger cities, then from the emerging, usually the smaller cities. We note that consensus is generally higher among criminal justice than among non-criminal justice personnel. It is highest among law enforcement, corrections, and comprehensive crisis intervention agencies. The differences between criminal justice and non- criminal justice personnel is significant at the 0.5 level using a t-test of means (Table 2-5). Table 2-4. Average Degree of Consensus on Gang Definitional Criteria by Area. Average Degree Area of Consensus Albuquerque* .53 Benton Harbor .78 CA-CYA* .83 Chicago* .62 Chino* .67 Cicero .75 Columbus .52 Detroit* .56 East LA* .63 El Monte* .50 Evanston .78 Flint .61 Fort Wayne .71 Indianapolis .69 Inglewood* .56 Jackson .83 LA City* .63 LA County* .54 Long Beach* .70 Louisville .64 Madison .78 Miami .51 Milwaukee .69 Minneapolis .64 New York* .60 Oakland* .83 Philadelphia* .67 Phoenix* .56 Pomona* .50 Reno .67 Rockford .79 Sacramento .67 Salt Lake City .58 San Diego* .67 San Jose* .67 San Francisco* .58 Santa Ana* .64 Shreveport .78 Sterling .50 Stockton* .72 Tallahassee .67 Tucson* .56 * Chronic Areas Table 2-5. Average Degree of Consensus on Gang Definitional Criteria by Respondent Category. Respondent Average Degree Category of Consensus Law Enforcement .69 Prosecution .61 Court .63 Probation .56 Corrections .67 Parole .61 School Security .44 School-Academic .49 Youth Services .51 Comprehensive .67 Family Services .52 Grass Roots .60 Planning .46 The t statistic for the difference in means between criminal justice and non-criminal justice categories is -2.54, significant at 0.05 level. The standard deviation for criminal justice respondents is .248; for non-criminal justice respondents is .260. Table 2-6 indicates that there is most agreement within area and by respondent category on the importance of the following gang definitional components: adults involved, non-criminal behavior, and property crime. There is a moderate amount of agreement on such components as residence in the same locality and participation in drug crimes. There is least agreement on importance of the idea of gang structure. Apparently some respondents believe a gang is an organized structure, others do not. The components of gang violence and use of symbols seem to be of moderate importance. Perhaps of even greater importance for our present discussion is that agreement on these definitional components is systematically higher on a city or site basis than by types of respondent or professional perspectives. The differences are statistically significant at the 0.01 level (See Table 2-6). The t statistic for the difference in means is significant at the 0.01 level. Table 2-6. Average Degree of Consensus for 42 Areas and 13 Respondent Categories by Definitional Criteria. Average Average (42 Areas) (13 Categories) General Group .63 .55 Gang Structure .35 .25 Non-Criminal .91 .89 General Criminal .39 .26 Drug Crime .75 .66 Violence .59 .46 Property Crime .83 .81 Collective Symbols .52 .53 Individual Symbols .62 .54 General Symbols .57 .27 Adults Involved .92 .94 Reside Same Locale .73 .72 Average Degree Standard of Consensus Deviation Over 42 Areas 0.651 .098 Over 13 Respondent 0.574 .083 Categories This pattern suggests that the influence of area is greater than that of the particular discipline in the determination or selection of the components of gang definition. In other words, it is not so much the profession of the person, how he was trained, or his agency affiliation which determines the definition of gang as where he or she is located by city or locality in carrying out his job. There appears to be a stronger geographical or physical proximity than a professional network effect. We will return to the notion of networks later in Chapter 5. Finally, we plotted the relation of the areas, cities, counties, sub-areas, and institutional sites to each other in terms of social distance. The 12 criteria of gang definition were used to compare the "definitional distance between each pair of multi-respondent geographic areas from each other." After applying a set of computational procedures, we were able to assign each site a value on each of two principal dimensions. We found that respondents in our chronic areas seemed to emphasize the general criminal or anti-social nature of the gang. The emerging cities tended to emphasize the symbolic aspects in their description of gangs. The chronic gang problem cities also more often defined their notion of gangs in drug problem terms, while the emerging city respondent defined the idea of gang slightly more often in organizational structure and general gang characteristic terms (See Appendix Table 2-7 and Appendix Figure 2-1). Thus respondents in the cities with long-term youth gang problems appear to emphasize the criminal and drug trafficking character of the problem. The cities with more recent gang problems see the problem in classic symbolic and group terms. This difference may