Using NCVS To Study Youth Victimization

Because violence is a relatively rare event in random samples of the population, the large sample size of NCVS is useful. Equally important, the sample is designed to be nationally representative; households are chosen on the basis of census information rather than published telephone numbers or random dialing procedures, which often produce biased samples. Participation is voluntary, yet more than 90 percent of households surveyed in 1995 agreed to participate in NCVS, making it one of the most representative social surveys in the country. This Bulletin is based on approximately 19,000 interviews with youth ages 12–17.

NCVS data contain family details that are difficult to find in other surveys of youth. For example, NCVS captures information on family income, size, length of residence, and configuration. Because the area-identified NCVS data contain census tract codes for each household or family, numerous indicators about a community (such as neighborhood poverty rates) can be linked to each household and youth in the survey. This wealth of information makes it possible to study how individual, family, and community characteristics are related to violent victimization among youth and whether these patterns are similar for youth living in different types of families.

For this Bulletin, a person is considered to have been the victim of violence if he or she reported at least one incident of attempted or completed assault, robbery, sexual assault, or rape during a 6-month period in 1995.³ Simple assault includes attempted or completed attacks without a weapon; aggravated assault includes attempted or completed attacks with a weapon and completed attacks with serious injury. Robbery includes attempted or completed thefts by force or threat of force. Sexual assault and rape include attempted or completed attacks involving unwanted sexual contact, verbal threats, or forced intercourse.

Most violent crimes involving youth are simple assaults (approximately 72 percent), followed by aggravated assaults (17 percent), robberies (8 percent), and sexual assaults and rapes (3 percent). The majority of these crimes (about 68 percent) are attempted rather than completed incidents of violence (Bureau of Justice Statistics, 2000).

No data can address all of the questions raised by a phenomenon as complex as violent victimization, and NCVS is no exception. Although NCVS has many strengths, its findings have been criticized in the past for underestimating rape and sexual assault, nonstranger assaults, and, more generally, crimes against females (see Bachman and Saltzman, 1995, and Kindermann, Lynch, and Cantor, 1997, for a discussion of these issues). In response to these criticisms, the survey instrument was redesigned, and the new measures have shown estimates of female assaults that are comparable to the estimates provided by the National Violence Against Women Survey (Tjaden and Thoennes, 2000). This Bulletin draws on data from 1995 because that was the year when the redesigned survey was first fully implemented.⁴

Several limitations of the data should be noted. First, although the sample is much more representative than most youth surveys, it does not include youth living in institutional settings, such as a juvenile detention facility, or homeless youth. Second, children younger than 12 years old were not interviewed because of their perceived inability to provide valid and reliable responses to the standard NCVS questions. The victimization of children younger than 12 appears to differ from that of older youth: crimes against young children are much more likely to involve family members, whereas crimes against adolescents are more likely to involve acquaintances (Finkelhor and Ormrod, 2000). Thus, although the data reveal a great deal about stranger and acquaintance violence among the majority of adolescents, it is difficult to determine whether the patterns reported here apply to children younger than age 12 or to crimes such as family violence or child abuse (Office of Juvenile Justice and Delinquency Prevention, 2000).

Measuring Family Characteristics

As described above, NCVS is a householdbased sample that contains important details about individual families (e.g., family income, size, and length of residence in the neighborhood). The overall configuration of each household or family is coded according to 32 different categories. The main elements of these original categories delineate whether a family is headed by a husband and wife or by a single female or male and whether any children, other relatives, or nonrelatives reside in the household. The relationship between each youth and the primary adult in the household is also coded. By combining these three measures, researchers can distinguish families in numerous ways. Preliminary analyses of the statistical validity of the survey’s family categories, however, indicated that youth could be combined into two major family types. Initial analyses also showed that this could be done without masking important distinctions in either victimization risks or individual and community characteristics.

The first family type consists of youth living with two married parents. According to NCVS sample information, nearly 71 percent of youth ages 12–17 live with two married parents, and the vast majority of those youth (97 percent) are the children of the adults. The remaining 3 percent of youth in this group are typically relatives of the married couple. Among all youth living with two married parents, few differences were found between youth living in households with parents and children and youth living in households with parents, children, and others (such as grandparents). Consequently, additional distinctions among families composed of married couples were not made.

The second family type consists of youth living in a single-parent or other family arrangement. According to NCVS sample information, about 25 percent of youth live in single-parent families; most of these youth (91 percent) are the children of the unmarried adult. Approximately 20 percent of children in these single-parent families live with their father, whereas nearly 80 percent live with their mother. It is important to note that no significant differences in victimization were found between youth living with single mothers and those living with single fathers. Therefore, single-mother and single-father families are combined in these analyses.

Also included in this second category are the 4 percent of all youth who live in other family arrangements, including those living with a grandparent, adult brother or sister, aunt, uncle, cousin, or nonrelative. No significant differences in victimization risk were found between youth living in single-parent and these other types of families. Because the relatively small number of youth in this “other” category does not allow for statistically reliable assessments of risk, youth in single-parent and other types of families were combined into one category. Using the categories described above, NCVS data indicate that approximately 71 percent of youth ages 12–17 live with two married parents and 29 percent live in a single-parent or other family arrangement.⁵

Measuring Community Characteristics

Communities are complex places that can be described in numerous ways. This Bulletin presents two strategies to describe them. The first strategy relies on a commonly used summary index known as “community disadvantage” (see Lauritsen, 2001; Sampson, Raudenbush, and Earls, 1997). This index was created by using factor analysis, a statistical technique that helps researchers create summary indexes. The results of factor analysis led to the creation of a single reliable index based on five census tract attributes: (1) the percentage of persons living in poverty, (2) the percentage of female-headed families with children, (3) the percentage of persons unemployed, (4) the percentage of households receiving public assistance, and (5) the percentage of the population that is black. ⁶ By using a single summary index, it is possible to understand how violence is generally associated with area characteristics.

The second research strategy is to study the relative influence of specific aspects of community disadvantage, such as poverty, family composition, and racial and ethnic composition. This strategy is used to determine which aspects of community disadvantage might be most strongly associated with the risk of violence. Both of these strategies were used, and the findings are discussed below.

How Families and Communities Influence Youth Victimization	OJJDP Bulletin • November 2003