THE FACTS ABOUT GANG LIFE IN AMERICA TODAY:
A NATIONAL STUDY OF OVER 4,000 GANG MEMBERS
a special report of the
National Gang Crime Research Center
CO-PRINCIPAL INVESTIGATORS:
George W. Knox, Ph.D.
Jodet Marie Harris, Ed.D.
Thomas F. McCurrie, M.S.
Alice P. Franklin Elder, Ph.D.
Edward D. Tromanhauser, Ph.D.
John A. Laskey, M.S.
James G. Houston, Ph.D.
Avery L. Puckett, B.A.
Shirley R. Holmes, Ph.D.
Sandra S. Stone, Ph.D.
Janice Joseph, Ph.D.
Barbara H. Zaitzow, Ph.D.
Mary S. Jackson, Ph.D.
Curtis J. Robinson, M.A.
Dorothy Papachristos
Robert E. Morris, MD
G.V. Corbiscello
Zheng Wang, Ph.D.
Elizabeth Gail Sharpe, MSW, CSAC
Diane Gunvalson
Thomas McAninch, Ph.D.
Melissa C. Caudle, M.Ed.
John V. Baiamonte, Jr., Ph.D.
Greg Zavala
Laura Rizzardini, M.A.
Dana Caseau, Ph.D.
John F. Kenney, B.A.
Kathleen Aquino
Jeffrey H. Doocy, M.P.A.
Copyright, 1997: National Gang Crime Research Center.
ABOUT THE AUTHORS
Kathleen Aquino has been working with incarcerated youth in Los Angeles for the past six years. She provides Health Education classes and Peer-to-Peer training. The main facility where she works is Los Padrinos Juvenile Hall. Kathleen is currently working towards a Masters degree in Public Health at California State University, Long Beach.
John V. Baiamonte, Jr., Ph.D., is director of the Jefferson Parish (Louisiana) Criminal Justice Coordinating Council, and the author of several books and articles on ethnicity, crime, and corrections. He holds the rank of adjunct associate professor at Southeastern Louisiana University and Loyola University of New Orleans.
Dana Caseau, Ph.D., is an Assistant Professor of special education at California State University, Fresno. She has over 20 years' experience, as a teacher, administrator, and therapist, working with students who have learning, emotional and/or behavioral problems. Her research interests include programming for students at risk, parent involvement, behavior management, social skills training, and inclusion of special needs students in the mainstream.
Melissa C. Caudle, M.Ed., is recognized as a leading expert in school violence, is the author of Crisis Alert System Implementation, and is an international and national speaker on school violence, gangs, and alternative education. She has sixteen years of experience in education as a principal and teacher working directly with delinquents and gangs.
G.V. Corbiscello, is a gang expert who works in the Monmouth County Sheriff's Office, in New Jersey. He has written a number of gang profiles such as the profile of the New York Latin Kings (see: Volume 4, No. 2, 1997, Winter issue of the Journal of Gang Research).
Jeffrey H. Doocy, M.P.A., is a Senior Investigator with the California Department of Corporations. His current research and publications have focused on economic components of criminal organizations, investigative methodology, and complex frauds.
Alice P. Franklin Elder, Ph.D., administers the Office of Correctional Research for the Ohio Department of Youth Services. As a sociologist, she is interested in the study of social organization of gangs. Her current research examines how gang organizations imitate legitimate social forms. She is a reviewing editor for the Journal of Gang Research.
Diane Gunvalson is the Program Director of the Valley Lake Boy's Home in Breckenridge, Minnesota where she has worked since 1979. She is currently working on her Master's at North Dakota State University.
Jodet Marie-Harris, Ed.D., is an assistant professor in Special Education and Rehabilitation Services, School of Education, Rehabilitation Counseling Program, at Jackson State University in Mississippi. She is also a reviewing editor of the Journal of Gang Research.
Shirley R. Holmes, Ph.D., is an associate professor of Education and Special Education at the North Georgia College - Dahlonega, Georgia. Her research interests include bullying, gangs and deviant behavior among school children, gang behavior in rural areas, as well as prevention and intervention programs.
James G. Houston, Ph.D. is an associate professor of criminal justice at Appalachian State University in Boone, North Carolina. He has over 20 years experience in prisons, jails, and community services and is author of Correctional Management: Functions, Skills, and Systems. He has been involved in numerous gang research projects and has completed a book entitled Criminal Justice and the Policy Process for Nelson-Hall, Inc.
Mary S. Jackson, Ph.D., teaches criminal justice at East Carolina University in Greenville, North Carolina.
Janice Joseph, Ph.D., is an Associate Professor in criminal justice at Richard Stockton College of New Jersey. Her research interests include violence against women, youth violence, juvenile delinquency, gangs, and minorities and criminal justice.
John F. Kenney is the Deputy Chief of Security of the Hampden County Correctional Center in Ludlow, Massachusetts. He has worked for the Sheriff's Department for 14 years and rose through the correctional officers rank to his present position. He holds a B.S. in criminal justice from Elmira College (N.Y.).
George W. Knox, Ph.D., is the founder and the executive director of the National Gang Crime Research Center. He authored the first full textbook on gangs (An Introduction to Gangs) now in its third edition and he has over 20 years experience in gang research.
John A. Laskey, M.A., is an instructor in the extension division of the Department of Criminal Justice at Chicago State University. He is active in providing gang training to a number of groups and organizations.
Thomas McAninch, Ph.D., is Professor of Criminal Justice at Scott Community College in Bettendorf, Iowa. He is a member of the Quad City Council of Police Chiefs. He toured China and has written articles on Chinese prisons. He has assisted police departments in designing strategies to control gang activity.
Thomas F. McCurrie, M.A., is the Managing Editor of the Journal of Gang Research which is now in its fourth volume. He has been active in a number of gang research projects over the years as a co-founder of the National Gang Crime Research Center. Prior to his work with the National Gang Crime Research Center he served as the Chief Investigator for the Cook County Public Defender's Office.
Robert E. Morris, M.D., is Associate Professor of Pediatrics at UCLA in Los Angeles and senior physician at the Central Juvenile Hall in Los Angeles, California. His research and publications involve HIV risk behaviors and risk reduction. He is responsible for the pediatric and internal medicine training program at the Los Angeles juvenile halls.
Dorothy Papachristos is a co-founder of Communities Dare to Care, a Chicago program that works the streets of the city and sate working with at-risk youth and their families. Dorothy goes into communities and assists them in organizing themselves to combat gangs, drugs and violence in their communities.
Avery L. Puckett, B.A. has been a Nashville, TN, Police Officer, Sheriff's Investigator, and Polygraph Examiner with over twenty years of investigative experience. At the Sheriff's Department, he developed an innovative jail based gang intelligence program and received a grant for funding. He has instructed college level courses on gangs as adjunct faculty, and also serves as a reviewing editor of the Journal of Gang Research. He is now a gang and polygraph consultant.
Laura Rizzardini, M.A., is a graduate student in Sociology and a research assistant in Criminal Justice at Loyola University in Chicago. In addition, she works as a research assistant at the Cook County Adult Probation Department in Chicago. Her work has included the study of gangs, gender and juvenile delinquency, and probation programs.
Curtis J. Robinson, M.A., is the Resource and Development Coordinator for the Lake County, Indiana Juvenile Court. He is also the gang officer for the court where he has been employed for 26 years. He has also been a member of the gang intelligence unit on the Gary Police Department and has also served as the assistant superintendent of the juvenile detention center in Lake County.
Elizabeth Gail Sharpe, MSW, CSAC, is a clinical instructor in criminal justice at East Carolina University. She has over ten years of experience with the North Carolina Department of Corrections in the areas of probation, parole, and chemical dependency.
Sandra S. Stone, Ph.D., has 20 years' experience in social services as a direct service provider, teacher, researcher and administrator. Her primary areas of interest are criminal justice, family violence, juvenile delinquency and public policy. She is currently an assistant professor in the Department of Sociology/Anthropology, State University of West Georgia, Carrollton, GA.
Edward D. Tromanhauser, Ph.D., is the chair of the Department of Criminal Justice at Chicago State University. He co-authored the book Schools Under Siege (1992) which analyzed the role of gangs and gang violence in today's school environment. He has held positions in a number of national and regional criminal justice and criminology professional organizations.