also represent a phase in the development of the youth gang problem from that of a traditional to a more criminalized and sophisticated, although not necessarily more visible, form. The multi-dimensional scaling approach was also used to plot the relation of the different types of justice system, school, and community-based agency respondents to each other in terms of two- dimensional space. We find judges and youth and family service agencies scale low on the definitional components of collective symbolism and involvement in drug crime. The comprehensive crisis intervention service agencies score high on the use of collective symbols in their identification of gangs. Grass-roots and planning agencies score highest of all categories of respondents in the use of violence and property crime in their definitions of the gang. Corrections scores lowest on this dimension. Prosecution, parole, and to some extent law enforcement appear to equally emphasize a variety of components of the definition of gang and are more or less centrally situated along the social distance dimensions, particularly in their use of such components as collective symbols, involvement in drug crime, general group characteristics, gang structure, violence and property crime (See Appendix Tables 2-10, 2-11, and 2-12). ------------------------------ SUMMARY: GANG DEFINITION Our procedures and findings in regard to a definition of the term gang may be summarized as follows. Two hundred thirty-six open-ended responses as to definitions of gang were content analyzed into 48 empirically anchored items or concepts, which were then grouped into twelve categories: 1) general group characteristics; 2) gang organization characteristics; 3) non-criminal group activities; 4) drug crime; 5) violent crime; 6) property crime; 7) general criminal activity; 8) symbols--collectively owned; 9) personal in nature, and still other more 10) general symbolic references; 11) the presence of adult members; and 12) residence in the same area. We then factor analyzed these twelve sets of characteristics to try to obtain fewer and more succinct patterned meanings to the notion of gang. We came out with five factors: a set of specific criminal activities factor, a symbolic factor, a more amorphous general delinquency factor, a group or organizational factor, and a residence or neighborhood factor. We could not account statistically (only 41.4 percent) for as much variance or meaning as we would have liked. Next we returned to use of our original twelve conceptual component grouping of what a gang was based on respondent definitions. We were interested in how much consensus there was on use of these components by respondents on the basis of area and professional affiliation or discipline. We found that consensus on use of gang definitional components tended to be a little higher in emerging than in chronic areas. Consensus was significantly higher among criminal justice than non-criminal justice respondents. In general there was greater agreement on definition of the term gang by area than by respondent categories. The difference was statistically significant at the .01 level. Finally, we used a multi-dimensional scaling procedure to locate area and then respondent categories in social distance space in terms of the component dimensions. We compared the definitional distance of our respondents from each other on an area and professional or disciplinary basis. We found that respondents in the chronic gang problem cities emphasized the general criminal or anti- social nature of the gang. Respondents in the emerging cities emphasized symbolic and organizational aspects in their definitions of gangs. ------------------------------ THE GANG MEMBER We used the same procedures and steps in analyzing the term gang member as we did in examining the definition of the concept gang. The content analysis of open-ended responses revealed nine different subsets or dichotomous variables of ways to identify gang members: symbols/symbolic behavior, self-admission, association with gang members, gang type of criminal behavior (M.O.), location of residence, police identification, informant identification, and other criminal and non-criminal justice identifiers (See Table 2-13). Table 2-13. Methods of Identifying Gang Members Method Respondents Citing % Symbols/Symbolic Behavior 124 51.0 Self Admission 113 46.5 Police Identification 102 42.0 Informant Identification 97 38.9 Association with Gang Members 94 38.7 Type of Criminal Behavior 82 33.7 Other Criminal Justice Identifiers 46 18.9 Other Non-criminal Justice IDers 39 16.0 Location or Residence 34 14.0 For those who make their own identification of gang members, symbolic behavior, patterns of association, types of criminal behavior, and hangouts or residence are important. For those who use someone else's definition of membership, a statement by the gang member about himself is the most commonly cited source, but also important are