Click for next page ( 60


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 59
Social Ecological Risks of Violence Against Women Research on violence against women has focused primarily on indi- vidual characteristics of victims or offenders or on event-level or situational characteristics of assaults and other forms of violence. For many years, it was widely thought that violence against women was "classless"; that is, women were victimized at high rates regardless of their socioeconomic status or the neighborhood in which they lived (Straus et al., 1980~. Findings of research in the 1970s indicated that the spatial dimensions of violence against women were unimportant since intimate-partner violence and rape were both well distributed by race and social class. This neglect of social context in research on violence against women is surprising given the current widespread recognition of the significance of community characteristics and neighborhood effects for violence gen- erally (Fagan, 1993), for the actual socioeconomic distribution of both vic- timization and offending behavior by women (Rennison and Welchans, 2000), and for other related social problems. Indeed, recent research pro- vides evidence for community effects on violence and social problems that may be gender-specific, including child abuse (Coulton et al., 1995; Coulton and Padney, 1992; Zuravin, 1989; Garbarino and Sherman, 1980), teenage pregnancy (Sullivan, 1993, 1989), the prevalence of marital versus cohabiting living arrangements (Tucker and Mitchell-Kernan, 1995; Sullivan, 1993), and the prevalence of female-headed households (Stokes and Chevan, 1996; Tucker and Mitchell-Kernan, 1995~. The remainder of this chapter reviews what research tells us about the importance of social 59

OCR for page 59
60 RESEARCH ON VIOLENCE AGAINST WOMEN context for both intimate-partner violence and sexual assault by strang- ers, and describes research and data needs in this important area of study. A number of recent studies have shown that rates of violence against women vary across such social areas as census tracts and neighborhoods (Miles-roan, 1998~. Moreover, the geographic distribution of violent vic- timization of women overlaps to a large degree with that of male victim- ization (Fagan et al., 2002~. This finding is perhaps easier to understand once we confront the fact that most violence against both men and women is perpetrated by men (Tjaden and Thoennes, 2000~. Some studies have shown that violent victimization of women, and particularly the risk of injury or fatality, is concentrated among poor, non- white populations. Prevalence estimates for stranger-perpetrated violence may be conservative for the neighborhoods in which these women live, and this exposure to violence by strangers also may contribute to factors that characterize violent offending by women (Baskin and Sommers, 1998~. Even intimate-partner violence appears to be susceptible to neigh- borhood effects. Block and Christakos (1995) found that the hardening, or general violence, of the inner city had an impact on female victimization by partners. Other research has shown that the effects of criminal legal sanctions for domestic violence covary with neighborhood context (Benson et al., 2003; Wooldredge and Thistlethwaite, 2002) or that the availability of ser- vices leads to declines in intimate-partner homicide (Dugan et al., 2000; Browne and Williams, 1989~. Moreover, just as individuals change over time in their rates of violent offending or victimization, so, too, do vio- lence rates in cities and neighborhoods (Taylor and Covington, 1988~. For example, Medina (2002a) discusses the idea that neighborhoods have a natural history of violence, with rates increasing and decreasing over time in relation to changes in both social structural factors and other social problems, such as drug or violence epidemics. These simple empirical facts, when framed in the context of theories of the risk of community-level violence, invite closer examination of the social ecological contexts of violence against women. The studies de- scribed above have extended advances in theory and research on commu- nities and crime to the study of violence against women. These studies have applied social structural theories and dynamic theories of social or- ganization and social control to explain variation across neighborhoods or cities in rates of violence against women. Even with this knowledge, however, virtually no research has compared social or neighborhood risks of violence for female and male victims. Nor has research examined the links between the patterns of victimization of women and men in a broader analysis of the spatial ecology of violence.

