Rural Isolation,Small Towns,and the Risk of Intimate Partner Violence

Abstract

Ethnographic research from the United States on gender-based violence showing that rural isolation exacerbates intimate partner violence (IPV) is at odds with estimates from nationally representative victimization surveys which indicate that the incidence of IPV in settlements conventionally characterized as rural is similar to or less than the incidence for urban settlements. One possible reason for this discrepancy—that the conventional metropolitan statistical area–based measure of settlement type fails to distinguish isolated rural areas from other nonmetropolitan places—is put to test in this study. Pooled data from 578,471 women interviewed a total of 1,672,999 times in the National Crime Victimization Survey (NCVS) between 1994 and 2015 were used in this study to consider the risk of IPV across a measure of settlement type that differentiates nonmetropolitan settlements into dispersed rural areas or residentially concentrated small towns. Logistic regression estimates of semiannual IPV prevalence were modeled using generalized estimating equations and robust standard errors to compensate for repeated measures and for the complex sample design of the NCVS. After adjusting for age, race/ethnicity, year, and time in sample, these analyses indicated that women from dispersed rural settlements had a lower semiannual risk of IPV (2.31 per 1,000 [95% confidence interval [CI] = [2.02, 2.64]]) than women from small towns (3.30 per 1,000 women [95% CI = [2.82, 3.87]]) or women from the urban core (2.60 per 1,000 [95% CI = [2.44, 2.77]]). Contrary to the ethnographic record, the results of this study indicate that women living in rural isolation are at a lower risk of IPV victimization relative to other American women and that women from small towns—the urbanized portions of nonmetropolitan counties—have been most at risk of suffering physical violence committed by an intimate partner.

Keywords

intimate partner violence prevalence NCVS settlement type

Qualitative research points to geographic isolation as a key determinant of intimate partner violence (IPV) against women residing in rural communities in the United States. Be it up an Appalachian holler (Websdale, 1998) or far out in an Alaska Native village (Shepherd, 2001), it is thought that rural isolation puts women’s lives in peril because of the impunity afforded their abusers: where neighbors are distant (i.e., where capable guardians are absent), abusers can act without fear of legal repercussion. While the ethnographic record is compelling, a connection between rural geographic isolation and an increased risk of IPV has not been established in victimization survey estimates. Furthermore, the risk of IPV against rural women has yet to be considered in light of a broader research stream on the socioeconomic blurring of the line between the urban core and the most populous rural settlements (Burton et al., 2013; Eason, 2012). At issue in this study are variations in risk of IPV across settlement types—where settlements can best be thought of as the places within which cultures are rooted (Jordan, 1966) and everyday human interactions take place (Camm & Irwin, 1979). In this article, I will consider data from the National Crime Victimization Survey (NCVS) to estimate variations in the risk of IPV across settlement types using a measure that distinguishes isolated rural settlements from the more populous nonmetropolitan settlements that are conventionally coded as rural despite their relatively high residential densities.

IPV in Rural and Nonmetropolitan Settlements

The results of research on rural/urban differences in IPV rates in the United States are largely equivocal. Of the eight different comparisons of the incidence of IPV by settlement type using nationally representative samples, only three have found that rates were higher in urban settlements than in those that are rural (Greenfeld et al., 1998; Rennison et al., 2013; Truman & Morgan, 2014); otherwise, differences in IPV incidence have not been statistically significant (Bachman, 1994; Bachman & Saltzman, 1995; DeKeseredy et al., 2012; Rennison & Welchans, 2000; Wolf Harlow, 1991). This inconsistency is unlike what is found more generally for the United States where urban areas are more violent than rural areas (Truman & Morgan, 2016; Weisheit & Wells, 2005).

Ethnographic research on gender-based violence in rural settlements provides some insights into why the risk of IPV against rural women may be at a level similar to that for urban women. Of these lessons, the most salient deal with the effects of geographic isolation on IPV which is seen as exacerbating the risks of IPV because it allows batterers to act with impunity (Websdale, 1998), because it serves to immobilize victims (Gagné, 1992), and because it renders pertinent services inaccessible (DeKeseredy & Schwartz, 2009; Shepherd, 2001). In other words, geographic isolation is thought to result in rural women leading “spatially restricted lives” (Warrington, 2001, p. 365) that remove them from neighbors’ informal guardianship, that interfere with their separation, and that frustrate efforts to receive assistance. At its core, this connection between rural isolation and increased risk of IPV is rooted in routine activities theory (Felson & Cohen, 1980). Isolation reduces capable guardianship and increases the convergence in time and space of a victim and her motivated abuser.

Problematic Conflation of Nonmetropolitan and Rural

As with most considerations of rural/urban differences based on public use data, available measures in the NCVS are less than optimal indicators of settlement type. The conventional measure of settlement type¹ does a poor job of distinguishing rural settlements from that are nonrural, which, as a result, makes it difficult to identify those rural settlements that are truly isolated. The measure in question, named V2129 and labeled “MSA status” in the NCVS public use data published by the Bureau of Justice Statistics (BJS, 2016), is based upon the components used by the Office of Management and Budget (OMB) to classify counties as metropolitan statistical areas (MSAs) in which the principal cities of MSAs are “urban,” areas outside principal cities within MSA counties are “suburban,” and nonmetropolitan counties are “rural” (Duhart, 2000).