identification through police, informants, other criminal justice institutions, and other non- criminal justice. We performed a GLS factor analysis on these nine ways of identifying gang members for the 243 respondents who answered this item, using the same method employed in the analysis of gang definition characteristics. Four significant factors resulted that explain 36.4 percent of the covariation among the variables, which again is not a great deal of explanatory power (See Appendix Table 2-14). The first factor is totally dominated by the component of symbolic behavior (See Table 2-15). There is a moderate positive loading on type of crime that a suspect is engaged in and a slightly stronger negative loading on the use of police identification. The definition of a government planner who scores high on this factor mentions "turf and colors." Factor 2 carries the heaviest loading for other non-justice system institutional identification but also has positive loadings on police identification and other legal institutional identification. Criminal behavior loads negatively on Factor 2. A probation officer who scores high on Factor 2 and defers to police agencies, school officials, parents, and probation department investigation. His sources of identification are court officials, police, and other youths in the community. In other words, a youth is considered a gang member mainly because some other organization says he is. Factor 3 loads most strongly on identification by gang member association. Other strong to moderate loadings include informant identification, criminal behavior, and self-identification. A police officer scores highest on Factor 3. His methods of identifying gang members include "association with gang members, information from Boys Clubs, schools and law enforcement agencies, methods of questioning, and checking photos of all gang and delinquent groups on file." Factor 4, accounting for less than 4 percent of the explained variation, loads most heavily on a combination of sources, other criminal justice, police, and self- identification, and location. Table 2-15. Rotated Factor Matrix from GLS Factor Analysis of Tetrachoric Correlation Coefficients of Gang Incident Definition Characteristics Factor 1 2 3 4 Association .052 -.104 .742 -.062 Self ID .037 .012 .237 .354 Police ID -.211 .236 .047 .331 Informant ID -.076 .141 .359 .084 Other Crim. Justice ID -.017 .360 .014 .397 Other Institution ID .023 .749 -.012 .106 Location .048 .012 .042 -.365 Symbolic Behavior .984 -.032 .051 -.168 Criminal Behavior (M.O.) .136 -.232 .281 .080 ------------------------------ DIFFERENCES IN GANG MEMBER IDENTIFICATION Once more we explore the issue of consensus, as it may provide greater understanding of how and under what conditions a person is defined as a youth gang member, i.e., on which components there is greatest agreement. We analyze our nine dichotomous criteria of gang membership again first by area and second by respondent category. We determine whether there is greater consensus by locality or respondent category on use of gang member definitions. We use the same procedure and measure of degree of definitional consensus, ranging from 0 for no consensus in definition to 1 for perfect consensus. In an examination of the area-level data, we find once more that consensus is slightly higher within emerging cities than within chronic cities. The differences are not statistically significant (See Table 2-16). However, when we examine the degree to which there is consensus in the respondent categories in use of components of the definition of gang member, we find that unlike the definition of gang, there is more consensus among non-criminal justice agencies than criminal justice agency respondents. Consensus is highest among grass-roots organizations and comprehensive youth service intervention programs and lowest for probation, parole, and law enforcement officers. The differences between criminal justice and non- criminal justice personnel is significant at the 0.05 level, using a t-test of means (See Table 2- 17). Table 2-16. Average Degree of Consensus on Membership Criteria by Area. Average Degree City of Consensus Albuquerque* .48 Benton Harbor .41 CA-CYA* .63 Chicago* .51 Chino* .56 Cicero .56 Columbus .38 Detroit* .56 East LA* .22 El Monte* .47 Evanston .70 Flint .48 Fort Wayne .50 Indianapolis .59 Inglewood* .61 Jackson .70 LA City* .57 LA County* .31 Long Beach* .64 Louisville .56 Madison .85 Miami .39 Milwaukee .33 Minneapolis .56 New York* .67 Oakland* .63 Philadelphia* .51 Phoenix* .56 Pomona* .47 Reno .56 Rockford .39 Sacramento .50 Salt Lake City .44 San Diego* .40 San Jose* .33 San Francisco* .63 Santa Ana* .56 Shreveport .63 Sterling .48 Stockton* .44 Tallahassee .89 Tucson* .63 * Chronic Areas Table 2-17. Average Degree of Consensus on Member Identification Methods by Respondent Category. Respondent Average Degree Category of Consensus Law Enforcement .34 Prosecution .40 Court .49 