The primary unit of analysis in this study, as explained in the first chapter, is a locality for which local authorities report problems with youth gangs. Information on gang-problem localities was collected over a 25-year period. Most of the information was obtained from seven types of sources, four major and three minor. The major sources were youth gang survey reports, media reports, databases, and interviews. The minor sources were conferences, academic literature, and routine police reports.

Different sets of sources were used during three phases of the study. During phase 1 (1974 through 1979), operations were conducted under the auspices of the U.S. Department of Justice's Law Enforcement Assistance Administration and Office of Juvenile Justice and Delinquency Prevention (OJJDP) and the Harvard University Law School. During phase 2 (1980 through 1993), the author operated as an independent investigator. During phase 3 (1994 through 1998), operations were supported by the National Youth Gang Center of Tallahassee, FL.¹

Table A-1 lists the seven types of data sources and indicates which were used during each phase.

Youth gang survey reports, media sources, conferences, academic literature, and routine police reports were used during all three phases. Interviews were conducted during phases 1 and 3, and databases were used during phase 3. The following sections describe the nature and use of each of the seven types of data sources. Detailed descriptions of sources and methods used primarily or exclusively during phase 1 are included in Miller, 1982, appendix E. These will not be repeated here but will be described in abbreviated form where appropriate.

Youth Gang Survey Reports

Tables A-2 and A-3 list 29 printed sources containing lists of names of localities that were designated by the producers of the reports as having problems with youth gangs between 1975 and 1997. Title, sponsoring organizations, and date of issuance are specified for each report. The list is not exhaustive, as the tables themselves reveal. Only those reports that were obtained and examined directly by the author are included. Designations such as "Report 2" and "fifth edition" in some of the titles indicate the existence of earlier reports. The California Investigators Report, for example, indicates that four reports were issued prior to 1993; the Virginia State report of 1996 notes the existence of three previous surveys—January 1992, September 1992, and August 1994. None of these were obtained by the author.²

These reports are divided into two categories, "unrestricted circulation" and "restricted circulation." The first type was made available to the general public with no restrictions. The second contained notices such as "for official use only," "for law enforcement only," and "confidential." These reports were made available to the Institute for Intergovernmental Research with the condition that specific content such as the names of survey respondents, gangs, or gang members would not be made public. The information contained in the present Report conforms with these conditions.

Although the 29 reports listed here do not represent a complete set of all such reports, they probably include most of the youth gang surveys conducted between 1975 and 1997 and thus provide a basis for some summary statements on the yearly frequency of the reports, the identity of the sponsoring agencies, and the terms used to refer to the kinds of groups that were the objects of the surveys.

The earliest of the listed reports was published in 1975. After a 7-year gap, one report was issued each year for 1982 and 1983. After a 6-year gap, 2 reports were issued each year in 1991 and 1992, 4 each year in 1993 and 1994, 10 in 1995, 3 in 1996, and 2 in 1997.

These figures indicate a substantial increase in the number and yearly frequency of the reports, with 1995 as the peak year. The trend suggests that official agencies paid relatively little attention to youth gangs during the 1970's and somewhat more attention in the 1980's, with a major surge of attention in the 1990's. What appears here as a slacking off after 1995 may reflect the fact that post-1995 reports were not yet available at the time of writing.

Examination of the agencies that conducted or supported the reports shows that the majority of reports (14) were conducted, sponsored, or cosponsored by the Federal Government, operating through one or more of its numerous branches and subbranches.

Three agencies—the U.S. Departments of Justice and the Treasury and the Federal Bureau of Investigation (FBI)—provided sponsorship. Within the Department of Justice, reports were produced under the auspices of the National Drug Intelligence Center, the National Institute of Justice (NIJ), and OJJDP. Within the Department of the Treasury, reports were produced by the Bureau of Alcohol, Tobacco and Firearms.

The diversity of investigative agencies reflects in part a struggle by the Federal Government to determine the proper jurisdiction for youth gang crime, after many years of a Federal policy that maintained that gangs were a local and not a Federal responsibility. Federal support of only three reports during the 13 years between 1975 and 1989, following the 1975 report that claimed serious youth gang problems were prevalent in the United States and predicted a substantial increase in these problems, indicates that the Federal Government took a long time both to recognize its legitimate interest in the control of gang crime and to allocate increased Federal resources to its prevention and control.