Zheng Wang, Ph.D., is an Assistant Professor at Texas Southern University. He also teaches about Asian gangs at the Houston Police Academy. Dr. Wang serves as co-chair for the Mayor's Asian Anti-Gang Advisory Board in Houston, Texas. He has an extensive gang research background and publication record. His main interests include Asian gangs, organized crime groups, and quantitative gang analysis.
Barbara H. Zaitzow, Ph.D., is an assistant professor of criminal justice at Appalachian State University, in Boone, NC. Her current work involves assisting local and state officials seeking the increased use of intermediate sanctions as alternatives to imprisonment. Her primary research areas include female criminality, corrections, and alternatives to incarceration.
Greg "Papa Z" Zavala is a state and federal consultant on gang issues. He is the author of the Street Smart Gang Awareness Program, and Gangs in the Hood Awareness Handbook. He has also taught at Yuba College when he retired from the California Youth Authority. He has been investigating gangs for over 27 years. Greg and his late brother have produced an anti-gang video.
ACKNOWLEDGMENTS
In order to survey over 10,000 confined juveniles and adults in 17 states the researchers are indebted mostly to a large number of individuals who work at the facilities used in our national sampling strategy. While 85 facilities were surveyed, at the time of releasing this report we are authorized to acknowledge only some of these personnel, such as: Terry Timm of the Lincoln Hills juvenile correctional facility in Wisconsin; Warden Jim Cooke and the staff of Corrections Corporation of America at the Metro Nashville Detention Center; Mr. Frank Gunter, Operations Director for the North Carolina Division of Prisons and to Dr. Stephen Kiefer, Research Analyst for the North Carolina Department of Corrections and all the superintendents and support staff who facilitated the research and data collection in the State of North Carolina;
We are thankful from the quality of work from the only two paid personnel during the project, two data entry technicians whose work involved data reduction and data verification during the summer of 1996: Mary Chambers and Debra Mitchell. We are most thankful for the large volume of unpaid labor in data entry from Jodet-Marie Harris, Ed.D. as well.
Executive Summary of Findings
Some 29 researchers collected data in 17 states from 85 different correctional facilities (prisons, boot camps, juvenile institutions, etc). This national sample includes N = 10,166 confined offenders, of which 4,140 are self-reported gang members (i.e., they report having joined a gang). We believe it is one of the largest gang research projects ever.
The research was designed to be truly national in scope. It includes data from the north (North Dakota, Minnesota, Wisconsin), from the northeast (Massachusetts, New Jersey), from the eastcoast (North Carolina), from the southeast (Florida), from the south (Georgia, Tennessee, Mississippi, Texas, Louisiana), from the westcoast (3 areas of California), and extensively from the midwest (Ohio, Illinois, Indiana, Iowa).
Here are some of the major differences that emerged in Chapter 3 comparing gang members with non-gang members:
* Gang members were significantly more likely to have the "super predator" personality trait that they always or usually "get what I want even if I have to take it from someone".
* Gang members were significantly more likely to be bullies in school.
* Gang members were less likely to regularly attend church.
* Gang members were significantly less likely to avoid situations involving the risk of personal injury.
* Gang members were significantly more likely to want to demand that their needs be met.
* Gang members were more likely to come from a mother-only household.
* Gang members were more likely to perceive themselves as part of the underclass.
* Gang members were significantly more likely to sell crack cocaine.
* Gang members were significantly less likely to have completed minimal educational credentials (high school degree or GED).
* Gang members were significantly more likely to report having fired a gun at a police officer.
* Gang members were significantly more likely to report it has been easier since the Brady Bill went into effect to acquire illegal guns.
* Gang members were significantly less likely to see the deterrent value of prosecuting juveniles as adults.
* Gang members were significantly less likely to agree with the suppression value of prosecuting gangs as organized crime groups.
* Gang members were less likely to believe in God, and more likely to claim they were on "Satan's side".
* Gang members were significantly more likely to have been involved in organized drug dealing.
* Gang members were significantly more likely to have close friends and associates who were gang members.
* Gang members were significantly more likely to get disciplinary reports while in custody.
* Gang members were significantly more likely to report engaging in fights while in custody.
* Gang members were significantly more likely to report starting a fight or attacking someone while in custody.
* Gang members were significantly more likely to report carrying an improvised weapon (knife, etc) while in custody.
* Gang members were significantly more likely to report having threatened a staff person while in custody.
* Gang members were significantly more likely to attempt to smuggle drugs into the correctional facility.
* Gang members were more likely to report that gangs do seek to influence staff members to bring in drugs/contraband into the correctional facility.
* Gang members were less likely to see the value of a "zero-tolerance" approach in preventing gang recruitment.
* Gang members were significantly more likely to report that a connection exists between prison gangs and juvenile institution gangs.
* Gang members were significantly more likely to have a parent who has served time in prison.
* Gang members were significantly less likely to report having adequate parental supervision as children.
The factual picture that emerged from our analysis of a large national sample of gang members showed these trends about gang life in the United States today:
* Two-thirds (65%) were still active in gang life.
* Many (44.8%) have tried to quit gang life.
* Over half (59.8%) have held some "rank" in their gang.
* Over half (59.3%) have a special language code in their gang.
* Two thirds (67.9%) have written rules in their gang.
* Four-fifths (84.3%) report their gang has adult leaders who have been in the gang for many years.
* Half (58.5%) have committed a crime for financial gain with their gang.
* Half (58.6%) report regular weekly meetings in their gang.
* A fourth (27.9%) report weekly dues are paid in their gang.
* Decisions to join the gang: a fourth (25.3%) indicated that making money was very important; 16.3 percent felt that seeking protection was very important.
* Joining versus volunteering: about half and half, 54.5 percent were recruited, and 44.5 percent volunteered to join the gang.
* The gang gets about a third of the proceeds from drug sales income of its gang members; the estimate therefore being that two-thirds to three-fourths of all illegal drug sale money being disposable income for the "sales force" of the individual gang members.
* Paranoia about the super structure of control: a fourth (28.5%) felt that some outside person/organization or force controls the action of their gang.
* Most (79.6%) would quit gang life given the right circumstances of being given a "second chance in life".
* Four-fifths (82.4%) report that their gang has sold crack cocaine. This should settle the issue about whether gangs or gang members do or do not engage in the illegal sale or distribution of crack cocaine.
* About third (35.2%) have effectively concealed their gang involvement from their parents.
* Most (83.3%) of the gangs have female members.
* About half (45.7%) of the gangs do have female members in a leadership capacity; although, we would not assume this is in a top leadership role, but is probably a more supportive or middle-management role.
* About a fourth (25.8%) indicated the crimes they committed were mostly for the benefit of the gang; about three-fourths (74.2%) indicated most of their crime was committed for their own individual benefit.
* About half (50.1%) claimed their gang does have ties to real forms of organized crime.
* About two-thirds (71.3%) felt that the gang has kept its promises to them.
* A fourth (24.6%) have made false 911 calls to the police emergency telephone number in connection with gang activities.
* Most (70.5%) felt that shooting at a police officer "would be really stupid because of the heat it would bring upon my gang", yet 29.5% felt that shooting at a police office would bring them status and more reputation in their gang.
* Over half (61.0%) reported that the gangs that exist inside correctional institutions are basically the same gangs that exist on the street.
* A third (32.5%) report they have never met the top leader of their gang.
* A new twist on unrealistic expectations: Two-fifths (39.1%) felt they would someday be the top leader of the gang they are in.
* A fifth (21.6%) felt gang membership has affected their religious beliefs.
* A third (36.6%) report the compulsion theory of crime: having been told by their gang to perform an act they knew was wrong.
* Over half (59.7%) indicated they their nicknames were picked for them by their gang friends; most (80.8%) got their nicknames before they were incarcerated.
* Few indicated that their father (9.7%) or their mother (6.9%) actually encouraged them to join a gang.
* Half (55.5%) indicated their parents would be embarrassed to learn of their gang involvement.
* About half (59.9%) agreed or strongly agreed with the idea that if they wanted to they could quit the gang.