Probation .30 Corrections .44 Parole .33 School Security .53 School Other .49 Youth Services .49 Comprehensive .70 Family Services .50 Grass Roots .73 Planning .52 The t statistic for difference in means is -2.54, significant at the 0.05 level. The standard deviation for criminal justice respondents is .248 and for non-criminal justice respondents is .260. Table 2-18 indicates that there is most consensus across areas and across the different respondent categories in regard to components of the definition that indicate the individual identified as a gang member lives in a particular neighborhood and when the source of identification is another institution. Apparently there is least consensus on the importance of symbolic behavior as a way of identifying a gang member and most consensus on components that respondents are less likely to use. Again, we find that agreement of the definitional components is generally higher on a city or site basis than by types of respondents or disciplinary perspectives (See Table 2-18). These differences are quite large and statistically significant at the .001 level. As with the definition of gang, we plotted respondent definitions of gang member, particularly the 9 criteria, in social distance terms. We did this for cities or areas as well as for types of respondents. We found that most cities tend to use a variety of criteria in identifying gang members. The respondents in the chronic problem cities rely less on police and other criminal justice institutional identification than in the emerging gang problem cities. Furthermore, the emerging cities tend to be outliers, some relying very heavily, some not at all, on police and other criminal justice agency identifications. The multi- dimensional scaling analysis is somewhat more revealing in reference to respondent categories. It locates criminal justice respondents generally higher on the dimension which emphasizes specific types of crimes gang members commit or by their M.O. Non-justice entities, such as grass-roots, comprehensive intervention service, and family service agencies are generally more likely to rely on symbolic behaviors to identify youth as a gang member (See Appendix Tables 2-19 and 2-20 and Appendix Figures 2-3, 2-4). Table 2-18. Average Degree of Consensus for 42 Areas and 13 Respondent Categories by Member Identification Criteria. Average Average (42 Sites) (13 Categories) Symbolic Behavior .41 .39 Association .48 .51 Criminal Behavior .45 .51 Location/Residence .75 .81 Self-ID .41 .41 Police ID .44 .45 Informant ID .41 .53 Other Legal ID .70 .69 Other Institutional ID .74 .67 Average Degree Standard of Consensus Deviation Over 42 Areas .744 .139 Over 13 Respondent .625 .062 Categories The F statistic for the difference in standard deviations is significant at the 0.01 level. The t statistic for the difference in means is significant at the 0.001 level. ------------------------------ SUMMARY: GANG MEMBER DEFINITION The content analysis revealed the following nine ways to identify gang members: symbols/symbolic behavior, self-admission, association with gang members, gang type of criminal behavior (MO), location of residence, police identification, informant identifications, other criminal justice identifiers, and other non-criminal justice identifiers. A factor analysis was conducted and resulted in four significant factors that explained 36.4 percent of the covariation among the variables, not a great deal of explanatory power: symbolic behavior; other non-justice and to some extent police identification; gang member association; and a combination of sources, including other criminal justice, self- identification, and location. Our inquiry into the degree of consensus by respondents on gang member identification revealed that consensus is slightly higher in emerging than in chronic gang problem cities. There tends to be more consensus among non-criminal than criminal justice agency respondents. Consensus occurs most often in relation to where the individual lives and least often on the importance of symbolic behavior in the determination of who is a gang member. Again we find that agreement on gang member definitional components is significantly higher on an area than on a type or category of respondent basis. It is difficult to find different patterns of identification of gang members comparing emerging and chronic gang problem cities. Some emerging cities tend to be somewhat uni-dimensional, some relying very heavily, some not relying at all on police and criminal justice sources. Ways of identifying gang members are more distinctive among categories of respondents. Criminal justice respondents tend to emphasize specific types of crime or MO's of reported offenders. Non-justice system respondents generally rely more on symbolic behaviors to identify youth as gang members. It is possible that some of these patterns of youth gang member definitions would be clarified, as with the definition of gang, if