Line
Methodology

Survey sample

The representative sample for the 1998 National Youth Gang Survey included 3,018 police and sheriff’s departments in four divisions grouped by area type:

  • All police departments serving cities with populations of 25,000 or more (large cities).

  • A randomly selected representative sample of police departments serving cities with populations between 2,500 and 24,999 (small cities).

  • All suburban county police and sheriff’s departments (suburban counties).

  • A randomly selected representative sample of rural county police and sheriff’s departments (rural counties).
The representative sample for the 1998 National Youth Gang Survey included 3,018 police and sheriff ’s departments in four divisions grouped by area type.

The universe of large cities (1,216) and suburban counties (660) was included in the survey sample for two reasons. First, the 1995 National Youth Gang Survey revealed that gang activity in the United States is most often reported in jurisdictions with large populations. Second, previous research on gangs focused mostly on large population areas. Therefore, including areas with large populations in the survey allowed for comparisons with samples from previous surveys.

The random samples of small cities and rural counties were selected by using a formula developed by Cochran (1977, see appendix C). Implementation of the sampling method produced the following sample sizes: 399 jurisdictions from a total of 8,740 cities with populations between 2,500 and 25,000, identified by the U.S. Department of Commerce, Bureau of the Census, and 743 rural counties from a total of 2,356 included in the Federal Bureau of Investigation’s Crime in the United States, 1994: Uniform Crime Reports (Federal Bureau of Investigation, 1995).

The comparative sample of 1,951 police and sheriff’s departments was composed of jurisdictions that were surveyed in 1995 but were not included in the 1996–98 representative sample. These jurisdictions will not be surveyed in future years.

After followup calls were conducted, the response rate increased to 88 percent for the representative sample and 86 percent for the comparative sample.

Survey instructions specifically asked that sheriff’s departments report only for their “unincorporated service area.” Any contracted jurisdictions were excluded from the reporting. This was done in an effort to avoid sheriff’s departments reporting for cities and towns within their county that were already in the survey sample. During the process of “cleaning” the data, whenever it was determined that the agency might have responded inappropriately (e.g., more gangs than gang members) or had included other jurisdictions in their responses, NYGC contacted the respondent and clarified the responses.

All jurisdictions included in the sample were cross-referenced with a Bureau of the Census database to determine accurate and current populations. Each jurisdiction was assigned a Federal Information Processing Standards (FIPS) Code generated by the U.S. Office of Management and Budget. Each FIPS Code is unique and is linked to the most recent Bureau of the Census population estimates. The 1998 survey used population estimates for 1994 because they were the most current estimates available at the time the sample was developed.

Each city and town was assigned a FIPS Code that corresponded to the entire population of that area.3 Counties were assigned populations for their unincorporated areas. FIPS Code language refers to the unincorporated area of a county as the “balance of” the county and excludes the populations of incorporated cities and towns within the county. A few counties do not have a “balance of” population because there are no cities or towns within the jurisdiction. In such cases, the jurisdiction was assigned the population of the entire county.

Response rate

In March 1999, surveys were mailed to agencies in both the representative and comparative samples. Surveys were addressed to the respondent from the previous year or to the chief of police or the sheriff. Within the first few months, the response rate was approximately 50 percent, with surveys being received either by mail or by fax (a self-addressed, postage-paid envelope and toll-free fax number were provided to each survey recipient). After followup calls were conducted, the response rate increased to 88 percent for the representative sample and 86 percent for the comparative sample. Response rates varied by area type for the representative sample (see figure 1).

Extrapolation/Estimation

To provide the most accurate nationwide perspective of the extent of the gang problem, it was necessary to estimate:

  • The number of jurisdictions with gangs.

  • The number of gangs.

  • The number of gang members.

Figure 1: Survey Sent and Returned, by Area Type, 1998

Extrapolation for nonrespondents in large cities and suburban counties was necessary to produce the most accurate nationwide estimate possible.

To estimate the number of jurisdictions reporting gangs in each area type, the percentage of agencies reporting gangs was multiplied by the total number of jurisdictions included in the group from which the sample was derived.

Estimating the number of gangs and gang members for small cities and rural counties involved the following steps:

  • Calculate the mean number of gangs and gang members for agencies responding either “yes” or “no” to the question regarding the number of gangs and gang members.

  • Multiply the total number of jurisdictions from which the sample was derived by the percentage of agencies responding either “yes” or “no” to the question regarding the number of gangs and gang members.

  • Multiply the product of step 2 by the mean calculated in step 1.