* About half (54.0%) agreed or strongly agreed that they feel protected and loved by their gang.
* About half (53%) report knowing active gang members who work in criminal justice (i.e., "moles" for the gang).
* Over a third (40%) have fought with rival gang members while in custody.
* Over a third (37.8%) have used "legal letters" to communicate with fellow gang members.
* About two-fifths (39.1%) have known males in their gang who forced females to have sex.
Chapter 4 will develope Model 1 of the gang/STG classification system for correctional institutions. Using a threat analysis rating system, varying from a low of "zero" to a high of "five", the higher the score on this risk assessment the greater the need for close security. A consistent trend was revealed: throughout the spectrum of the risk assessment scale for gang/STG membership, the higher the threat, the larger the problem for correctional managers. The Model 1 gang/STG classification system appears to have substantial validity, and was tested using a number of critical incident variables (fighting with inmates, attacking or provoking other inmates, threatening staff/correctional officers, carrying improvised weapons, attempting to smuggle drugs into the correctional facility, disciplinary reports, etc).
Chapter 5 provides a detailed analysis of the female gang member. Part 1 of the analysis provided an extensive comparison of male gang members with female gang members. A number of variables proved to be significant different in this comparison of male gang members and female gang members. Part 2 compared gang members and non-gang members within the confined female population. This also produced a number of significant differences. Part 3 analyzed the confined female population using the Model 1 gang/STG classification system. A key finding is that the baseline difference comparing confined males and confined females: the males are more violent generally.
Chapter 6 provides an analysis comparing juvenile and adult gang members. Juveniles were defined as those 17 years of age or under. Adults were defined as those 18 years of age or older. The analysis used only active gang members. Adults, somewhat predicatably, tended to be more pessimistic about their future, more likely to have held rank or leadership position in the gang, to report that their gang has a special language code, to feel protected and loved by being in a gang, and to not view "seeking protection" as a reason for joining the gang.
In chapter 7 we analyze the gang as a social organization and how the variation in organizational sophistication impacts on gang life and gang members. A scale of organizational complexity in gangs was developed and used to compare informal versus formal styles of gang organizations. The analysis revealed that the level of organizational sophistication in a gang impacts on its member's behavior in a number of very significant ways.
Generally, the member of the more formalized gang tends to represent a higher threat in terms of crime, violence, etc than compared to members of less organized gangs.
Chapter 8 provides an analysis of family factors in relationship to gang behavior. An additive index of family dysfunction was created for the purpose of examining "strong" versus "weak" families. The gang member from the more dysfunctional family generally posed a greater threat in terms of violence, was more disruptive once confined in a correctional institution, founding gang life more satisfying, was more likely to be involved in drug crimes, and was more highly committed to gang life. A gang member from a highly dysfunctional family apparently makes for an "ideal" gang member: a more loyal and dedicated soldier for the gang.
Chapter 9 examines the "super predator" syndrome, by providing an analysis of the most violent of the violent offenders operating in the United States today. Only 6.2 percent of the large national sample met the criteria developed for being a super predator, but this 6.2 percent accounted for a large share of problems. The super predator concept appears, like the notion of the "chronic recidivist", to help us account for the higher threat profiles among the gang member population. While small in their percentage of the overall confined population (6.2%) the super predator shows a consistent pattern of behavior: more likely to engage in crime and violence, and less likely to be deterred by the threat of stiff legal sanctions.
Chapter 10 gives the conclusions and recommendations of this report. The primary recommendation is that Congress pass legislation that upgrades the FBI's Uniform Crime Report, effective immediately, for one small enhancement designed to capture two new variables for all persons arrested in the United States: (1) is the arrestee a gang member, and if yes (2) which gang does the arrestee belong to. The justification for this needed change is the need for a more factual understanding of the scope and extent of gang member involvement in crime and violence in the United States today.
TABLE OF CONTENTS
Page(s)
About the Authors................................ ii - vi
Acknowledgements................................. vii
Executive Summary of Findings.................... viii-xiii
Chapter 1: Introduction and Overview............ 1 - 7
Chapter 2: Methodology.......................... 8 - 30
Chapter 3: Descriptive Statistical Results
With a Comparison of Gang Members
and Non-Gang Members................. 31 - 63
Chapter 4: Model 1 of the STG/Gang Classification System For Adult and Juvenile
Correctional Populations............. 64 - 77
Chapter 5: The Female Gang Member............... 78 - 99
Chapter 6: A Comparison of Juvenile and Adult
Gang Members.........................100 - 108
Chapter 7: An Analysis of Gang Organization and
its Impact on Gang Life.............. 109 - 124
Chapter 8: Towards an Understanding of Family
Factors Among Gang Members: A Large
Scale Empirical Assessment...........125 - 140
Chapter 9: The Super Predator: The Most Violent
of the Violent Criminal Offenders
Operating in the United States Today.141 - 155
Chapter 10: Conclusions and Recommendations.....156 - 165
Endnotes......................................... 166
CHAPTER 1: Introduction and Overview
INTRODUCTION
In 1995 a group of researchers brought together by the National Gang Crime Research Center met for the purpose of organizing what may be the largest piece of gang research ever conducted. It was an ambitious project that sought a large national sample of gang members. In discussing how to most effectively bridge the gap between current knowledge needs in the area of "gang problems" and the current state-of-the-art in gang research, the researchers quickly came to a point of consensus that would finally serve as the organizing goal for the project. What was really needed was a large national project that would help to finally clarify the facts about gang life in America today. Out of this came Project GANGFACT.
This chapter explains how Project GANGFACT was organized to provide for both research and service, and how it was possible to carry out such a large scale national research project.
What is Project GANGFACT?
Project GANGFACT is an acronym for Project Gang Field Assessment of Crime Threat. It seeks to clarify the facts about gang life in the United States today. The project was organized in 1995 by the National Gang Crime Research Center. The project includes researchers from a diverse and interdisciplinary spectrum (criminology, health care, special education, criminal justice practitioners, etc). The project was designed to be totally open, where anything could be studied in relationship to gangs. As a result, it is certainly an eclectic approach to hypothesis testing.
What is the National Gang Crime Research Center
The National Gang Crime Research Center (NGCRC) was founded in 1990 and its track record of research productivity has grown considerably since then. The NGCRC track record of organizing, completing, and disseminating useful information from national and local gang research projects has steadily expanded over the last five years. The Center publishes a professional quarterly journal, called the Journal of Gang Research, now in its fourth volume. The Center provides assessment, research, and training services about gangs and security threat/risk groups.
The Center was initially based at a university where several of the key staff also worked as criminal justice faculty. In 1996 the Center became an independent non-for-profit corporation. The mission statement of the Center remains the same: to promote and carry out research to more effectively educate the public about how to reduce the crime and violence from gangs and gang members.
Gang Members as the Unit of Analysis
There are a number of different ways of studying gang problems that can be found in the literature today. Some projects survey police chiefs or social workers, and ask their opinions on the scope and extent of the gang problem. Some projects analyze police records to see what can be learned from arrest reports about gang activity. Some projects study the community's attitude toward gangs and gang members and gang victimization. All of these really do not have the gang member as a unit of analysis where new primary data is being collected.
Project GANGFACT uses the individual gang member as the unit of analysis. The research strategy is the classic "self-report" methodology in survey or questionnaire research. The research strategy, it will be seen, is enhanced with additional attention to validity and reliability issues. The idea is to collect new primary data directly from gang members. Project GANGFACT did not target high schools, it targeted the confined offender population in 17 states.
Large National Sample Goals for Project GANGFACT
Project GANGFACT had the original goal of being able to develop a sample of about 5,000 gang members in 17 states. Due to factors of cost and expense, the full sample goal was not met. However, the large sample size of self-reported gang members that was developed in 17 states makes it, we believe, the single largest study of its kind ever undertaken and yet reported in the literature.
Researchers Involved in Project GANGFACT
We are a group of responsible criminological researchers with a positive track record of both good scholarship and good services to the agencies we work with. We are not being paid for this research, it is a probono effort. In fact, researchers had to pay many expenses out of their own pockets for this research. No one received any supplemental salary for the research reported here. The reason is easy to understand: we did not seek funding for this project, nor did we receive any. Supplies, photoduplication, amenities used to reward respondents, postage, communication, and travel were expenses assumed by the individual researchers alone.