OCR for page 59
SOCIAL ECOLOGICAL RISKS 6 INTIMATE-PARTNER VIOLENCE In a paper prepared for the workshop, Medina (2002b) observes that attempts to explain concentrations of violence against women using mea- sures of gender inequality point instead to the relevance of other commu- nity characteristics. In 1927, Mowrer suggested that spatial patterns of marital conflict, particularly divorce and desertion, can be explained within a social disorganization framework. Some authors have argued that the mechanisms at play are those emphasized by contemporary ver- sions of social disorganization theory. Pagan (1993) offers the most developed interpretation of the geogra- phy of intimate-partner violence consistent with this framework (see also Williams and Hawkins, 1992, 1989a, 1989b). For Fagan, the relevant di- mensions of community for understanding spouse assault include the level of social control within the community, the social networks within which people and couples are embedded, and the community's social capital. The concentration of social structural deficits in urban areas weak- ens both informal and formal social controls on spouse assault. The significance of social networks for spouse assault is illustrated by researchers' understanding of both the risk factors involved and the im- portance of victims' ability to invoke informal social controls. Social isola- tion is a risk factor for domestic violence (Stets, 1991; Thompson et al., 2000), but the effects of social isolation also increase the risks of partner violence at the community level. Growing residential segregation and iso- lation of residents from the social and economic institutions that repre- sent mainstream society weaken the influence of the larger society on in- terpersonal behaviors, particularly among the minority groups segregated in inner-city areas. The absence of patterns of husband-wife interaction more commonly transmitted by working families embedded in stable kin- ship or friendship networks facilitates the transmission and reification of more-violent norms (Massey and Denton, 1993~. SEXUAL ASSAULT BY STRANGERS Whereas the study of the spatial dynamics of physical violence against women by their intimate partners has been dominated by a focus on the "neighborhood" and structural factors, the study of stranger rape has been dominated by the "place" and opportunity research tradition. Medina (2002b) found an extensive body of research examining the spatial preda- tory patterns of the search for victims in which serial rapists engage. These crimes are more likely to be reported than the much more common nonstranger sexual assaults. Serial rapists, moreover, have a tendency to use the same space repeatedly (LeBeau, 1987), to the point where they

OCR for page 59
62 RESEARCH ON VIOLENCE AGAINST WOMEN influence the definition of those areas with the highest incidence of rape (LeBeau, 1985~. This research documents the distance from the rapist's home to the points where victims are located and raped (Myers et al., 1998; Warren et al., 1998; Kocsis and Irwin, 1997; LeBeau, 1992,1979; Costanzo et al., 1986; Rand, 1984~. Medina (2002b) suggests that by inverting the findings of this research, one can deduce the probable residence of an offender from information about the known crime locations, their geographic connec- tions, and their characteristics. Indeed, police departments in several countries are using computerized geographic profiling in their investiga- tions of serial sexual assaults with an acceptable degree of success (Rossmo,1995~. This intervention may not have applicability, however, to the majority of sexual assaults, which, according to Tjaden and Thoennes (2000), are perpetrated by intimate partners. RESEARCH NEEDS The findings and questions resulting from recent research on commu- nity effects suggest first steps toward integrating theory and research on the social ecological characteristics of violence and on the epidemiology of violence against women. The research agenda that emerges from a con- sideration of this work addresses a series of important scientific ques- tions, as detailed below. Social and Spatial Epidemiology of Violence Research is needed to estimate the extent of variation in violence against women across census tracts or small neighborhoods, police pre- cincts or districts, or other theoretically meaningful social area aggrega- tions. For example, it may be found that rates of violence against women vary at the census tract or neighborhood level, but that these differences are masked at higher units of aggregation, such as planning districts com- prising several tracts. Such research may also reveal that there are differ- ences in area risks for different forms of violence against women. A corol- lary question is whether these differences persist over time, or whether there are variations in the stability of violence rates for different area units or different types of victimization. That is, an important research question is whether life history and ethnographic methods can be applied to neigh- borhoods or other small social areas to identify factors that elevate or at- tenuate rates of violence against women and if so, which forms of vio- lence are thus affected. Research designs should specify which areal units are important for which theoretical perspectives, what data are available at those levels, and whether critical data are lacking. A related question is the interdependence of violence against women

OCR for page 59
SOCIAL ECOLOGICAL RISKS 63 and violence against men in the same social areas. Research should ex- plore this question by estimating the covariance of the two violence rates over time at different units of spatial aggregation. For example, it is im- portant to know whether increases in rates of violence against men el- evate the risks of violence against women over time. To determine whether correspondence in these rates is spurious, it is important to con- struct better ecology models that take gender into account. Social Area Risks of Violence Against Women Research is needed to determine which features of area composition influence rates and types of violence against women. As suggested above, understanding the social structural, social organizational, and social con- trol capacities of neighborhoods is critical to explaining differences in rates of violence against women. A related question is whether changes in these social area features predict changes in violence rates over time. For ex- ample, research should explore whether changes in marriage rates or wages for women predict changes in rates of violence against women. Similarly, research should examine whether an increase in men's unem- ployment rate or educational attainment is associated with an increase in violence against women (Macmillan and Gartner, 1999~. Often research- ers studying neighborhood ecology must deal with aggregation bias that may confound individual effects with effects of neighborhood context and therefore fail to detect true ecological effects. This problem can be mini- mized by using independent measures of individual status (e.g., employ- ment, education) and neighborhood context (Simcha-Fagan and Schwartz, 1986; Sampson et al., 1997~. An important task for this research is to estimate whether there are gender-specific dimensions of a neighborhood that elevate violence risks for women beyond the overall risks of violence in that area. Thus, for example, research should test whether sex ratios, women's participation in labor markets, wage differentials, or other gender-specific social struc- tural features of neighborhoods are associated with particular risks of violence against women. A further question is whether these gender-spe- cific area risks vary for different forms of victimization. Both epidemio- logically and theoretically, research should formulate separate models for violence against women and against men, and then explicitly test their interdependence. Distribution of Services One dimension of social ecology is access to different types of ser- vices. The availability of services (e.g., shelters, counseling) has been linked to variation in rates of intimate-partner homicides against women