The first of the 10 regional agency reports in table A-2 (sponsored by the California Gang Investigators Association) appeared in 1993, but its designation as "fifth edition" indicates 1988 as the initial year of a series of yearly reports. The first of the eight listed State government reports, Gangs in Texas Cities, appeared in 1991. It is likely that other States conducted surveys that were not obtained by this study. Three academic institutions, Harvard University, the University of Chicago, and West Virginia University, worked in conjunction with NIJ and OJJDP to produce reports. Finally, 2 of the 29 reports were cosponsored by private consulting firms.

The extended dispute over the proper responsibility for youth gang problems was paralleled by an extended dispute, discussed in the first chapter of the Report, over the proper term for the groups that were the objects of the surveys. The titles of the tabulated reports cast some light on the terms used by the producers of the reports. Of 32 different titles, 26 contain the word "gang." Of these, 11 use the term "gang" with no modifying adjective. Eight use the term "street gang" or "criminal street gang," and seven use the term "youth gang." Of the seven titles using the term "youth gang," four appeared before 1990, whereas none of the titles using "street gang" appeared before that date. This reflects, in part, the increasing participation in report production by law enforcement agencies, since "street gang" has been for many years the term favored by these agencies.

Media Reports

Of the 3,260 gang-problem localities tabulated in the final chapter of this Report, approximately 20 percent were derived either exclusively or in conjunction with other sources from reports in the media. The term "media" here refers to the widest range of documentary materials issued for public consumption, including daily and weekly printed newspapers, magazines, published books, radio accounts, television accounts, and the very wide variety of online documents available through the World Wide Web and other Internet platforms.

Although media sources make up only one of the seven types of data sources used here, it is unusual for a Report of this type to use the media so extensively. The major reason for using media sources relates to the historical nature of the present study. Survey research-based methods operate within a limited span of time; respondents supply information that is available when they are queried. The 25-year timespan covered by the primary analysis in the present Report, in common with historical research, requires retrievable documents over an extended time. For substantial periods of this study, media reports were the only relevant evidence available.

The gang survey reports cited in the previous section represent the most desirable type of source document, in terms of both quality and coverage. However, as shown in the last section, such reports were issued in only 12 years of the 25, leaving 13 years with no survey-based data on gang localities. Media sources were also used in the 1982 gang survey report. Of the 286 estimated gang-problem localities cited in this Report, only 26 were obtained directly through site visits and onsite interviews; the others were obtained primarily from media sources.

Like all data sources, media sources have weaknesses and strengths. Media reporting of youth gang activity varies in quality. Some reports are detailed and accurate, others incomplete and questionable, with all degrees in between. Whether or not events occurring in a particular gang locality will be reported depends to a greater degree on the perceived newsworthiness of the situation than on any desire for comprehensive coverage.

The steady and ongoing nature of media reporting, with continuity of publication assured by market forces (readers and advertisers) rather than by difficult to obtain and relatively infrequent support through grants or other forms of public funding, is one of its strengths. Another very important advantage of media information is the fact that the originating source of information can be identified and verified quite easily.

In contrast to survey research methods where the identity of the original data provider, the respondent, is usually not revealed, media reports identify the original data provider or providers, specify their role or position, and give the date the information was provided. This makes it possible for any interested person to verify the accuracy of the information by retrieving the original account through records or databases maintained by the publishing entity.

Examples of Media Reports

Tables A-4, A-5, and A-6 list the names and positions of persons who provided information on the existence of youth gang problems for 77 of the 3,699 gang-problem localities cited in the final chapter. Each of the three tables covers a different decade—one each for the 1970's, 1980's, and 1990's. The name and position of the information provider and date of information, along with the localities characterized as having gang problems, are listed for 20 information providers for each of the 3 time periods. The 60 tabulated examples do not represent any kind of sample of the universe of gang-problem localities but were chosen to illustrate the variety of positions of the information providers, the range of geographic regions of the localities, and the variety of media sources.