The training backgrounds of the researchers in Project GANGFACT cover a number of social science disciplines as seen in the section "About the Authors".
Most of the researchers are associated with universities. However, some are also practitioners in the field of criminal justice.
The Strains and Struggles of the Researchers
Most research reports do not include such a section, as most research of this nature is often subsidized by a large government grant or by foundation funding. This project did not have the benefit of such funding, nor did the researchers feel they could wait for funding before carrying out the research. For some of the more active members of Project GANGFACT there were, naturally, strains and struggles. One researcher lost a job, to some extent because of the project. Others struggled to cover small and large expenses that had to be paid for out of pocket. However, in spite of enormous obstacles, the researchers were able to carry out their goals of service to Project GANGFACT. To say that the project in all cases went smoothly would be inaccurate. There were tensions, strains, and struggles. To a large degree these problems were worked out through discussion and dialogue and pooling resources from members of the national project.
The Task Force Approach Used in Project GANGFACT
The researchers involved in Project GANGFACT are providing their services, skills, and work in probono fashion: they are not paid, rather they are doing this work as a public service. This approach is therefore a means of carrying out major large scale national research without specific funding for it from government or other sources. This approach also allows for being able to rapidly disseminate findings and results.
This "task force" approach developed by the National Gang Crime Research Center has a long history of previous success. It includes other large scale research projects carried out in recent years such as: Project GANGPINT, Project GANGECON, and Project GANGGUNS.
In the task force approached developed by the NGCRC, each member of the consortium takes individual responsibility for a significant data collection contribution towards the national sampling goal. This is done in their own local areas or in geographical areas not reached by other members of the consortium. All costs are not reimbursed, but are considered local donations to the overall project. The data reduction tasks are handled by the NGCRC. It is the duty of the consortium members to develop strong positive ties to local correctional authorities for purpose of research access. Follow-up after the data collection is essential to maintaining ongoing positive relationships with host sites.
Benefits to Host Sites
Host sites are those criminal justice agencies (jails, programs, juvenile detention centers, prisons, long term juvenile correctional institutions, etc) who allow a Project GANGFACT researcher to collect local data. A host site receives a major service free of charge: a customized local research report based on an analysis of that local data collected; this is a confidential report for the eyes of the site administration. Project GANGFACT provides this local, rapid, useful feedback at no expense and as a public service to the host agency.
Option to Remain A Totally Anonymous Site
A host site can remain completely anonymous: that is, we do not have to identify it in the acknowledgements section of the final report. About a month before the report is to be released, a copy is provided to the host site for review: at which time the host site is asked whether it wants to be acknowledged (names, titles, etc). If we do not hear from the sites after providing the advance copy of the report and asking them if they want to be acknowledged, then we assume they want to remain anonymous, and thus we will not identify the site or staff.
Protection of Human Subjects
The research strategy used by Project GANGFACT is that of an anonymous questionnaire. It poses no harm to human subjects because it is completely anonymous. Thus, no person who completes the survey can ever be identified, in as much as no identifying information (name, etc) is used in the questionnaire.
Still, the research project had to be approved by a number of Institutional Review boards in the states affected. In some states legal opinions had to be obtained that would exempt the researchers from parental notification policies. In California, a detailed court motion had to be filed along these same lines in order to obtain approval for access to incarcerated juveniles. Access to incarcerated adults proved a somewhat easier task.
Process For Collecting Data
The best process we have found that works best in secure settings is to rapidly distribute the surveys as quickly as possible. Thus, if it takes an hour to get all the surveys distributed to the population, at the end of distribution, the local researcher simply returns to the first point of distribution for collection. Thus, we have been able in this process to survey an entire medium sized jail in about an hour. We have been able to survey the largest juvenile detention facility in the United States in two hours using this technique. Thus, there is very little disruption: indeed, it provides the respondents something positive to do.
Where the anonymous survey process works most effectively is where the respondents are also able to be "rewarded" for their assistance. Most sites used in Project GANGFACT therefore made use of such rewards. These rewards varied from providing a $1.00 donation to the inmate's account in an Iowa jail, to providing bags of chips or candy bars to respondents inside juvenile detention centers.
FORMATION OF THE PROJECT GANGFACT TASK FORCE
The Project GANGFACT research task force was formed in 1995 and involved invitations to a number of gang researchers and experts to pledge their support and labor in a unique knowledge development project. It was agreed in advance that this would have to be an unfunded type of research, because in the current overall state-of-the-art gang research the greatest need is for information that will truly help to clarify the facts about gang life in America today. Some of the disagreement in the gang literature today is not theoretical, it is empirical and has to do with what gangs do or do not do, what behaviors or crimes gang members do or do not engage in. It was agreed that to really understand the multifaceted nature of gang life in America today, that multiple social contexts would have to be studied (i.e., boot camps, jails, juvenile facilities, programs, etc). It was agreed that in addition to being a research project, that this would also be a probono service project as well: specifically, that we would provide a detailed "site report" reflecting a complete analysis of all data to each and every site that cooperates with our research efforts. The objective of these "site reports" was to provide useful rapid feedback describing trends and important findings about the site population. The site populations included boot camps, jails, prisons, juvenile correctional facilities, etc.
It was also agreed that to be able to speak effectively to the wide variety of hypotheses that a large task force group might want to test, that a very large national sample would be needed. It was agreed by the Project GANGFACT research task force members that this study would have as its goal the inclusion of at least five thousand (N = 5,000) self-reported gang members in representative geographical areas of the United States (west, north, south, east). The full goal was not met: while over 10,000 confined persons were surveyed, our gang sample is just over 4,000.
In this type of large scale research there would be hard costs that could not be avoided: travel, lodging, photoduplication, honorariums to the respondents, etc. It was agreed by Task Force researchers that any individual expenses in travel and per diem and time would be donated to this project, that is the researcher would basically pay his or her own way to the extent they were able to do so. No one was to be paid any salary for work. Thus, all time devoted to this project was not reimbursed in any sense. Further, the out of pocket expenses spent on the respondents, involving honorariums (cookies, potato chips, etc) would be expenses that the Task Force members themselves would not be reimbursed for.
So facing a situation where there would be a lot of work, and no income, and additionally perhaps substantial out-of-pocket expenses that would not be reimbursed, Project GANGFACT can be said to represent a true alternative paradigm to traditional funded research.
It is difficult to estimate the amount of labor and effort that went into this project, but is was obviously extensive. Of course, the researchers in their capacities as educators were able to use some small resources involving photoduplication from their respective universities. Outside of this limited assistance, no federal, state, or other government or private foundation support existed for this project. It was, in short, a research team designed to accomplish a goal as an end in itself: not for compensation, but because it needed to be done.
SAMPLING GANG MEMBERS IN MULTIPLE SOCIAL CONTEXTS
One of the goals of the Project GANGFACT research task force was to ensure that gang members were studied in a variety of social contexts. The universe became easy to define when we asked the general question: where could we most likely find some gang members today? The first answer was: custody, those in jail, etc. Thus, at an early stage in the research process various social contexts were identified, contacted, and persuaded to cooperate with our research mission.
The types of social contexts used in the Project GANGFACT task force research therefore consists of the following types of facilities:
(1) Jails
(2) Adult prisons
(3) Boot camps
(4) Local juvenile detention centers
(5) Long term juvenile correctional institutions
(6) Private residential facilities and private correctional
facilities, DUI and Work Release centers, etc.
Many of these contexts are sites previously developed by co-principal investigators in previous projects organized and carried out by the National Gang Crime Research Center. Some of these have been sites used by the NGCRC for several years in a variety of research projects. It is not uncommon for sites to be cooperative with NGCRC projects because of the service component that goes hand in hand with NGCRC research projects such as this one, and mostly because of the responsible prior history of NGCRC projects such as this.