The convention of conflating nonmetropolitan with rural is an unintended extension of the process of MSA delineation that results in a measure of settlement type with limited face validity. As OMB administrators have pointed out on numerous occasions (e.g., Arbuckle, 1998; Mulvaney, 2018), the standards for delineating MSAs “do not produce an urban-rural classification” (Sunstein, 2010, p. 37246) and are instead intended solely for the purpose of identifying metropolitan counties. Going beyond the OMB’s intended purpose results in a measure that categorizes places in ways that do not correspond conceptually. For instance, many settlements are coded by the MSA status variable as suburban even though they are culturally rural (e.g., Amish communities outside of Lancaster, Pennsylvania or the farms of the San Joaquin Valley) or ecologically desolate (e.g., the Mojave Desert of California’s Imperial and San Bernardino Counties; DuBois et al., 2019). Similarly, the conventional coding of nonmetropolitan counties as rural ignores a sizable proportion (42%) of their populations living on residential blocks classified by the U.S. Census Bureau (2012a, 2012b) as urban. As a result, for many women from nonmetropolitan areas that are conventionally coded as rural by scholars of the NCVS, the settlements within which they reside should not be characterized as isolated.

The conflation of rural with nonmetropolitan by the conventional NCVS measure is problematic not only because it fails to distinguish isolated settlements but also because it does not differentiate the most populous nonmetropolitan settlements that have begun to resemble the urban core in other ways that would be expected to increase the risk of IPV (Schneider et al., 2016). Over the past few decades, the pains of industrial restructuring associated with globalization and the various financial crises have hit hardest in those nominally “rural” settlements. This has resulted in a concentrating of poverty in the most populous nonmetropolitan settlements at a rate that is greater than that of other types of settlements (Thiede et al., 2018). Two larger developments stand out in this regard. First, there was the growth of concentrated disadvantage in the most populous nonmetropolitan settlements (Foulkes & Schafft, 2010) resulting from a proliferation of mobile home parks (Salamon & MacTavish, 2017) and from the in-migration of underprivileged urbanites seeking affordable housing (Lawson Clark, 2012). The second development is the effects of the 2007–2009 “Great Recession” on unemployment that exacerbated a preexisting decline in the nonmetropolitan manufacturing base leaving the most populous nonmetropolitan settlements with the steepest decline in employment and the lengthiest recovery. The pains of unemployment were most acutely felt in nonmetropolitan counties with urban areas of 10,000 or more residents than in metropolitan counties or in other nonmetropolitan counties (Hertz et al., 2014). In terms of levels of concentrated disadvantage, the blurry line distinguishing rural and urban for these most populous nonmetropolitan settlements relative to the urban core is the size of the population so afflicted (Eason, 2012).

A pair of theoretical approaches has been used to connect concentrated disadvantage to the risk of IPV. First, social disorganization theory links an elevated incidence of violence and disorder to economic hardship through community breakdown brought about by losses of collective efficacy and social capital and increases in neighborhood turmoil in the form of high levels of residential mobility, family disruption, population density, and ethnic heterogeneity (Bruinsma et al., 2013). The second theoretical approach views IPV as a reaction to the challenges to men’s masculinity brought about by economic hardship and industrial restructuring (DeKeseredy et al., 2007), whereby unemployment is a threat to precarious manhood (Michniewicz et al., 2014) that, in turn, increases the likelihood of IPV perpetration (Reidy et al., 2014). This “crisis of masculinity” is seen to be especially deep in rural areas where the social and economic transformations of the past three decades have formed the most substantial threat to masculine identity (Carrington & Scott, 2008).

Limits of Incidence Studies

The other primary limitation of earlier comparisons of IPV rates across settlement types is a lack of adjustments for variations in demographic structures associated with the risk of IPV. Two characteristics in particular, age and race, vary across settlement types and are associated with IPV. Generally speaking, residents of nonmetropolitan areas are older and have less racial and ethnic diversity. There is an inverse association between urbanization and age. On average, according to the 2010 U.S. Census, adult women and older living in settlements conventionally defined as rural were more than 4 years older (M = 49.6 years) than adult women living in urban settlements (M = 45.3 years; U.S. Census Bureau, 2012a, 2012c). There is less racial and ethnic diversity as one moves from urban to rural. According to the 2010 U.S. Census, non-Hispanic Whites made up five out of six (82.7%), nearly three fourths (71.7%), and roughly half (51.7%) of adult women and older residing in settlements conventionally defined as rural, suburban, and urban, respectively (U.S. Census Bureau, 2012a, 2012c, 2012d).