race/ethnicity as well as age control factors were interrelated. ------------------------------ GANG-RELATED INCIDENTS The gang-related incident is the most important criterion for determination of the scope and severity of the youth gang problem in a particular locality, especially based on police definition. It is the measure most frequently used by the police, media, and researchers, as well as by policy makers, to assess the scope and severity of the youth gang problem. It is also, relatively speaking, the most reliable indicator since it focuses specifically on criminal activity. Nevertheless, the concept of gang-related incident is clearly dependent on the particular definition of gang and gang member in a locality or by type of respondent. Problems of validity and reliability are also expected to be compounded because of political and policy considerations about the impact of the definition of gang incident, especially on a city basis. A content analysis suggested nine definitional components or subsets of characteristics of a gang incident. The presence or involvement of a gang member was the most frequently reported criterion in identifying an event or an incident as gang- related among 61.4 percent respondents. Gang function, i.e., a criminal activity based on gang member motivation or gang interest, was used as a criterion by 39.1 percent of the respondents. Modus operandi (MO) is another criterion used by 37.8 percent of the respondents. Certain crimes such as drive-by shootings were almost universally regarded as gang-related. However, the type of MO used varied considerably in different cities. Some respondents reported treating all auto thefts as gang-related. In one city, a respondent reported that he considers all drug crimes as gang-related. The involvement of multiple offenders per se at an event (15.9 percent) and evidence of the involvement of rival gang youth (15.9 percent) were sufficiently frequent responses for us to establish these as separate categories. Identification of an activity by another institution or group as gang related (11.2 percent) and by where the event occurred (10.7 percent) were criteria reported by some respondents. Least frequently reported as the basis for respondents deciding that an activity was gang related were claims by the gang itself that an incident was gang related (4.7 percent) and simply that several youth were involved in the activity that was not necessarily of a criminal nature (4.3 percent) (See Table 2-25). Thus, at this first level of analysis, the key characteristics of the gang incident definition seemed to be whether a youth gang member was involved (sometimes called gang-related incident); whether the event furthered gang function or development (sometimes termed gang motivated incident); and whether a certain commonly recognized criminal gang act occurred, such as a drive-by shooting, i.e., modus operandi. Table 2-25. Definitions of a Gang-Related Incident Criteria Respondents Citing % Gang Member Involved 150 64.4 Furthers Gang Function 91 39.1 Gang Modus Operandi 88 37.8 Incident Involves Multiple Offenders 37 15.9 Rival Gangs Involved 37 15.9 Other Institution/Group Identifies 26 11.2 Location of Incident 25 10.7 Gang Claims Incident 11 4.7 Youth Involved 10 4.3 A GLS factor analysis of the tetrachoric correlation coefficients of these nine measures of gang incidents produced five factors that combine to explain 43.3 percent of the covariance among the variables (Appendix Table 2-26). The explanatory power of these factors for gang incidents is somewhat more than those of gang and gang member, yet still explains less than 50 percent of what goes into a definition. The first factor loads very heavily on the involvement of a gang member in the incident. It has a fairly strong negative loading on other institutional identification. The second factor loads very heavily on the characteristic of modus operandi or distinctive type of offense by the participants in the incident. There are small negative loadings on institutional identification, gang members involved, and a smaller positive reference to rival gang involvement (See Table 2- 27). The third factor loads heavily on the pair of variables that indicate a number of persons were present and that they were young people. There is a positive loading on rival gangs involved, and moderate negative loadings on institutional identification and that the gang claims the incident. Table 2-27. Rotated Factor Matrix from GLS Factor Analysis of Tetrachoric Correlation Coefficients of Gang Incident Definition Characteristics Factor 1 2 3 4 5 Member Involved .986 -.128 -.088 .042 -.011 Modus Operandi -.073 .996 -.025 .011 -.009 Gang Function -.001 -.107 -.028 .676 -.116 Number Involved .069 .046 .708 .066 -.148 Rival Gangs .024 .102 .181 .282 .154 Youth Involved .016 -.064 .467 -.032 .098 Gang Claims Incident .071 -.005 -.142 -.046 -.027 Institutional ID -.329 -.133 -.167 -.337 -.072 Location .005 -.007 .039 .000 .440 Identification by gang function dominates the fourth factor. It also loads negatively on institutional identification and positively on the involvement of rival gangs, probably engaging in gang conflict. The fifth factor explains a relatively small part of total variation and loads most heavily of any of the factors on location. It loads somewhat negatively on number of individuals involved and positively on the involvement of rival gangs (See Table 2-27). Thus the order of importance of these criteria for the definition of gang incident shifts as a result of the factor analysis. The gang member related component remains most important, but the importance of gang function is reduced, and the modus operandi criteria becomes more important. The tendency is thus to an expansive definition of the idea of gang incident. The criterion of gang function tends to be relatively restrictive. The following is illustrative of the variety of gang incident definitions obtained and how they load on the different factors. A police lieutenant who includes in his methods of identifying a gang incident "known gang members involved" scores highest on Factor 1. He also includes in his definition "gang MO" and "multiple youth suspects", and, hence, shows a relatively high score for factors 2 and 3. A public school administrator and a deputy district attorney score highest on Factor 2. The school administrator states in his definition that "two or more identified gang members are involved in an incident that results in a hostile and/or illegal situation. We try to determine if the incident is related to gang function. Most incidents involve gang fighting." This respondent also has relatively high scores on factor 1, as well as Factor 4. The deputy district attorney states that his agency identifies a gang incident as one in which "the suspect and/or victim is a known gang member. The incident is motivated by an exchange of gang challenges. Colors, graffiti, and signing may be involved. This includes classic gang tactics such as drive by shootings and attacks involving multiple assailants." Likewise he also scores high on Factors 1, 3, and 4. An assistant planning director of a local United Way scores highest on Factor 3. She states that her agency identifies a gang incident as one that "involves a significant number of youth." A police sergeant from a southern city and the founder of a grassroots community organization score highest on Factor 4. The police sergeant identifies a gang incident as "violence between two groups that escalates to the killing or assault level. The conflict is due to group identity. Crimes are a primary part of group identification." The grass- roots community organization respondent states that a gang incident is when a "gang fights for turf to sell drugs and over gang recruitment." Both these latter respondents score fairly low on factors 1 and 2 and moderately high on Factor 5. Despite our factor analysis, there still seems to be a good deal of overlap in these components or they tend to be employed jointly in a definition of gang incident. ------------------------------ DIFFERENCES IN DEFINING GANG INCIDENT BY AREA Next we analyze our nine dichotomous criteria for classifying an incident as gang-related first with location as the unit of analysis and second with respondent category as the unit of analysis. As with definitions of gang and gang member, we are interested first in the degree of consensus on these components by respondents within sites and then by category of respondents across sites or areas. We are especially interested in whether the area or the type of respondent category contributes more to consensus on components used in the definition of a gang incident. Again we use our measure of consensus, which ranges from zero (non- consensus) to one (perfect consensus). Contrary to our findings comparing consensus on definitions of gang and gang member, there is at least as great consensus in chronic gang cities as in emerging gang problem cities on what the criteria of a gang incident are. This is so despite the fact it is more difficult to achieve consensus in chronic problem cities, which tend to be larger and from where we had more respondents. In part, this could be because chronic gang problem cities have longer experience in and better information systems for dealing with gangs and gang members. The differences comparing means is not statistically significant between these two kinds of gang problem cities (See Table 2-28). Again, we observe that consensus is slightly higher among criminal justice than among non-criminal justice personnel. It is highest among court and prosecutors, and quite high among law enforcement officers, considering the large number of police personnel in that respondent category with so much greater potential for dissonance. Still there is no statistically significant difference in degree of consensus comparing criminal and non-criminal