SUMMARY
In this chapter we have endeavored to explain why and how Project GANGFACT was carried out. In short, the purpose was to clarify the facts about gang life in the United States today. There are competing explanations and viewpoints on gang life and this project sought to clarify the facts. The factual basis for such an analysis comes from a rigorous methodology that is truly one of the most exceptional undertakings of its kind. Without government or foundation funding of any kind the scholars and researchers involved in this project were still able to interview over 10,000 offenders in 85 different sites in 17 different states, of which N = 4,140 were self-reported gang members. This report provides the preliminary results from this study.
CHAPTER 2: METHODOLOGY
INTRODUCTION
The purpose of this chapter is to explain the research methodology for Project GANGFACT. The chapter outlines the research process from beginning to end. Special attention is paid to issues of validity and reliability.
ITEM DEVELOPMENT AND PRETESTING THE SURVEY INSTRUMENT
All members of the Project GANGFACT research task force at an early stage in the research process developed specific hypotheses they would explore and test. This meant that every researcher developed and submitted specific questions or items to be included in the survey instrument. These questions in the preliminary item pool were then distributed for review, critiqued, revised and then finally tested in a pretest of the finalized survey instrument. With N = 29 different researchers in this national gang research consortium, obviously there were many different types of hypotheses that would be explored in the research, even though the primary theme examined aspects of gang prevention and gang intervention generally.
The pretesting of the instrument was conducted in a high gang density juvenile correctional institution in the midwest. This is a famous site for the Chicago-school of criminological research. The site staff were very skeptical that the youths could in fact complete the survey, but most were able to complete it in a very short period of time. In fact this facility containing nearly 500 youths in short term detention were completely surveyed in a very short period of time: the researchers were able to get in and out of the facility within a two hour period. Several members of the research task force were present for this pretesting, they made observations, and debriefed some of the respondents (i.e., asking them to report questions they did not understand, words they did not understand, phrases they did not understand, etc) and some of the staff. Through this process the survey instrument was further modified to make necessary changes identified from the pretesting. Actually very few changes had to be made. The pretest sample was known in advance to contain gang members. We expected that there would be some respondents who would not know about some of the detailed issues of gang life: any non-gang member would not be privy to the socialization and training afforded by a gang (i.e., learning its language or subcultural argot, its rules, its code, etc).
After the survey instrument was field tested, we felt comfortable as described in the section on validity below that the instrument was capable of measuring what it sought to measure.
SAMPLING OVER 4,000 GANG MEMBERS
The type of research that samples only from one city, or one state, has historically been a source of confounding and confusing research results in the gang research arena. We felt we needed to capture gang members where they can really be found: on the street, and in custody. Our research strategy was one that therefore focused on a variety of social contexts in order to obtain a sample of 5,000 gang members. Mostly, we sought to have representative national data and therefore we focused our research strategy on multiple states, in large and small jurisdictions.
We did not reach our original goal of N = 5,000 gang members. But we came respectably close in generating a gang member sample of over 4,000.
Figure 1 below shows the type of social context by the sample size of gang members from these sites in twenty states. Thus, some 85 different sites in 17 states were used for data collection in the research reported here.
FIGURE 1
TYPE OF SOCIAL CONTEXT BY SIZE OF GANG SAMPLE
Self-Reported Gang Member?
Missing NO YES Total
******* ******* ****** ******
Louisiana Training 4 40 39 83
Institute
Rivarde Training 2 20 25 47
Inst. (Louisiana))
Florida Parish 1 11 13 25
Detention Center
(Louisiana)
Monmouth Co. Jail 5 147 54 206
(New Jersey)
Dade Metro Juvenile 5 110 48 163
Detention Center
(Miami, Florida)
Stockton, California 1 21 23 45
(Juvenile Center)
Indian River Juvenile 13 159 133 305
Correctional Center
(Ohio)
Riverside Juvenile 5 97 66 168
Correctional Center
(Ohio)
DeKalb County Jail 2 28 6 36
(Illinois)
Lincoln Hills Juven. 6 77 199 282
Correctional Center
(Wisconsin)
Site GA1: Juvenile 21 115 96 232
Correctional Inst.
(Georgia)
Texas Boot Camp 2 1 46 49
Figure 1: Continued
Site GA2: Georgia 3 97 54 154
Female Juvenile
Corr'l. Institution.
Site GA3: 9 164 141 314
Juv. Corr. Inst.
GEORGIA
Lake County, Indiana 4 19 42 65
Juvenile Detention
Center
Sciotto State Juv. 18 27 83 128
Corr. Inst. OHIO
Cuyahoga Juv. Corr. 14 157 147 318
Inst. OHIO
TICO, Juv. Corr. 31 132 103 266
Institution, OHIO
Site GA3: Juvenile 7 88 41 136
Correctional Facility
GEORGIA
Foothills Main 7 152 129 288
Corr'l. Facility
NORTH CAROLINA
Audy Home: Chicago 17 124 429 570
Cook County Temporary
Juvenile Det. Center
Scott Co. Jail 5 117 45 167
IOWA
Foothills Camp
NORTH CAROLINA 23 97 43 163
Central Prison 5 86 32 123
NORTH CAROLINA
Juvenile Residential 1 37 39 77
Programs in ND and MN
Parchman Prison 31 257 52 340
Parchman, Mississippi
Figure 1: Continued
Fresno, California 0 19 88 107
Juv. Det. Center
Secure Facility for 4 66 49 119
Convicted Juvenile
Felons in Southeast U.S.
Circleville Juv. 9 89 65 163
Corr. Institution
OHIO
Mohican Village 3 79 85 167
Juv. Corr. Inst.
OHIO
Maumee Juv. Corr'l. 1 72 52 125
Inst. OHIO
Chatauga Corr'l. 0 28 4 32
Center
NORTH CAROLINA
Metro Nashville 22 285 71 378
Detention Facility
(CCA)
Nashville, TENN.
Paint Creek Juv. 2 15 16 33
Corr. Center
OHIO
Cook County Jail 39 195 182 416
Chicago, ILLINOIS
Secure Facility for 6 196 15 217
Convicted Female Adult
Felons in the Southeast
U.S.A.
Secure Facility for 3 130 23 156
Pretrial Adults in
the Southeast U.S.A.
Nashville City Jail 20 233 39 292
CJC (Tennessee)
(Max Security Unit)
McLean Co. Jail 4 96 57 157
ILLINOIS
Figure 1: Continued
State Female Prison 0 14 17 31
(Texas)
S. Texas Co. Jail 0 27 9 36
Medium Security Jail 2 144 29 175
Nashville, Tennessee
Greene Corr'l. Center 3 73 20 96
North Carolina
New Hanover Corr'l. Ctr. 0 24 16 40
NORTH CAROLINA
Sandhills Youth Inst. 0 18 17 35
North Carolina
Harnett Corr'l. Ctr. 4 114 29 147
North Carolina
Wayne Corr'l. Center 1 38 12 51
North Carolina
Franklin Corr'l. Center 4 77 23 104
North Carolina
Sanford Corr'l. Center 1 7 5 13
North Carolina
N.C.C.I.W. (NC25) 3 91 15 109
North Carolina
Piedmont Corr'l Ctr. 5 63 15 83
North Carolina
NC07 Polk Corr'l. Inst. 10 79 48 137
North Carolina
Johnston Corr'l Inst. 4 32 5 41
North Carolina
Durham Corr'l. Ctr. 5 22 10 37
North Carolina
A secure pre-trial 0 15 12 27
juvenile detention
facility located
in the Southeast U.S.A.
Figure 1: Continued
A secure facility for 0 54 66 120
convicted juveniles in
the Southeast U.S.A.
A secure facility for 0 57 63 120
convicted juvenile
felons in the Southeast
U.S.A.
A secure facility for 10 52 53 115
convicted juvenile
felons in the Southeast
U.S.A.
Minimum Security Jail 4 113 17 134
Units in Nashville,
Tennessee
A Secure pre-trial 6 63 25 94
Juv. Det. Center
in the Southeast U.S.A.
Tulare Co. Juv. Det.074552
Center
Tulare Co., California
Caucteret Corr.Ctr. 5 56 17 78
North Carolina
A secure Co. Juv. 3 12 10 25
Det. Center Located
in the Southeast U.S.A.