Although the 60 gang locality information providers listed in tables A-4, A-5, and A-6 represent a relatively small percentage of all media information providers, their agency affiliations, reporting localities, and publication sources are fairly representative of the wide variety of agency personnel, localities, and publications in the larger set of media sources. The three tables list 77 different gang localities in 34 States and the District of Columbia. The 60 information providers were affiliated with seven types of agencies—police departments, sheriff's departments, governmental agencies (Federal, State, and city), prosecutor's offices (Federal, State, county, and city), judicial agencies (county, city, and district), public schools, and private service agencies. The media sources include 26 different publications— 19 daily newspapers, 4 magazines or journals, 2 wire services, and 1 online news service.

Sources of Media Reports

The following sections describe five types of media sources: national newsclipping service, online information retrieval, local newspapers, media articles and features, and media-initiated sources. The use of these types varied during the three phases of the study, depending on available resources and developments in computer-based data retrieval.

National newsclipping service. During phase 1 of the study, information on gang localities was received from Burelle's, a national newsclipping service whose staff receive and review all daily, weekly, and Sunday newspapers in the United States and an additional 4,000 specialized publications. The service was directed to clip all stories in which youth gangs or street gangs were mentioned, all stories involving illegal activities by three or more juveniles or youth, and all stories about programs or policies dealing with the prevention or control of youth gang or street gang problems. Clippings arrived weekly for 3.5 years—approximately 150 stories per month, totaling about 3,600 stories, which were filed by locality. Although some stories citing youth gangs during the 3.5 years were undoubtedly missed, examining all the newspapers in the country on a daily basis made it most unlikely that any locality experiencing gang problems would escape attention.

Online information retrieval. The advent and availability of desktop microcomputers with communication capabilities in the late 1970's provided a major new vehicle for obtaining news accounts of youth gang activities and gang-problem localities. The first major widely available online service (called time-sharing at the time) was The Source, a service that pioneered procedures later followed by other online services. Media stories transmitted to the online service could be searched by procedures that located all news accounts containing selected words or phrases. This made it possible to obtain nationwide news reports without using newspapers or clipping services. The Source online service provided two search and retrieve methods. The first, a "menu-driven" system, made it possible to access about 10 national daily newspapers—the Los Angeles Times, Minneapolis Star, New York Times, San Francisco Chronicle, Washington Post, and others. Starting in 1980, electronic editions of these papers were accessed each day and relevant stories selected for hard-copy printouts. Associated Press wire-service stories were processed the same way.

The second system used a keyword search method to select all relevant stories from the total output of United Press International (UPI), which carried about 1,000 to 1,500 stories a day from all parts of the country. When these stories were transmitted to subscribing newspapers, they were also transmitted to online service providers for direct access by any online computer. All UPI stories were searched for keywords. Two or more selector terms were permitted. For example, entering the word "gang" returned all stories containing this term; entering "Chicago and gang" returned all stories with both terms. The term "gang," entered on a daily basis, returned about 5 to 15 gang-related stories per day. Some of these involved adult, motorcycle, terrorist, prison, or other types of gangs in addition to youth gangs. The terms "juvenile," "teenager," "group," "youth," "murder," "robbery," "crime," and "delinquency" were also entered on a regular basis, alone or in combination.

A rapid scan feature made it possible to select abstracted stories for a full-text readout, and an optional print command provided hard-copy printouts. These printouts were filed and analyzed. Computerized news retrieval, while considerably less comprehensive than the national clipping service, made it possible to continue accessing nationwide press coverage of youth gangs on a reduced level after termination of the clipping service.

During subsequent periods, other online services became available. Searches were conducted using three of these: CompuServe, the Dow-Jones News Service, and America Online. The Source was later bought and absorbed by CompuServe, which in turn was bought by America Online. CompuServe, however, maintained independent operations.

These online service providers made it possible to use media sources without subscribing to clipping services. However, the capacity to obtain national-level gang-related media data on youth gangs was enormously enhanced by the advent of the Internet, first accessed in 1994, and the subsequent development and expansion of the World Wide Web, a special feature of the Internet.