OCR for page 59
64 RESEARCH ON VIOLENCE AGAINST WOMEN (Browne and Williams, 1989; Dugan et al., 2000), but only in state- or city- level analyses. Whether access to local services, and at what distance from event locations, can affect localized rates of intimate-partner violence is a critical research question with important implications for the planning and location of preventive services. Research should examine, for ex- ample, the relationship between violence "hot spots" and service loca- tions to assess distances that are barriers to the prevention or deterrence of intimate-partner violence. Research on the locations of other services, including counseling centers and medical services, also should examine this relationship. Research should consider whether access to gender-spe- cific services or general violence prevention services, and at what dis- tance from event locations, may make a difference in the occurrence of victimization. Social Area Effects on Sanctions and Services Social ecological factors may affect not only rates of violence, but also the efficacy of legal sanctions and social interventions. For example, the effects of arrest on the recurrence of intimate-partner violence vary with the social position of both victim and offender (Sherman and Smith, 1992; Maxwell et al., 2001~. But the clustering of persons of similar social posi- tions in neighborhoods or other homogeneous social areas suggests that areas themselves may influence aggregate rates of compliance with legal sanctions (see, for example, Mears et al., 2001~. Additionally, research has shown that dynamic processes of social control affect violence rates in neighborhoods (Dekeseredy et al., 2003; Sampson and Bartusch, 1998; Sampson et al., 1997), as well as rates of intimate-partner violence (Brown- . . ~ ng,1n press. Despite these advances in knowledge, however, research is needed to examine the covariation of individual and social area factors in the re- sponses of both victims and offenders to legal sanctions or social inter- ventions for violence against women. Sanctions may have a deterrent ef- fect in neighborhoods where the social and legal costs of offending are high to perpetrators, but may have the opposite effect in areas where high rates of poverty or social exclusion mitigate both the social costs of of- fending and the punishment costs of sanctions. Research on this topic should examine not only small areas (such as neighborhoods), but also police districts and other areas where law enforcement strategies are implemented and managed. Research is needed as well on area effects on different forms of violence against women.

OCR for page 59
SOCIAL ECOLOGICAL RISKS 65 DATA NEEDS To create a data infrastructure that can be used to test the above re- search questions, modifications will be needed in the sampling strategies employed in survey research and epidemiological studies so that social ecological factors can be incorporated. Stratified sampling designs in sur- vey and epidemiological research should include samples of social areas, as well as of individuals within areas. The selection of social areas should reflect the theoretical question at hand. Studies of the effects of policing, for example, might include both police districts and social neighborhoods as ecological sampling units. The types of data collected on social areas also should reflect theoreti- cal questions. Studies of informal social control, for example, should in- clude survey data from individuals who can report on the social organi- zation and dynamic processes of social control within an area (for an example, see Sampson et al., 1997~. To avoid aggregation biases, these samples should be independent of the samples of individuals from whom epidemiological data on the incidence of violent victimization are col- lected. To obtain accurate estimates of the covarying influences of indi- vidual and area characteristics, the methods of data analysis employed must estimate multilevel effects and avoid the problems of clustering of individual and neighborhood risks (i.e., endogeneity) that often occur in hierarchical studies of violent victimization. The advances in crime mapping methods discussed above can be ex- tended to the study of violent victimization of women. Administrative records from law enforcement and other criminal justice or social service agencies can be modified to include addressable data for events, victims, and offenders. Possible sources of information on victimization of women might also include probability samples on health and victimization and certain public health and vital statistics records (e.g., those who seek treatment, those who are murdered or commit suicide). The integration of public health and criminal justice records, if feasible, might provide more robust estimates of the prevalence of violence against women at various levels of aggregation that would avoid the biases of the respec- tive systems.