St. Bernard's Parish0 5 5 10
Juv. Det. Center
Louisiana
Hampden Co. Corr'l. 1 3 5 9
Center, Ludlow, Mass.
NC37, Eastern Corr'l. 0 14 1 15
Inst., North Carolina
Independence Hall, Juv. 2 8 12 22
Facility, Ohio
Freedom Center, OHIO 1 12 4 17
Figure 1: Continued
NC19, Vance Corr'l. 0 3 1 4
Center, N. Carolina
NC20, Umstead Corr'l. 2 6 3 11
Center, N.Carolina
NC21, Warren Corr'l. 1 18 3 22
Inst., N. Carolina
NC16, Person Corr'l. 0 4 4 8
Center, N. Carolina
NC15, Orange Corr'l. 0 0 3 3
Center, N. Carolina
NC27, Black Mountain 1 21 0 22
Corr'l. Ctr. for Women
N. Carolina
NC26, Raleigh Corr'l. 0 14 1 15
Center for Women
N. Carolina
NC28, Wilmington RFW 0 7 1 8
(Halfway House)
N. Carolina
NC35, Martin Corr'l. 0 4 3 7
Center, N. Carolina
NC12, Halifax Corr'l. 0 10 2 12
Center, N. Carolina
NC13, Granville Corr'l. 0 2 0 2
Center, N. Carolina
Co. Jail in Texas 1 31 10 42
(Anon.)
Jail Lock-up, Texas 0 3 19 22
(Anon.)
State Corr'l. Facility 1 41 16 58
(Texas)
Juvenile Halfway House 2 25 13 40
(Texas)
Figure 1: Continued
Central Juvenile Hall 4 107 286 397
Los Angeles, California
Los Padrinas Juv. Hall 0 112 296 408
Downey, California
TOTAL MISSING 441
TOTAL "NO" 5,585
TOTAL "YES" 4,140
TOTAL OVERALL NATIONAL SAMPLE 10,166
--------------------------------------------------------
In all contexts a saturation sampling technique was sought. This meant everyone in the social context was asked to participate in the research. Sometimes incentives were used, and this meant upwards of 90 percent of the populations in these contexts cooperating. Our sample of gang members therefore includes juveniles and adults. But, overall, mostly an offender population sampled while in custody.
INTERNAL CONTROLS ON DATA QUALITY
A number of precautions and safeguards were undertaken during the survey process to ensure the highest possible quality
in the data collected.
1. Covert Observation. During the actual collection of data at some sites there was the opportunity for covert observation. This involved several of the jail or juvenile correctional sites where it was possible to watch the inmates completing the questionnaires on closed circuit television or through observation areas. Thus, in some jail sites it was possible for the researchers to hand out pencils and surveys and then in a control room watch the inmate behavior in their cell areas on the security video monitors. In no case did we see collaboration or any systematic tampering (i.e., one inmate filling out more than one questionnaire). There was no evidence of any collective fraud on the part of inmates in completing the questionnaires. As in other settings, this was presented as a "very personal" survey. Almost all inmates and others surveyed in other sites were remarkably cooperative. In most sites, for example, there was always at least one researcher present at all times inside each classroom while the questionnaires were being filled out.
2. Overt Observation. Overt observation was the rule of thumb in all sites, as one or more of the principal researchers were on hand at all sites to watch and observe the process of data collection. This also afforded the opportunity of introducing another methodological safeguard to evaluate the quality of our data collection. Gender is a specific forced-choice item on the questionnaire, but it was also a variable coded during overt observation immediately after collecting the questionnaires. In all the jails, in the west coast site, and in Chicago sites, we took an additional overt observation precaution during the data collection process. This entailed physically marking all of the physical source documents with a code for gender. Thus, all male and female respondents could then be assessed in terms of attempts at deception with regard to gender. This code assigned by the researchers as an observation taken during a close social contact (i.e., collecting the survey instruments one at a time) was then able to be compared with the respondents forced-choice response. A random response pattern or a fraudulent response pattern could possible be evident in a case where the overt observation of gender did not match with the self-reported gender in the response to the question inside the survey about gender. Thus, lying about gender was seen in only a couple cases and where it was detected the entire survey instrument was not used. Thus, a few cases were eliminated for obvious attempts at deception.
3. Zero Tolerance for Data Entry or Transcription Errors. All survey data stored electronically for purposes of computerized statistical analysis were cross-checked against source documents (i.e., the survey instruments). The data was checked and re-checked and contains no validity threat from transcription errors in the data reduction process. Most of the data entry was performed by one of the Ph.D. researchers or a Ph.D. candidate statistical typist. Some of the data was keyed to disk by highly trained advanced students serving as interns to the National Gang Crime Research Center and their work was thoroughly checked.
4. Few Unusable Survey Instruments Detected. In most of the sites and social contexts used for data collection, a saturation sampling method was used: everyone in the site was asked to complete the questionnaire. Small honorariums were used in some of the sites, and in these sites we would could casually check the surveys to ensure they were fully completed before giving out the honorariums. In very few instances were unusable questionnaires returned. This is far less than one percent and typically involved someone who would check every response to every question, or some similar way of showing non-cooperation. This was a voluntary action to participate in the research, and for the most part a very large majority of the persons at all sites participated and provided high cooperation. The most hostile reaction the gang members had to the survey was the question we had about their expected life expectancy: at what age they expected they would die.
However, we got the distinct impression that most respondents including those in custody were highly motivated to complete the questionnaires, in one sense because this provided an interesting distraction from the boredom of routine regular activities. In only rare instances, then, did we obtain "tainted" survey instruments: those what were obviously fraudulently completed, or not capable of being interpreted, that is for the most part non-cooperative. Thus, no tainted data is included in our analysis because these very few cases where the respondent was less than cooperative their survey instruments were discarded.
5. An Acceptable Level of Trust Was Established.
While our approach was essentially the same with everyone regardless of the social context, in the jail and secure contexts extra efforts were made to provide an adequate introduction and explanation to the respondents. At least one or more of the researchers were typically on hand in the correctional environments studied, where they approached each cell-house area or living area and explained in detail the purpose of the survey research. In the correctional settings, it was not uncommon for joking comments to be heard from the respondents about criminal justice officials, or critical comments towards the criminal justice system generally. Friendly dialogue was common in all social contexts, because the researchers often took time after the survey to answer verbal inquiries, and listen to concerns and issues of the respondents. While the survey asks for no name, and its printed title is "THE 1996 ANONYMOUS NATIONAL YOUTH SURVEY", and while we explained verbally that we did not want their names because this was a very personal type of questionnaire; it was not uncommon in some instances for respondents to still write-in their names and provide other unsolicited information. Several offered to become paid informants, or desired personal interviews such as this respondent from a midwestern site: "I can tell you a lot more about the _______ gang, please ask Officer Smith if you can talk to me in private.". Such messages of "snitches" willing to sell their souls were common in all sites across the nation.
In nearly all research sites or settings, one or more of the co-principal investigators were always present, along with one or more research assistants who were always on hand in each room or area, and to ensure the privacy of the responses the respondents were told that their teacher or program supervisor would never actually see or touch the surveys. Thus, we collected all surveys directly from the respondents in and out of the correctional settings. A large number of students and volunteers assisted Project GANGFACT in 17 different states.
We feel on the basis of the above procedures respectful of the respondents in their social settings regarding the privacy of their responses, and based on our observations of the process of data collection, that a sufficient level of trust was established with respondents to get relatively honest answers.
We have only one caveat that all criminological researchers should be recognizing themselves: offenders have the tendency to over-report their positive attributes and under-report their stigmatizable attributes, and the kind of gang members we are studying are more often than not offenders. Regardless of social context this tendency operates in all areas of research on real offenders, and gang offenders are no different. We also recognize, and this tendency works to our methodological advantage, that offenders are more likely to honestly report the deviance of their friends and associates than their own deviance. Thus, many of the questions or focal areas of our research ask them about "others", i.e., their gang.
6. High Cognition on the Meaning of the Survey Items Implies Clearly We Are Measuring What We Purport To Measure. A large number of respondents, across social contexts but particularly those in custody, had the tendency to write notes and memoranda style comments in the margins of the survey instrument on a variety of issues. These are highly emotive comments implying clear cognition of the true meaning of the survey items or questions. Several examples of this kind of "running" commentary and shared written communication from respondents is helpful to review here to illustrate our assumption regarding this aspect of the strength of our methodology.