Extrapolation for nonrespondents in large cities and suburban counties was necessary to produce the most accurate nationwide estimate possible. Without extrapolation, the extent of gang activity in these areas would have been systematically underestimated. In addition, any change in the proportion of agencies responding for large cities and suburban counties in future surveys would likely have resulted in a commensurate change in the number of gangs and gang members for these areas, which could have led to a false conclusion that gang activity in these areas had increased or decreased.

To estimate the number of gangs and gang members for large cities and suburban counties, the arithmetic mean number of gangs and gang members per jurisdiction was calculated. These estimates were controlled for population by stratifying responding agencies in both area types into population groups (each based on a subsample of at least 40 agencies) and calculating a mean for each population group. The means by strata were used in lieu of missing data from nonrespondents.

Weighting percentages

Law enforcement agencies continue to be the best available and most widely used source of information for national gang surveys and other forms of criminal justice research.
In several instances, survey questions asked respondents to estimate the percentage of gangs or gang members who were involved in a particular activity or who met certain criteria. Specifically, this pertains to the questions regarding age, gender, race/ethnicity, drug gangs, and the analysis of an alternative definition of gangs. An important limitation to interpreting these types of responses is that these percentages do not reflect differences in the size (membership) of the gangs across the reporting jurisdictions. To account for this important factor, the percentages were weighted by the total number of gangs or gang members (whichever was appropriate) reported in each jurisdiction. Only the responses of agencies that reported an estimate of gangs or gang members (depending on the question) were used in this weighting procedure. This requirement decreased the number of agencies included in the analysis but increased the reliability of the translation of percentages into actual numbers of gangs or gang members.

Demographics

Respondents who reported gangs in 1998 were asked for specific demographic information: age, gender, and race/ethnicity (see appendix A). All responses were in the form of percentages. Demographic categories were defined as follows:

  • Age: younger than 15, 15–17, 18–24, and older than 24.

  • Gender: male and female.

  • Race/ethnicity: African American, Hispanic, Caucasian, Asian, and other.

Gang activity within each demographic category was analyzed by area type and geographic region (as defined by Crime in the United States, 1998: Uniform Crime Reports (Federal Bureau of Investigation, 1999); see appendix E).

Data limitations

Law enforcement agencies continue to be the best available and most widely used source of information for national gang surveys and other forms of criminal justice research. Criminal justice agencies usually are centrally organized and capable of developing systems for routine recordkeeping and reporting (Curry, 1995, 2000; Maxson, Klein, and Cunningham, 1993). However, law enforcement data have some important limitations.

First, many agencies do not collect data in a standardized manner. Databases, automated or otherwise, are becoming more widespread but are more commonly used for gathering information than for recording crime. In addition, the accuracy of responses in most surveys of law enforcement agencies often varies across jurisdictions, because responses are generally based on estimates. For this survey, instructions specifically asked respondents to base answers on records or personal knowledge. The previous (1997) survey asked respondents for their sources of information. The majority of 1997 respondents (53 percent) said they used both official records and estimates, 44 percent indicated they reported only estimates, and only 2 percent indicated they derived their responses solely from official records.

Definitions continue to pose problems for practitioners and researchers evaluating gang activity on a national level.

Second, responses to survey questions likely were influenced by the respondents’ perceptions of gangs in their jurisdiction. Each year, the survey is directed to the most recent previous respondent. If there has never been a response from the agency, the survey is directed to the chief of police or sheriff. Unfortunately, it is difficult to ensure that the same respondent or even the most appropriate official receives and responds to the survey. As a result, a number of different perceptions and opinions may be reflected in the responses of some jurisdictions. Political considerations also may affect responses, and a gang problem may be either denied or exaggerated (Curry, 1995).

Definitions continue to pose problems for practitioners and researchers evaluating gang activity on a national level. Little agreement has been reached on what constitutes a gang, gang member, or gang incident, despite efforts to gain a consensus (Spergel and Bobrowski, 1989). In light of these problems, the current survey did not seek to define gang terms narrowly. The survey defined a youth gang as “a group of youths or young adults in [the respondent’s] jurisdiction that [the respondent] or other responsible persons in [the respondent’s] agency or community are willing to identify or classify as a ‘gang.’” Respondents were asked to exclude motorcycle gangs, hate or ideology groups, prison gangs, and exclusively adult gangs.

In an effort to address definitional issues, the survey questioned respondents about how their agencies defined youth gang crime and what characteristics they considered important in defining a youth gang. The survey also offered respondents an alternative to the definition mentioned in the preceding paragraph and asked what percentage of gangs in their jurisdiction would fall within the alternative definition.

Previous

1998 Youth Gang Survey
OJJDP Summary
November 2000