IPV is more likely to be committed against women who are younger and/or who belong to a racial or ethnic minority. There is a negative association between age and the likelihood of IPV victimization. Over a lifetime, the risk of physical IPV victimization peaks when women are in their early 20s (Rivara et al., 2009). In the short term, the incidence (Rennison, 2001; Rennison & Rand, 2003; Truman & Morgan, 2014) and prevalence (Breiding et al., 2008) of IPV victimization are greatest for women in their late teens to early 20s. Women from some racial/ethnic minorities are more likely to experience IPV (Capaldi et al., 2012). American Indian/Alaska Native women (Rosay, 2016), Black women, and multiracial women (Tjaden & Thoennes, 2000) in particular have a greater than average risk of IPV victimization. However, for Hispanic (Kantor et al., 1994) and Asian women (Breiding et al., 2014), the risk of IPV is similar to or less than that of non-Hispanic White women.

The lack of consideration of age and race/ethnicity as confounders in comparisons of IPV in rural and urban settlements is largely a function of the limited statistical power of comparisons made using incidence rates.² Even though the NCVS surveys tens-of-thousands of respondents each year, and despite the fact that IPV is much too pervasive, the relatively low base rate of IPV over a 6-month reference period prevents the specification of zero-order associations with potential confounders. For example, Rennison and Rand’s (2003) attempted disaggregation of IPV incidence rates by age group (12–24 years, 25–54 years, and 55 or more years) and race/ethnicity fell short when there weren’t enough cases in the oldest age group for reliable estimates for Black or Hispanic women. They had similar difficulties for IPV incidence rates by age group across settlement types when there were too few victimizations of rural women age 55 and up for them to have been included in the estimate comparisons (Rennison & Rand, 2003).

The research presented in this article is designed to address both of the above shortcomings. To sort out the effects on IPV risk of rural isolation relative to that of residence in a more populous nonmetropolitan settlement, I considered a more fully specified indicator of settlement type that separates the most dispersed nonmetropolitan areas from those that have a greater degree of residential concentration. To consider the effects of potential confounders upon the association between rural isolation and IPV within the limitations of a low base rate phenomena, I used multivariate logistic regression models to estimate the net effect of settlement type on women’s risk of IPV victimization. Given the effects of race/ethnicity and the effects of age upon IPV, one would expect that the prevalence of IPV would be attenuated in nonmetropolitan settlements due to their older, less-diverse populations.

Method

Data and Measures

The data used in this study are from n = 578,471 females aged 18 and older interviewed a total of n = 1,672,999 times in the NCVS over the period 1994 to 2015 (BJS, 2016). Following a rotating panel design, the NCVS is administered in person and by telephone every 6 months to a nationally representative household sample by interviewers from the U.S. Census Bureau. In 2015, roughly 124,000 different respondents residing in nearly 85,000 households were interviewed a total of almost 190,000 times. As with the earliest victimization surveys (Biderman, 1967), a primary purpose of the NCVS is to obtain accurate estimates of the amount and kinds of crime committed nationwide (BJS, 2016). In addition to behaviorally specific screening questions about a set of common violent and property crimes, the NCVS asks follow-up questions regarding the nature of each incident reported—including harms experienced because of the incident, incident specifics such as time of day and type of location, and pertinent characteristics of individuals involved with the crime.

For this study, the most important NCVS questions are those regarding (a) violent offenses and (b) the relationships between victims and perpetrators. Each NCVS respondent was coded as having experienced IPV if she reported being the victim of a violent crime (assault, rape, sexual assault, or robbery) that was committed by an intimate partner (a current or former spouse, boyfriend, or girlfriend). Given that the power imbalances and inequalities often underlying IPV in opposite-sex couples are no less salient in same-sex couples (Eaton et al., 2008), incidents of IPV committed by same-sex partners were included in the analyses conducted for this study. Overall, 1.8% (95% confidence interval [CI] = [1.4%, 2.3%]) of adult female IPV victims reported at least one incident of violence committed by a same-sex intimate partner.

Breaking with BJS practice of including all females aged 12 and older in NCVS estimates of IPV (e.g., Bachman, 1994; Rennison, 2001; Truman & Morgan, 2014), in this study I have limited the population of interest to females aged 18 and above; most females below the age of 18 are not at risk of IPV because they are unlikely to have intimate partners. A large majority of youth below age 18 (Taylor et al., 2017), especially those age 14 and younger (Lenhart et al., 2015), are not involved in dating or intimate relationships. By limiting the analyses to adult females, this study follows the lead of the National Violence Against Women Survey (NVAWS; Tjaden & Thoennes, 2000), and the National Intimate Partner and Sexual Violence Survey (NISVS; Breiding et al., 2014).

The NCVS also includes information on household geography that was used to categorize respondents by settlement type. Two geographic categorical variables in particular—the MSA status variable considered above and a second measure named V2017 and referred to as “land use” (BJS, 2016)—were combined to separate geographically isolated nonmetropolitan settlements from those that are much more urbanized. While not referred to as such, the NCVS “land-use” variable is actually the U.S. Census Bureau’s dichotomous residential block-level urban/rural indicator. For the Census Bureau, a settlement with a minimum of 2,500 residents is categorized as “urban” when it has a core of blocks with population densities of at least 1,000 residents per square mile and peripheral blocks with minimum densities of 500 residents per square mile (Ratcliffe et al., 2016). Any residential block not meeting one of those criteria is categorized as “rural.” The Census Bureau’s urban/rural measure provides a much better indication of residential concentration than the MSA status variable. Conceptually, this measure gives us an “airship view” of “conurbation” (Fawcett, 1922, p. 111) from which sparsely populated areas are easily distinguished from those with a more densely concentrated population.