Not one survey respondent returned the survey instrument and claimed not to understand the questions. Not one written comment indicated a lack of understanding of the meaning of the question. These were, after all, very concrete questions.
7. Built-in "Lie Tests". In the Minnesota Multiphasic Personality Inventory (MMPI) full form of 566 true or false questions, a "lie scale" exists by being able to compare responses to a question early in the form with a similar question latter in the form. When someone intentionally engages in deceit, they often forget what they lied about before. Thus, it is possible to identify clear inconsistencies in this way.
Similar provisions were adopted for the present research methodology by building in such "lie tests" or tests of inconsistency into the survey instrument questions. Thus, like the MMPI once scored, our present survey instrument once analyzed provides the basis for identifying deceptive response patterns --- those that are clearly inconsistent or suggestive of deceptive responses.
The first test is one where we could capture any respondent who was paranoid enough to lie about present age and age at time of first arrest. One of the questions in the survey instrument asks "At what age were you FIRST arrested for any crime? When I was______years old". Another item in the survey instrument asks "How old are you today? I am _____ years old." A respondent who would engage in early intentional deceit in a response pattern to the survey instrument could therefore be detected by comparing these two items. Deceitful responses would be evident whenever the value of the age for first arrest exceeded the value of present age. A simple computer check allowed for directly testing for this type of systematic deception. In other words, a deceptive respondent might give the age of 17 for current age and give the age of 20 for age at time of first being forced to have sex later in the item order of the survey instrument. The results of this test were no detected deception of this blatant nature.
Several other hypotheses were tested to evaluate the validity of the data. These consisted of matched-pair items that were very similar in nature, that is these questions basically asked the same thing, but with the item phraseology being slightly altered. Most of these matched pair variables measuring the same thing were intermixed throughout the survey. By using contingency table or crosstabulation analysis, the relationship between these paired variables had better be statistically significant by the Chi-square test because one similar variable should significantly different a second similar variable if the respondents are being honest with us. We found these tests very significant with Chi-square values reaching very strong levels. Had these matched-pair lie tests not been significant, then we would have had to conclude that large scale lie behavior threatened the validity of the data.
A related type of validity test is that of internal consistency in terms of logically expected results. If the expected internal logical consistency is above 95 percent, then this could be considered a measure of the rigor of the methodology in a very large national sample. We can illustrate this with Table 1.
Table 1 uses two different questions from the Project GANGFACT survey. The logic here should be obvious regarding these items. Where a sample of N = 8,959 responses were available, what we find here is that only 4.7 percent of the responses are inconsistent with the logical expectation of internal consistency.
The 4.7 percent "inconsistency" is calculated by taking the 262 respondents who indicated they "Believe in God", but are on "Satan's side" (a logical inconsistency), and adding to it the 166 respondents who indicated they "do not believe in God", but are on "God's side" (another logical inconsistency). These N = 428 respondents with inconsistent answers constitute only 4.7 percent of the overall national sample.
The vast majority of the respondents (95.3%) show logical consistency in the pattern of their answers in regard to the test made in Table 1.
Table 1
The Distribution (N) of Believe in God by
Being on the Side of God or Satan
Which best Which Side Are You On?
describes you: God's Side Satan's Side
I BELIEVE IN GOD 8268 262
I DO NOT BELIEVE IN GOD 166 263
Chi-square = 2510.8, p < .001
In Table 1, such a check on internal logical consistency can be made in the overall sample through cross-tabulation of two similar items. The "inconsistency" in Table 1 would be anyone who claims to believe in God but then indicates they are on Satan's side, as well as anyone who claims to not believe in God who then claims to be on God's side. We would not want to argue that generally in the offender population anyone is going to always get the full truth about anything. But the "inconsistency" in Table 1 accounts for only 4.7 percent of the sample. This inconsistency is remarkably low given the large sample size (N = 8959) for this test.
In short, much attention in this research during the instrumentation phase was paid to the matter of structuring a variety of opportunities for the respondent to be deceptive or deceitful in a way that could be easily detected by data analysis. We know it is not customary for researchers when investigating such offender populations as included in the present research to do this, but we did feel that it was necessary to speak to this issue in as much as this was a large scale investigation involving an assortment of known offender groups.
OTHER ISSUES OF VALIDITY
As previously alluded to much attention to detail and many precautions were undertaken during the research process that were designed to enhance validity by protecting against threats to validity. These protective measures used will be described here. We conclude that the validity of the research reported here is higher than average for social research of the type conducted here. We begin by recognizing that generally in social research, and all criminal justice or criminological research, that the term validity is defined as the extent to which the researchers have measured what they purport to measure. Therefore, the ultimate assessment of validity goes directly to the issue of whether or not the survey instrument captures and effectively differentiates the population at-risk to gangs, whether it can effectively identify subgroupings within the gang population (i.,e., specific gangs), and whether the questions about gang life, economic issues, and other related factors or variables really measure what they say they measure. We further note that the ambiguity in language in the survey instrument was reduced during a field or pretest of the survey instrument. We further note that there were few who did not understand the questions in the survey that they were predicted to understand.
Obviously, we did not assume that non-gang members would understand much about the detailed dynamics of the economic infrastructure and financial aspects of gang life. We did, however, predict that gang members would both understand the meaning of such questions and be able to report their experience and beliefs about these specific aspects of gang life. We therefore report that in terms of the construct validity of the survey instrument itself that gang members clearly did understand and had little difficulty in providing responses to the nearly 200 variables in the survey instrument.
The validity issue of the length of the survey is a moot issue we feel. Our survey instrument is long, but generally can be completed in about 30 minutes by most respondents. The structure of the social settings in which the data was collected were such that no "pressure" existed to rapidly complete the questionnaires. The respondents in all social contexts had more time available to them than was needed for the actual completing of the anonymous questionnaire. Normally, about an hour was set aside, and few needed this much time. In some settings, the respondent could take as much time as needed. Thus, by the nature of the precautions taken during the implementation of the research, we rule out any fatigue or "length of survey deterioration" factor as a threat to the validity of this research.
Concurrent of criterion validity means examining a measurement in relationship to some other variable it should be highly predictive of. The most important aspect of the current research was defining who was or who was not a gang member. The way in which validity controls were implemented in the present research design therefore asked different versions of the same question for several variables. This also meant being able to induce much quality control: for example, making sure that someone who in one question reports they have ever joined a gang, and who in another variable indicates the exact name of the gang in an open-ended "fill in the blank" type of question (i.e., "What gang did or do you belong to?_______________), and who then indicates the type of alliance or nation status. Thus, any "Gangster Disciple" would in our sample have to also indicate a membership in the "Folks" nation. We found very little discrepancy between such variables, and therefore believe that our basic measures that differentiate gang members and non-gang members are very accurate. These are also, for the most part, "brand name" gangs: gangs common to the social contexts from which they were sampled (Crips, Bloods, and Sureno sets on the west coast, etc). One of the ways in which we were able to use a "criterion" validity approach was our access to probably the best and most current national directory of gangs in America today --- the National Geographic Guide to Gangs in the United States. This is a large computer file maintained by the National Gang Crime Research Center, it is updated from numerous sources (law enforcement, corrections, etc) every year and has monitored the gang proliferation problem for five years in a row. For a sample listing of this information useful in validating gang names for gang members, see the companion volume to An Introduction to Gangs (3rd and expanded edition, 1995): National Gang Resource Handbook: An Encyclopedic Reference (1995, Bristol, Indiana: Wyndham Hall Press). Thus, official sources providing names of numerous gangs in America were used to cross-check the self-disclosed data from respondents in the present research. As the analysis will reveal, the gang members in the present study are for the most part very well known gangs. Thus, for our most important variable of focus (gang membership), we were able to ascertain the validity of the self-reported gang membership by examining it in relationship to another of other validity control items (name of gang, gang nation alliance, type of rank held in the gang, etc). We did not encounter any cases that were impostors: only a gang member in such a gang would know the type of leadership positions in its unique hierarchy.