justice respondent groups (See Table 2-29). Table 2-30 suggests that there is greater agreement on the components of the definition of a gang incident which are least frequently used: when there are youth present, when there is self- admission, and when the incident occurs in certain parts of the city. There appears to be least agreement by area and among the different respondents on the use of gang function, modus operandi, or the involvement of gang members as identifying elements in a definition of a gang incident. Again, we see that influence of the city or area is considerably greater in the development of consensus on gang incident definitional components then respondent categories. Locality is a greater influence on what goes into a definition of the gang problem than the particular agency or respondent's discipline. These differences are statistically significant at the 0.01 level. The influence of the area on a consensual definition of the gang incident is even greater than for the definition of gang or gang member. Table 2-28. Average Degree of Consensus on Incident Criteria by Area. Average Degree Area of Consensus Albuquerque* .74 Benton Harbor .93 CA-CYA* .93 Chicago* .62 Chino* .56 Cicero .78 Columbus .41 Detroit* .78 East LA* .56 El Monte* .56 Evanston .70 Flint .67 Fort Wayne .56 Indianapolis .70 Inglewood* .78 Jackson .63 LA City* .59 LA County* .73 Long Beach* .82 Louisville .63 Madison .70 Miami .72 Milwaukee .67 Minneapolis .78 New York* .71 Oakland* .78 Philadelphia* .69 Phoenix* .78 Pomona* .69 Reno .67 Rockford .83 Sacramento .78 Salt Lake City .56 San Diego* .53 San Jose* .84 San Francisco* .78 Santa Ana* .67 Shreveport .78 Sterling .78 Stockton* .70 Tallahassee .78 Tucson* .78 * Chronic Areas Table 2-29. Average Degree of Consensus on Incident Identification Methods by Respondent Category. Respondent Average Degree Category of Consensus Law Enforcement .66 Prosecution .70 Court .78 Probation .56 Corrections .64 Parole .58 School Security .56 School Other .59 Youth Services .59 Comprehensive .63 Family Services .59 Grass Roots .60 Planning .65 Table 2-30. Average Degree of Consensus for 42 Area and 13 Respondent Categories by Definitional Criteria. Average Average (42 Sites) (13 Categories) Gang Member Involved .56 .26 Furthers Gang Function .46 .27 Gang Modus Operandi .46 .27 Multiple Offenders Involved .72 .63 Rival Gangs Involved .72 .68 Youth Involved .91 .95 Gang Claims Incident .93 .92 Other Institution Identifies .80 .79 Location of Incident .78 .85 Average Degree Standard of Consensus Deviation Over 42 Areas .7067 .108 Over 13 Respondent .6254 .062 Categories * The F statistic for the difference in standard deviations is significant at the 0.05 level. ** The t statistic for the difference in means is significant at the 0.01 level. Again, we plotted the relation of the cities and then the categories of respondent to each other, respectively, on a social distance map. The multiple dimensional scaling procedure first produced a clustering of the chronic gang problem cities and suggested a balance of multiple components by them that enter into a definition of gang incident. Differences in definition among some of the large cities, such as Los Angeles City, Los Angeles County, and New York were relatively limited while the gang incident definitions of the emerging cities were more uni-dimensional and tended to be at the extremes on the dimensions established in the analysis, for example, emphasizing gang membership or gang function (See Appendix Tables 2-31, 2-32, 2-34, 2-35 and Appendix Figure 2-5). Views of gang incident by the different types of respondents were also quite distant from each other in social space. There is a tendency, however, for respondents in the different categories to depend somewhat on other agencies probably with the exception of the police on the determination of whether an incident is gang related or not (See Appendix Figure 2-6). Thus, the distinctions that some of the very large and chronic gang problem cities draw as to the definitions of gang incident may be less than commonly regarded, at least compared to the variety of, or more disparate, definitions in the emerging or newer gang cities. ------------------------------ SUMMARY: GANG INCIDENT DEFINITION The gang incident is the most important indicator of the scope and severity of the gang problem. The respondents listed the following criteria for identifying and reporting a gang incident: involvement of a gang member, gang function (based on gang motivation or interest), modus operandi, involvement of multiple offenders, rival gangs involved, youth involved, gang claims the incident, other institutional identification, and the location of the event. A factor analysis further reduced the key component ideas of gang incident to: gang member participation, modus operandi, presence of a group of gang people, gang function, and location. There seemed to be little difference by type of problem city -- chronic or emerging -- on what the criteria of a gang incident were. Consensus was slightly but not significantly higher on the components of the definition among criminal justice respondents, especially judges and prosecutors, than among non-criminal justice personnel. The influence of area was much larger than that of respondent category on the level of consensus on definition of the gang incident. The difference was highly statistically significant. The attempt to distinguish emphases in definition, using multi- dimensional scaling, revealed a few differences by type of area or category of respondent. The definitions of the larger chronic gang problem cities were based on multiple criteria and were reasonably close to each other. Those of the emerging cities were more extreme in their emphasis, for example, either of gang membership or gang function. ------------------------------ SUMMARY This chapter examined the multiple meanings of the gang problem in terms of its key components: gang, gang member, and gang incident. The importance of a common definition was stressed for understanding the scope and seriousness of the problem across different cities and jurisdictions and for developing effective policies and programs of intervention. The variety of definitions used by respondents also suggests caution in the use of assessments of the problem including the findings presented later in this report. In general, there was more consensus on definitions based on area than by discipline or agency of the respondent. In other words, the fact that a respondent was located in a particular locality was more important than what his professional or agency affiliation was in determining the way he defined a gang, a gang member, or gang incident. The most frequently cited elements of a definition of gang were certain group or organizational characteristics, symbols, and criminal activities, especially violence, drug use and sales. The most frequently used elements of a definition of gang member were symbols or symbolic behavior, self- admission, identification by others, especially the police, and association with gang members. The most frequently cited ways of identifying or reporting a gang incident were the involvement of a gang member, gang function, and modus operandi. A variety of statistical procedures were employed, with limited success, to reduce the complexity and obtain understanding of the underlying factors of the definitions. We were able to account for slightly more of the underlying elements in the definition of gang incident than with the definition of gang or gang member. Respondents in the larger, chronic gang problem cities appeared to be in greater accord on what the elements in such a definition were than smaller, emerging gang problem cities. This somewhat painstaking examination of the way respondents define youth gang, youth gang member, gang incident indicates how complex, elusive, and even contradictory elements of the problem are. It serves as a reminder of the need to temper any particular assessment of the scope and seriousness of the gang problem. The youth gang problem at the present time is apparently many types of perceived gang problems. At the same time, there are certain aspects of the gang problem that are clearly identifiable and interpretable. The role of the criminal justice system, especially law enforcement as the lead identifier of gangs and gang problems is to emphasize serious criminal components. Community-based agencies, including schools and youth service agencies, generally identify a wider range and attribute less serious behaviors to gang problem. On the other hand, a labelling or mislabelling process that identifies minor delinquent or disruptive acts, or criminal behavior by gang members in non-gang structure contexts, as gang activity may occur, which then is conducive to a possible widening of the problem as specifically criminally gang induced. On the other hand, a failure to identify a gang problem in its more serious and special meaning, when in fact one exists, constitutes denial and may be conducive to the growth and development of the gang problem, if nothing or little is done about it. Some balanced consensually observed set of criteria of what is a gang, who is a gang member, and when a criminal act is gang related or motivated is obviously required. The various definitions, especially that of gang incident, should neither exaggerate nor minimize the scope and seriousness of the problem. The findings of the survey, reported in later chapters, also suggest that consensus on definition of the gang incident is an important condition for an effective approach in dealing with the gang problem, especially in chronic gang problem cities. ------------------------------ CHAPTER 3 GANG-CHARACTERISTICS In this chapter we describe and present a preliminary analysis of the scope of the problem based on respondent perceptions of selected gang characteristics. We prepare