Combining the two geographic measures results in five possible settlement types for which the risk of IPV was considered in this study. This combined settlement type measure retains the central city of the MSA status variable as the urban core while dividing its suburbs and nonmetropolitan areas according to Census Bureau categories of urban or rural residential concentrations. Metropolitan areas outside the central city are divided into suburbs and exurbs depending on if they have urban or rural residential concentrations, respectively. Similarly, areas coded as nonmetropolitan by the MSA status variable are categorized as either small town if they have an urban residential concentration or as dispersed rural (Demangeon, 1927) if their residential concentration is rural.

This separation of the higher density nonmetropolitan settlements (i.e., small towns) from those with less concentrated populations (i.e., dispersed rural settlements) makes it possible to sort out the effects on IPV risk of residence in a more populous nonmetropolitan settlement versus residence in an isolated rural settlement. Basic demographic comparisons of the settlement types as defined conventionally and for the combined measure are shown in Table 1. The division of the nonmetropolitan MSA status according to residential concentration clearly distinguishes dispersed rural settlements from those that are more urbanized. In terms of per-person population densities, the nonmetropolitan areas with 18.32 acres/person are divided into (a) less dense dispersed rural settlements with 32 acres/person and (b) more densely populated small towns with 0.45 acres/person. Comparatively speaking, roughly a third of the nonmetropolitan population resides in small town settlements that are, on average, 41 times denser than nonmetropolitan settlements more generally. Breaking down the noncentral city settlements of metropolitan areas (at 3 acres/person) by residential concentration results in a similar distinction between sparsely populated exurbs (14 acres/person) and more heavily populated suburbs (0.3 acres/person).

Table 1.

Demographic Characteristics by Settlement Type.

Data Source:	NCVS, Women aged 18+, n = 1,672,999			2010 Census
SettlementType Measure	% of Total Population	% White Non-Hispanic	Mean Years of Age	Trimmed Population Density (Persons/Sq. Mi.)^a
MSA status (V2129)
Central city	31.2	54.8	44.6	2,801
Suburb	50.9	75.8	46.9	218
Nonmetropolitan	17.9	84.7	48.6	35
Residential concentration (V2017)
Urban	77.6	65.7	45.9	2,350
Rural	22.4	88.5	48.3	28
Combined settlement type (V2129 × V2017)
Urban core	31.2	54.8	44.6	2,801
Suburb	40.0	72.2	46.7	1,974
Exurb	10.9	88.7	47.4	45
Small town	6.5	78.3	47.6	1,447
Dispersed rural	11.4	88.3	49.2	20

Source. U.S. Census Bureau (2012a).

Note. NCVS = National Crime Victimization Survey; MSA = metropolitan statistical area.

Excludes census tract segments with zero residents or that were larger than the state of Rhode Island (1,212 sq. mi.).

Data Analysis

In accordance with instructions for calculating standard errors using direct variance estimation methods (Shook-Sa et al., 2015), estimation of relative risk and prevalence of IPV across the five settlement types required creation of a person-level file combining information pooled for the years 1994 to 2015 from the NCVS incidents, persons, and households files. Starting with the NCVS incidents file, each incident was categorized as either involving a violent crime (i.e., assault, rape, sexual assault, or robbery) committed by an intimate partner (i.e., spouse, ex-spouse, or current or former boyfriend/girlfriend) or involving some other combination of offense and victim/offender relationship. Next, the total number of IPV incidents was aggregated for each semiannual interview, then collapsed into a dichotomous indicator of having been the victim of IPV during the semiannual interview period, and finally combined with the NCVS person-level file matching on the unique person identifier and year-quarter of survey administration variable. The Census Bureau’s urban/rural indicator was transferred from the NCVS household-level file onto the person-level file, matching on the unique household identifier and year-quarter of survey administration variable.

The particular challenges of using pooled data from the NCVS—both its complex stratified sample design with analysis weights and its clustering of individual respondents across multiple interviews—preclude the use of standard methods of statistical analysis. Given potential underestimation of the variance and an increased likelihood of Type I error, the estimates of IPV victimization risk for this study were instead calculated in SUDAAN (Research Triangle Institute [RTI], 2012) using generalized estimating equations (GEEs) with an exchangeable working correlation structure and robust standard errors to fit multivariate logistic regression models. SUDAAN is unique in dealing with multiple levels of clustering, thereby allowing for simultaneous consideration of correlated data within primary sampling units and across repeated measures (Driezen, 2016; Horton & Lipsitz, 1999). All analyses were conducted using person weights to adjust for nonresponse and the probability of selection.