The enhanced availability of the Internet coincided roughly with the major upsurge in gang problems in the late 1980's. The Web became a repository not only for electronic media news stories but also for an enormous variety of documents of all kinds, including many relevant to gangs. Using the Alta Vista digital search engine, entering the term "gang" in the late 1990's returned more than 64,000 documents; "street gang" returned 3,600; "youth gang," 1,300; "gang unit," 650; and "juvenile gang," 270. A major advantage of this type of search compared with the earlier electronic press searches was the absence of short-term database purges; Alta Vista retained gang-relevant stories for up to 5 years prior to the search date.

Gang-involved groups, associations, and organizations of many kinds set up and maintained Web sites. Hundreds of police and sheriff's departments developed their own Web pages, which generally included information on whether there was a gang officer, unit, or squad in the department, and some included activity reviews that provided details on gang problems. State and regional law enforcement associations developed Web sites detailing their activities, which often involved youth gangs. Gang task forces at State and local levels also set up Web sites detailing their activities. Many city and town councils published the minutes of council meetings on the Internet, some of which included discussions of youth gang problems and efforts to cope with them.

Towns and villages also developed Web sites; these were particularly valuable for the purpose of locating gang problems because many were too small to receive regular attention from major magazines and dailies such as Newsweek, the New York Times, and the Washington Post. Two of the statewide youth gang survey reports listed in table A-2 were found through Web searches. Even gang members developed their own Web sites and provided forums for discussing gang-related issues. Web searches were of particular value in keeping current with the post-1995 gang localities tabulated in the final chapter.

Local newspapers. During the site visit period of phase 1, local newspapers were used extensively to obtain background information on gang-problem localities and to obtain current information during the course of the visits. In addition to the local newspapers, the Boston Globe and the New York Times were reviewed on a daily basis during all three phases. The clipping service was terminated in 1978, and the first online retrieval service became available in 1980, leaving a 2-year gap. During this period, some limited national coverage was available from periodicals and local newspapers. An "out of town" newsstand in Cambridge, MA, provided hundreds of newspapers and magazines from around the Nation. Publications with stories on gang localities were purchased, clipped, and filed.

Media articles and features. In addition to news stories, a fair number of reports on gang-problem localities appeared in both the print and electronic media during the course of the study. Some of these were quite detailed, taking the form of multipart series prepared by a group of reporters, in some cases reporting from different cities. Features and articles of this kind appeared in Life, the New York Times, Newsday, Newsweek, Time, U.S. News & World Report, and elsewhere.

Reports on gang localities also appeared as programs in the electronic media. This type of program appeared on all the major television networks— generally focusing on gang problems in particular cities. Local radio and television stations also ran programs on local gangs. Like some of the print media articles, some of these programs represented collaborative efforts by numerous reporters, editors, writers, and producers. Notes were made on as many of these programs as possible. In some cases, producers provided the contents of the programs through transcripts or videocassettes.

Media-initiated sources. During phase 1, the author participated periodically in the preparation and production of media pieces on gangs and gang-problem localities. Largely because of extensive publicity accompanying the publication of the 1995 National Youth Gang Survey Summary, writers, reporters, and producers initiated contacts with the author in connection with stories, articles, features, and programs dealing with gangs, and these contacts developed useful information.

Sometimes interest in an upsurge of gang activity or new developments in youth gang crime (e.g., media concern with "wolfpacks" in the late 1970's) provided the impetus for requests by reporters for background information. A frequent by-product of these contacts was information not obtained through other sources. Participation by the author in radio and television programs also provided information. A common format for such programs was a panel discussion featuring a group of informed people. Most of these programs included gang members along with specialists such as police department gang-squad personnel, youth service workers, and legislators.

Databases

Computerized database programs became widely available during phase 3 of the 25-year data collection period. The capabilities of these programs were ideally suited to the task of recording and analyzing gang-problem localities and related information. Many of the reports listed in tables A-2 and A-3 were based on information recorded in computerized databases, but the printed reports resulting from those databases, rather than the databases themselves, provided the gang city information reported here. In some instances, however, the original databases were available to the National Youth Gang Center. These will be discussed following a description of the master database used in the present Report.