The present research can rule out a threat to the internal validity of the design based on history. The reason this is true is that all the data was collected in a short period of time during 1996 (Spring to fall, 1996) covering about a six month period. The hidden benefit of not being a federally funded type of research project is that also there were few if any obstructions to the research process, and that the results could be reported in a relatively short time frame as well. Thus, the findings are very much reflective of the current social reality (i.e., we did not have to wait a couple years to report our findings). For the same reasons, maturation was not a threat to the validity of our research design, because as stated all data collected occurred in a short period of time nationally in all sites, sometimes simultaneously.
The issue of testing as a threat to validity is common to all surveys on known offenders and all self-report surveys. Completing surveys in some of the contexts was a common expectation, particularly among students and youths in juvenile correctional settings. Even the jail inmates had much experience in completing such questionnaires and "surveys". A number of precautions were taken to ensure the validity of the research design by always having researchers on-site during all data collection, typically several members of the larger team were present to assist with data collection. Mostly this involved explaining to the potential respondents that this was a completely anonymous survey, it did ask very personal questions, and that was the reason we did not want anyone's name, and that the "we" consisted of: mostly university professors.
Wherever possible we tried to off-set the disruptiveness of having unknown persons intruding on their social contexts distributing surveys and pencils by offering some type of consumable amenity as a small reward or honorarium for completing the questionnaire. Thus, in most contexts this was viewed as a pleasant distraction. For these and other reasons discussed in the report we do not feel that testing was a major threat to the validity of the present research.
As discussed above as well, the pretesting or field testing of the survey instrument was designed to eliminate any ambiguity in words, phraseology, expressions, and writing. These are forced-choice questions that are not double-barrelled questions, they are pinpointed to specific issues or measurements of background and behavioral experience, beliefs, and attitudes. We posit that the validity of the present study is therefore acceptable for studies of its kind regarding any threat to validity from instrumentation. Another reason that we can rule out instrumentation as a major threat to the validity of our research is that a number of our variables are direct replications of previous research found to be acceptable and reported in the literature --- as is discussed at appropriate points in this report, where such literature is specifically cited.
One of the strengths of the present research regarding validity is how we overcame the potential threat to validity from differential selection of subjects. In gang research, as is common in criminal justice and criminological research on offenders generally, the common situation is to have only one social context to study the human aspect of interest. In the present research, as explained earlier, the plan in advance was to develop and use multiple social contexts for purpose of data collection. Thus, gang members were studied in a large number of social contexts where we could reasonably assume we could find them: in adult correctional settings, in juvenile correctional settings, in community programs, in probation caseloads, work release centers, boot camps, etc. Further, the geographic representation of the sample was intended to be able to examine variations across a large span of the United States: including data collection sites in the west (northern and southern California), in the north (North Dakota, Minnesota, Wisconsin), the midwest (Indiana, Iowa, Illinois, Ohio, etc), the south (Texas, Mississippi, Tennessee, etc), and the east coast (Massachusetts, New Jersey, Florida, North Carolina). A complete list of states is provided in the site cross-reference table (Figure 1). Therefore in a comparative assessment our present research is stronger than most in this regard to taking precautionary measures to ensure that the threat to validity in terms of differential selection of subjects included a broad cross-section (i.e., across different social contexts) and geographical spectrum. Our analysis can therefore speak to issues of comparison not capable in much previous gang research (i.e., comparing west coast Crips/Bloods with midwest People/Folks, etc). The issue of mortality for validity is of minimal concern to the present study and is therefore not a major threat. This was, after all, a "snapshot" survey design: we only sought out a cross-section of data; this was a multi-state multi-context cross-sectional survey research design. It was not intended to be a longitudinal design, with follow-up measurements. The only way in which mortality might therefore negatively impair on the research design would be if our sampling was limited to only one type of social context, or if we only used one site for each type of social context. The fact is we used multiple social contexts and for some of these (i.e., jails, juvenile correctional facilities) we used multiple sites within these social contexts as well. Thus, if someone "missed out" from such a site, chances are this is minimal in terms of not being able to capture the social reality of these social contexts. After all, within the specific social context sites, the research plan called for a "saturation" sampling method: everyone in the jail, the prison, the juvenile center, etc, was asked to complete the questionnaire. Very few persons refused to complete the questionnaire. We tried to structure our data collection in correctional settings where we did not interfere with court calls or visiting, thus we often had to be at the jail late at night on some occasions, and almost always on the weekends, requiring travel and overnight stays in various cities for some of the researchers. Through previous research experience the researchers knew how to structure the data collection process to be as minimally disruptive to the security and other concerns of correctional facilities. We do know that with a long survey instrument such as that used in the present research project that there may be "item mortality", this is not a matter of "attention deficit disorders", it is a matter of simply losing the respondent at some later point in the survey instrument, thus resulting in some cases missing data for those items towards the end of the survey instrument. The present research is, therefore, not unlike other similar survey research designs in having the common problem of some missing data on the many variables measured.
Finally, the issue of regression as a threat to validity is viewed as minimal in the present research and not a major threat. Measurement error was not a major problem, given the fact that among gang members our variables designed to elicit the nature of their economic experiences in and associated with gang life were questions or items or variables that are both replicable and having little if any ambiguity. No cognitive bias exists in this regard to the variables used in the research. We did not simply include extreme cases: such as those highly cooperative youths on a street corner who might suddenly become very interested in a research project when a person of higher social class and social power arrives on the scene to offer aid and assistance --- material and psychic benefits in nature. The fact is our gang analysis covers the complete gang risk continuum as the analysis will reveal.
The gang member respondent was particularly prone to write lengthy and unsolicited comments in the margins of the surveys. This highly affective arousal signaled clearly that the respondents understood all too well the meaning of the questions. Sometimes this running commentary of unsolicited remarks directed feedback to the researchers in various ways, explaining subtle nuances, some of which will be discussed in this report at appropriate points in the presentation of results.
These were questions that gang members clearly understood, enough so that often such members would strike up conversations and seek out attention from the researchers at almost all the sites. The typical gang member respondent was very curious that anyone would ask such specific questions about gang life today. Thus, it was not uncommon for the researchers to stay around the site for additional time spent answering direct questions from the youths, this was particularly true in the juvenile correctional setting.
To recap, in many of the common threats to the internal validity of research such as that reported here (history, maturation, testing, instrumentation, differential selection of subjects, experimental mortality, statistical regression), the precautions taken in the research design, the scope and extent of the research effort (i.e., covering several states, different social contexts, and a large sample), render these threats to be viewed as minimal.
RELIABILITY
The issue of reliability is the matter of the "quality" of the data. The informed reader will recognize that the term reliability in research means basically "do we get the same results with repeated measures". Over time it is possible, indeed probable, that "gang life" and the gang problem itself can, could and might change. For example, the gang problem has expanded and proliferated in recent years in the United States. In another sense, the meaning of reliability in the type of social research conducted here often means "would different researchers going to the same places using the same questions get the same results". We argue, by the nature of the methodological rigor and level of effort in the present research design, that this would in fact be the case to a high degree. In other words, when we issue what are called "site reports" or rapid information summaries back to the host sites that provided access for data collection, that our data truly does reflect the social reality of their environment. Most who have received these site reports agree with us in regard to the critical issues: specifically, gang density, and the scope and extent of the gang member problem in their populations.
This is not to say that the problem may not escalate or deteriorate in the years ahead. That is not the nature of our research purpose to predict the future. Rather our intent is limited to the nature of our research methodology --- a cross-sectional survey design using large samples --- of simply describing the current situation in these various social contexts. Given the rapid feedback, that is little time delay arose between time of data collection and the reporting of findings, we also argue that our research has high reliability in terms of the volatility of such data: ours was recently collected, and quickly reported. Our generalizations are to the present, not the future --- as we recognize the gang problem is a dynamic and not a static problem.
But the trained researcher will also recognize that the methodological matter of reliability is really the simple and testable issue of whether the same measurement techniques used in different research settings or at different points in time produce the same results. We can give and test an example of this aspect of reliability. Different questions at different points in the item order of the survey produced almost identical results. These are discussed in the chapter describing the descriptive findings on our gang member sample.