The adjusted prevalence of IPV victimization (i.e., the adjusted probability of being a victim) was calculated using SUDAAN’s predmarg postestimation statement (RTI, 2012) to estimate the predictive margins for each settlement type. Generally speaking, to calculate the predictive margins for a specific predictor category, the observed values of the confounders for each case are used to estimate the probability of each case experiencing the outcome when all cases are treated as belonging to that specific category of the predictor variable, and then averaging across the probabilities for all cases. For instance, the marginal effect of residing in the urban core on the risk of IPV would be calculated by taking the mean of the estimated marginal probabilities for every NCVS respondent based on age, race/ethnicity, and year surveyed while also treating all respondents as if they lived in the urban core (regardless of their actual settlement type). These marginal effects are then calculated in a similar manner for all categories on the predictor variable. Model-adjusted relative risk ratios were estimated using the adjrr option of the predmarg statement and the statistical significance of pairwise comparisons of adjusted prevalence rates across settlement types were performed with the pred_eff statement.

Although the annual prevalence for violent and property victimization more generally for those interviewed twice per year is not statistically different than that for those interviewed only once in the first half or second half of the year (Lauritsen & Rezey, 2013), for IPV there are clear differences with the risk being substantially greater for those with a single interview. Because of those differences, the estimates reported below are made for a semiannual basis.

Results

To begin with, as shown in Table 2, the risks of IPV victimization across the conventionally defined settlement types (i.e., MSA status) were considered. Corresponding with earlier research, both unadjusted and regression adjusted estimates of victimization indicate that the semiannual rate of IPV for women who reside in nonmetropolitan settlements (2.72 per 1,000) is essentially the same (p = .451) as that for women residing in central cities (2.59 per 1,000). While women from nonmetropolitan and central city settlements were at similar risk of IPV victimization, both were at a greater risk of victimization relative to suburban women. Holding age, race/ethnicity, year of survey administration, and time in sample constant, the semiannual rate of IPV for suburban women (2.25 per 1,000) was 12% and 17% less (both p < .01) than that for women from central city and nonmetropolitan settlements, respectively.

Table 2.

Unadjusted and Logistic Regression Adjusted Semiannual Prevalence of Intimate Partner Violence Against Women Aged 18 and Above (n = 1,672,999) by Settlement Type, United States, 1994 to 2015.

SettlementType Measure	Unadjusted		Adjusted^a
SettlementType Measure	Relative Risk Ratio	Rate per 1,000 [95% CI]^b	Relative Risk Ratio	Rate per 1,000 [95% CI]
MSA status (V2129)
Central city	1.08	3.12 [2.93, 3.32]^c	0.95	2.59 [2.43, 2.76]^c
Suburb	0.79	2.29 [2.14, 2.45]	0.83	2.25 [2.11, 2.41]
Nonmetropolitan	1.00	2.89 [2.61, 3.19]^d	1.00	2.72 [2.45, 3.02]^d
Residential concentration (V2017)
Urban	1.30	2.81 [2.68, 2.95]	1.17	2.53 [2.41, 2.65]
Rural	1.00	2.17 [1.96, 2.40]	1.00	2.16 [1.96, 2.38]
Combined settlement type (V2129 × V2017)
Urban core	1.35	3.12 [2.93, 3.33]	1.13	2.60 [2.44, 2.77]^e
Suburban	1.03	2.37 [2.22, 2.54]^e	1.00	2.32 [2.17, 2.48]^e
Exurban	0.87	2.00 [1.73, 2.31]^e	0.86	1.99 [1.73, 2.29]^e
Small town	1.69	3.90 [3.33, 4.57]	1.43	3.30 [2.82, 3.87]
Dispersed rural	1.00	2.31 [2.01, 2.64]^f,g	1.00	2.31 [2.02, 2.64]^f,g,h

CI = confidence interval; MSA = metropolitan statistical area.

Results adjusted for race/ethnicity, age, year, and time in sample.

Unless otherwise noted, all absolute differences in prevalence rates between settlement type categories are statistically significant p < .05.

Prevalence difference with nonmetropolitan settlements not statistically significant (ns).

Prevalence difference with central city settlements ns.

Prevalence difference with dispersed rural settlements ns.

Prevalence difference with suburban settlements ns.

Prevalence difference with exurban settlements ns.

Prevalence difference with urban core settlements ns.

There were clear differences in the risk of IPV when comparisons are made using the Census Bureau’s urban/rural settlement type indicator with a greater risk of IPV for women residing in settlements with an urban residential concentration. After adjusting for age, race/ethnicity, and year of survey administration, the logistic regression estimates indicate that the semiannual rate of IPV victimization for women from urban settlements (2.53 per 1,000) was 17% higher (p = .001) than the semiannual rate for women from rural settlements (2.16 per 1,000).

Similar to what was found when considering Census Bureau’s urban/rural settlement type indicator by itself, comparisons using the combined settlement type measure indicate that residence in a settlement with low levels of residential concentration does not put women at greater risk of IPV victimization. When the MSA status nonmetropolitan settlements are differentiated by rural or urban residential concentration, it is women from the latter category who were at the greatest risk of IPV victimization; the logistic regression adjusted semiannual rate of IPV against women from small town settlements (3.30 per 1,000 women) was 43% greater (p = .001) than the similar rate for women from dispersed rural settlements (2.31 per 1,000). The disaggregation of nonmetropolitan settlements by residential concentration also provides evidence that the risks of IPV are greatest not for women living in the urban core but instead for women from small towns; the regression adjusted semiannual IPV rate for women from small town settlements was 27% higher (p = .014) than the rate for women living in the urban core (2.60 per 1,000). Differences in IPV risk for women from small town settlements relative to women from exurban or suburban settlements were even more substantial. The regression adjusted semiannual IPV rate for small town women was 42% higher (p < .001) than the rate for suburban women (2.32 per 1,000) and 66% higher than the rate (p = .024) for women from exurban settlements (1.99 per 1,000).