National Youth Gang Database

A database configured for recording gang-problem localities and related information was created in 1990, using Ashton Tate's dBASE II database program. The original youth gang record included 14 fields, as follows: location (city, county), city population, State, date of information, time period of information, type of record (report, incident), gang problems reported (definite, probable, possible), number of gangs reported, number of gang members reported, number of homicides reported, race/national background/gender, source of data, and "detail," a text field for recording additional details. This database was exported to an upgrade, dBASE IV, in 1992 and finally to Microsoft's Access database, through several upgrades to version 2.0.

Subsequent versions of the database were modified to fit a variety of purposes, including one designed specifically to generate lists of localities for the mailing lists of the 1995 National Youth Gang Survey (National Youth Gang Center, 1997). Another version focused on longitudinal analysis and expanded the gangs present city population and county population fields to allow separate entries for each of the three decades. Examples of one version of the database record containing 22 fields are displayed in the next section. The total number of fields in later versions of the database was about 35, and the number of records about 9,900. About 4,900 of these were records of cities, towns, and villages, and the rest were records of counties. The cumulative total of gang city records shown in the second chapter is 1,487, about 3,400 records fewer than the total number of city records. This latter group of city records includes two categories. The first consists of about 1,700 records, imported from other databases, of cities that did not report gang problems. The remaining 1,700 records are additional records for the specific gang-problem cities enumerated in table 1. Evidently, many gang cities in the database had more than one record; in fact, some had five or more. The above figures show that the average gang city had slightly more than two records each. For larger cities, the average number of records per city was considerably higher. For example, gang cities with populations of more than 100,000 had about 3.5 records per city.

Each additional record for the same city contained items of information different from those included in the original city record. These multiple records served two major purposes: recording changes over time in numbers of gangs, city size, and other items, and strengthening or weakening the validity of recorded data by using multiple information sources.

For example, Fort Worth, TX, had separate records for 1983, 1984, 1987, 1988, and 1991 to record the number of gangs reported for each of these years—13, 32, 87, 67, and 175, respectively. Multiple sources for the same locality were used for Anaheim, CA, where gang problems were reported for 1975 (Miller, 1975); for 1988 (Spergel et al., 1990); 1989 and 1990 (Fox, 1994); and 1993 (Curry, Ball, and Decker, 1995). Similarly, in Garden Grove, CA, gang problems were reported for 1975 (Miller, 1975); 1981 (M. Davis, Garden Grove Police Department); 1982 (J. Nunez, San Jose Police Department); 1988 (Spergel et al., 1990); 1989 and 1990 (Fox, 1994); and 1993 (Curry, Ball, and Decker, 1995).

Using multiple sources to report the same item of data serves a function similar to that of the triangulation method discussed in the first chapter of this Report. Given the likelihood of differences between respondents, the use of multiple sources representing different agencies and interests increases confidence in the validity of collected information and reduces the risk of obtaining inaccurate information, which can occur when relying on a single source or respondent.

No matter how many records were available for a single locality, each was coded as a new gang locality only once, as defined in the first chapter, and was so tabulated in all relevant tables. Unlike locality counts in some studies (e.g., Miller, 1982; Klein, 1995), the frequency tabulations in the present Report include no estimates; each citation of a gang-problem locality is documented by one or more specific source citations. This feature explains a discrepancy between the figure of 286 given for the number of gang-problem cities in the 1970's in Miller (1982, table 4.3) and the figure of 201 appearing in tables 2 and 3 in the present Report. The difference of 75 between the earlier and later figures results from estimates for the 1970's in the "U.S. Cities Under 100,000" category (Miller, 1982).

Partial data were available for eight States, and estimated undercount percentages were applied to these figures. The States and estimated number of gang-problem cities were as follows: Connecticut, 4; Florida, 3; Illinois, 14; Massachusetts, 24; Michigan, 3; New Jersey, 4; New York, 12; and Pennsylvania, 11. Undercount percentages were based on the assumption that additional numbers of uncounted gang-problem localities would be present in areas adjacent to the major gang cities of the 1970's—New York, Boston, Chicago, Philadelphia, and others. It should be noted here that survey data obtained in the 1990's by the National Youth Gang Center indicated the existence of 270 gang cities in the 1970's—only about 6 percent fewer than the 286 cited by Miller (1982).