Discussion and Conclusion

The above results indicate that women from small towns—the urbanized portions of nonmetropolitan counties in the United States—are most at risk of suffering violence committed by an intimate partner. Even after accounting for the demographic structures of the different settlement types, the prevalence of IPV against women from small town settlements exceeds that for women who reside in dispersed rural areas, for women from the suburbs and outlying exurbs of metropolitan counties, and, most unexpectedly, for women who live in the urban core.

The results of this study concerning the prevalence of IPV are in line with earlier research considering its incidence across the conventional measure of settlement type (DuBois et al., 2019). There is a similar risk of IPV victimization for women from “rural” (i.e., nonmetropolitan) and “urban” (i.e., metropolitan central cities) settlements while both are at greater risk relative to women from “suburban” (i.e., metropolitan noncentral city) settlements. Rural/urban differences in IPV prevalence are much more apparent when settlements are categorized according to the U.S. Census Bureau’s indicator of residential concentration than when such categorization is made in reference to location within an MSA. The results of this study indicate that women from settlements recognizable as rural have less of a risk of IPV victimization than do women from built up settlements that are clearly urban when seen from an “airship view.”

Although the estimates reported above indicate that the greatest risk of IPV victimization is for women from small towns, those results do little to explain why that is the case. Living in the type of settlement that has experienced more than a fair share of economic decline over the past three decades does increase the risk of IPV, but in no way can that be taken to mean (a) that the specific settlements where the victims resided had more than typical levels of economic decline or (b) that the victims’ partners were more likely to have experienced job losses and the like. Any connections to social disorganization theory and/or a masculinity crisis to explain the higher risk of IPV for small town women are tenuous and likely to be ecologically fallacious.

While difficult to explain, the similarity of IPV prevalence in nonmetropolitan settlements (especially those with concentrated populations) and the urban core is not an aberration but is instead comparable to what has been found in many other settlement type comparisons across the social and health sciences. The notion of settlement types running along a continuum of urban to suburban to rural that ranges from the malignant to the idyllic is largely outdated. Over the past 20 to 30 years, there has instead been a convergence in outcomes for those living in rural and urban settlements while the lives of suburbanites are markedly healthier, wealthier, and safer than the rest. Indeed, relative to urban and nonmetropolitan settlements, it is the suburbs that are idyllic in terms of lower levels of asset poverty (Fisher & Weber, 2004), greater food security (Guerrero et al., 2014), less neighborhood decay (York Cornwell & Hall, 2017), greater education funding (Miller & Votruba-Drzal, 2013), better physical health (Eberhardt & Pamuk, 2004), more physical activity (Parks et al., 2003), reduced obesity (Ramsey & Glenn, 2002), less alcohol abuse (Borders & Booth, 2007), decreased opioid-related mortality (Rigg et al., 2018), and longer life expectancy (Elo et al., 2018).

The results of this study provide some clarity regarding the more general issue of whether women are safer when residing in urban or rural settlements. This is best dealt with by answering the question put forth by Edwards (2015) in her review of the literature on IPV across settlement types in which she asks if “the rural-urban-suburban divide” is “myth or reality?” (p. 359). As always, the answer would depend on how those terms are operationalized. If that divide is based on the U.S. Census Bureau’s urban/rural delineation using residential concentration, then the results of this study should be taken as evidence of the reality of that divide because women from urban settlements have a much higher risk of IPV than women from rural settlements. However, if the divide is based only on the delineation conventionally used to represent urban, suburban, and rural settlements in analyses of the NCVS, then the results of this study suggest that any such urban/rural divide is a myth and that our attention would better spent on understanding the suburban/non-suburban divide in the risk of IPV.

As always, there are a number of limitations that should be noted when considering the results of this study. One issue is that the estimates of IPV prevalence presented here generally are less than what has been found in other research using nationally representative samples. For instance, the annual prevalence of intimate partner rape and physical assault of 1.5% reported in the NVAWS (Tjaden & Thoennes, 2000) would still be almost twice as high as the 1995 NCVS estimated semiannual IPV prevalence of 0.41% (95% CI = [0.37%, 0.46%]) even if it were halved for a semiannual basis. Similarly, even when adjusted to a half-year, the magnitude of difference for 2011 between the NISVS annual IPV prevalence estimate of 4.0% (Breiding et al., 2014) and the NCVS semiannual estimate of 0.19% (95% CI = [0.16%, 0.23%]) is much more substantial. Although there are many explanations for these differences (e.g., the NCVS is an omnibus crime survey with general screen questions, confirmatory incident reports, and preventive procedures against telescoping, whereas the NISVS and NVAWS were specially designed to measure gender-based violence with specific questions about physical violence committed by an intimate that are primed by question sets on a partner’s psychological aggression and/or coercive control; Centers for Disease Control and Prevention, 2016; Rand & Rennison, 2005; Rosay, 2016; Tjaden & Thoennes, 1999), none provide an indication that the underestimation of IPV is unevenly distributed across demographic and geographic categories. As preferable as it would be to use the NISVS or NVAWS to consider the effects of settlement type on IPV risk, neither provides such information about respondents’ residences.