To provide illustrations of the onscreen appearance of the gang-problem locality records, several examples of one version of the data record including information from the 1970's, 1980's, and 1990's are illustrated. Table A-7 provides explanations of the field abbreviations.

Incorporated Datasets

Gang locality listings from three databases were imported directly into the dataset used for this Report. The first of these, provided by G. David Curry, was compiled in connection with his 1992 and 1994 national gang surveys conducted under the auspices of NIJ and the University of West Virginia.³ The second dataset, provided by Cheryl L. Maxson, was produced by Maxson and Malcolm Klein in connection with a 1992 national survey of youth gang migration, conducted under the auspices of NIJ and the University of Southern California, Social Science Institute.⁴ The third dataset was based on a subset of the Uniform Crime Reports Supplementary Homicide Reports: 1976-1992, prepared by Eugene Pond of the National Youth Gang Center in 1995. Any locality reporting juvenile or gang-related homicides to the FBI was assumed to have experienced gang problems.⁵ Almost 40 percent of the 2,193 gang-problem localities cited in the second chapter were obtained from these three datasets.

Interviews

Interviews were conducted during phases 1 and 3 of the study to gather information on gangs and gang localities. Most of the interviews took place during the site visit portion of phase 1. Interviews were of two types: face-to-face and telephone. A total of 131 face-to-face interviews were conducted in 26 localities with staff members of 173 different agencies. Many of these were group interviews—for example, all or most staff members of a probation department would take part in an interview. A total of 458 individuals participated in interview sessions. Three kinds of telephone interviews were also conducted during phase 1, including interviews with site-surveyed locality personnel prior to the site visits, with site-surveyed locality personnel subsequent to the visits, and with individuals familiar with the local gang situation in approximately 50 localities that were not visited.

Most of the site-visited localities that did not report gang problems at the time of the visit were asked in a final followup survey whether such problems had emerged at a later time. Phone calls to the 50 localities not visited were made primarily to obtain information on the presence or absence of youth gang problems. All California cities with populations of 100,000 or more and a number of California counties (e.g., Los Angeles, Orange, San Bernardino, and Ventura) were included. These calls sought to confirm or discount media reports of local gang problems. Additional information on the specific details of the site visit interviews is contained in Miller (1982, appendix E).

Most of the phone interviews during phase 3 were also conducted to check on media reports of gang problems; others were made to cities whose size suggested the presence of gang problems, but where no evidence of such problems was available. A series of calls were made to Sergeant Wesley McBride of the Los Angeles Sheriff's Department, primarily to determine the status of California localities that had reported gang problems during phase 1, but for which no evidence of later problems was available.

Conferences

During phases 2 and 3, the author attended several conferences where gang-problem localities were discussed. Names of previously unknown gang-problem localities were obtained either directly from presenters familiar with the localities at issue or from conference participants who knew of the existence of youth gang survey reports. Reports not already on hand were obtained by subsequent requests to the issuing agencies.

Routine Police Reports

Every stage of the criminal justice processing system generates data on offenses and offenders, from initial reports of violations through release from parole. Extensive attrition of offense information occurs as one proceeds from earlier to later stages; information recorded during the final stages of the process includes only a very small proportion of the offenses dealt with during the earliest stages.

One body of information collected during the earliest stages of the process records the enormous number of acts and events reported to or by local police in the course of their daily activities. Recorded incidents have two major sources: citizen complaints (generally received by phone and relayed by radio to patrol officers) and incidents observed directly by police in the course of patrol. Records of these incidents take various forms, including handwritten records by individual policemen, logs kept by the department, and computerized incident listings, often coded by locality, type of incident, and other characteristics. In some communities, police log information regularly appears in local newspapers.

These reports can be of value in identifying localities with youth gang problems. In many communities, a very substantial proportion of all incidents handled by the police involve groups of youth, some of which are designated as gangs. During phase 1, routine police reports for selected periods were obtained for all of the site-surveyed localities and also for about 250 other U.S. localities both during and preceding phase 1. Monitoring local police reports continued during phases 2 and 3.

Gang-Problem Locality Record

5 See Fox, 1994.