Protocols designed to protect respondent confidentiality in the public use NCVS data set preclude consideration of the local context for understanding the mechanisms underlying the associations presented above. It was only for the earliest days of the National Crime Survey (the predecessor of the NCVS) that neighborhood sociodemographic information was included in public use data files for all types of settlements (BJS, 1998). The current lack of local contextual data also limits the ability to distinguish the effects of varying levels of settlement densities upon the risk of IPV victimization. It is possible that women from the most isolated of rural settlements are at a greater risk of victimization relative to women from settlements that fall just under the Census Bureau’s criteria for classification as rural. While the U.S. Census Bureau’s dichotomous residential block-level urban/rural indicator provides some indication of residential concentration, by no means is it the best possible measure geographic isolation. More generally, the combined settlement type measure used in this study is but a first step toward recognition of the heterogeneity of rural settlements. There are many other settlement characteristics such as adjacency to metropolitan areas, the presence of amenities, and the type and strength of economic base that could have an effect on the risk of IPV that should be examined.

As with characteristics of settlements other than their type, it was also not possible to consider any of the more obvious demographic or socioeconomic correlates of IPV as confounders because of various problems with the measures available in the NCVS. For example, household income was not included in the analysis because it is unreported for roughly 30% of NCVS respondents (Harrell et al., 2014). Level of education was not included because its age-dependence makes it unsuitable as an indication of socioeconomic status; the lack of a high school degree for 18-year-olds is not the same marker of low socioeconomic status as it would be for 25-year-olds. Difficulties establishing time order on the marital status variable—marital status is measured on the date of the survey, not the date when a victimization occurred—precluded its consideration. It would not be possible to determine whether women were at greater risk because they were already separated or whether their marital separations were a result of being victimized by their partners while still married. Without inclusion of these potential confounders or the abovementioned settlement characteristics, the results of this study should not be taken as the final word on the risks of IPV associated with settlement type.

Limitations aside, the methods used for this article provide new direction to the study of the correlates of victimization using the NCVS. To date, NCVS research has largely been limited either to case–control studies of victimization aftermaths (e.g., Klein et al., 1997) or to comparisons of victimization incidence. While the latter has been used to approximate the likelihood of being victimized, estimates of the incidence rate (i.e., the number of acts of victimization per population) are best reserved for their original purpose of establishing the crime burden and determining the extent to which those offenses ever come to the attention of the police (Biderman, 1967). The main problem with using incidence statistics to denote risk is that a varying proportion of victims, somewhere between 14% and 24% depending on the year according to the NCVS (Oudekerk & Truman, 2017), are victimized multiple times such that the number of crimes committed against the population far exceeds the number of victims. This is especially true for crimes committed against intimate partners: two thirds of all incidents of IPV recorded in the NCVS between 2005 and 2014 were committed against just one third of victims (Oudekerk & Truman, 2017). At best, disaggregated incidence estimates only provide an indication of the number of cases from a given population subgroup that could be dealt with by the police.

This research demonstrates the twofold benefits of using prevalence estimates when considering patterns of victimization with NCVS data. First of all, unlike what is the case with estimates of incidence, the prevalence rate provides for a description of patterns of victimization in terms of individual risk. The results presented above establish, for the first time using the NCVS, differences in the risk of IPV victimization of women depending upon the settlements within which their day-to-day lives are lived. The other main benefit of prevalence estimates is the ability to model the relative effects of competing influences upon victimization risk. In the present study, this involved multivariate logistic regression models used to estimate the effects of settlement type upon IPV risk net of the known correlates age and race/ethnicity; statistically significant differences under one circumstance—IPV risks for urban core women relative to risks for women from dispersed rural settlements—were attenuated when adjusted for demographic controls.

The statistical methods used in this study advance the use of the NCVS as a tool for understanding variations in victimization risk more generally. Although previous research has used GEE to account for repeated measures in the NCVS (Conaway & Lohr, 1994; Lauritsen et al., 2018), they have never been estimated simultaneously with methods that account for the effects of the NCVS’s multistage stratified sampling upon statistical inference. Use of this combination of statistical methods to deal with NCVS data will finally allow us to speak in terms of which types of households are most likely to experience property crimes or which types of individuals face the greatest risk of violent victimization.

From a conceptual standpoint, this study is an improvement over earlier research on the effects of settlement type upon IPV victimization. The combined settlement type measure used in this study does a better job than the conventional measure of capturing places as homogeneous units within which everyday interactions occur. While nowhere near optimal, a combination of metropolitan classification and residential concentration better captures rural isolation than either measure does on its own. When used to consider the prevalence of IPV, this combined measure of settlement type indicates that it is not women from rural isolated areas but rather women from small towns that are most at risk of suffering violence at the hands of an intimate partner.

Footnotes

Acknowledgements

The author would like to thank Janet Lauritsen for sharing her expertise in working the NCVS—particularly in terms of making the data amenable to prevalence estimation. Of course, the author is responsible for any errors or omissions made in this article.

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Kathryn O. DuBois

Notes

Author Biography

Kathryn O. DuBois is an associate professor in the Department of Criminal Justice and Criminology at Washington State University Vancouver. Her research deals with the causes of and responses to gender-based violence in rural, remote, and indigenous communities.

References

Arbuckle

D. R.

(1998). Alternative approaches to defining metropolitan and nonmetropolitan areas. Federal Register, 63, 70526–70561.

Bachman

(1994). Violence against women: A National Crime Victimization Survey report. Bureau of Justice Statistics.

Bachman

Saltzman

L. E.

(1995). Violence against women: Estimates from the redesigned survey. Bureau of Justice Statistics.

Biderman

A. D.

(1967). Surveys of population samples for estimating crime incidence. Annals of the American Academy of Political and Social Science, 374, 16–33.

Borders

T. F.

Booth

B. M.

(2007). Rural, suburban, and urban variations in alcohol consumption in the United States: Findings from the National Epidemiologic Survey on Alcohol and Related Conditions. Journal of Rural Health, 23, 314–321.

Breiding

M. J.

Black

M. C.

Ryan

G. W.

(2008). Prevalence and risk factors of intimate partner violence in eighteen US states/territories, 2005. American Journal of Preventive Medicine, 34, 112–118.

Breiding

M. J.

Smith

S. G.

Basile

K. C.

Walters

M. L.

Chen

Merrick

M. T.

(2014). Prevalence and characteristics of sexual violence, stalking, and intimate partner violence victimization: National Intimate Partner and Sexual Violence Survey, United States, 2011. Morbidity and Mortality Weekly Report, 63, 2–18.

Bruinsma

G. J. N.

Pauwels

L. J. R.

Weerman

F. M.

Bernasco

(2013). Social disorganization, social capital, collective efficacy and the spatial distribution of crime and offenders. British Journal of Criminology, 53, 942–963.

Bureau of Justice Statistics. (1998). National Crime Surveys: National sample, 1973-1983—Codebook for full files (1973-1976) (6th ed.). Inter-University Consortium for Political and Social Research.

10.

Bureau of Justice Statistics. (2016). National Crime Victimization Survey, concatenated file, 1992-2015. Inter-University Consortium for Political and Social Research.

11.

Burton

L. M.

Lichter

D. T.

Baker

R. S.

Eason

J. M.

(2013). Inequality, family processes, and health in the “new” rural America. American Behavioral Scientist, 57, 1128–1151.

12.

Camm

J. C. R.

Irwin

P. G.

(1979). Space, people, place: Economic and settlement geography (3rd ed.). Longman Cheshire.

13.

Capaldi

D. M.

Knoble

N. B.

Shortt

J. W.

Kim

H. K.

(2012). A systematic review of risk factors for intimate partner violence. Partner Abuse, 3, 231–280.

14.

Carrington

Scott

(2008). Masculinity, rurality and violence. British Journal of Criminology, 48, 641–666.

15.

Centers for Disease Control and Prevention. (2016). National Intimate Partner and Sexual Violence Survey (NISVS): General population survey raw data, 2010—Survey instrument. Inter-University Consortium for Political and Social Research.

16.

Conaway

M. R.

Lohr

S. L.

(1994). A longitudinal analysis of factors associated with reporting violent crimes to the police. Journal of Quantitative Criminology, 10, 23–39.

17.

DeKeseredy

W. S.

Donnermeyer

J. F.

Schwartz

M. D.

Tunnell

K. D.

Hall

(2007). Thinking critically about rural gender relations: Toward a rural masculinity crisis/male peer support model of separation/divorce sexual assault. Critical Criminology, 15, 295–311.

18.

DeKeseredy

W. S.

Dragiewicz

Rennison

C. M.

(2012). Racial/ethnic variations in violence against women: Urban, suburban, and rural differences. International Journal of Rural Criminology, 1, 184–202.

19.

DeKeseredy

W. S.

Schwartz

M. D.

(2009). Dangerous exits: Escaping abusive relationships in rural America. Rutgers University Press.

20.

Demangeon

(1927). La géographie de l’habitat rural [The geography of rural settlements]. Annales De Géographie, 36, 97–114.

21.

Driezen

(2016). Analysis of longitudinal survey data: Introduction to generalized estimating equations with examples from the ITC survey. International Tobacco Control Policy Evaluation Project.

22.

DuBois

K. O.

Rennison

C. M.

DeKeseredy

W. S.

(2019). Intimate partner violence in small towns, dispersed rural areas, and other locations: Estimates using a reconception of settlement type. Rural Sociology, 84, 826–852.

23.

Duhart

D. T.

(2000). Urban, suburban, and rural victimization, 1993-98. Bureau of Justice Statistics.

24.

Eason

J. M.

(2012). Extending the hyperghetto: Toward a theory of punishment, race, and rural disadvantage. Journal of Poverty, 16, 274–295.

25.

Eaton

Kaufman

Fuhrel

Cain

Cherry

Pope

Kalichman

S. C.

(2008). Examining factors co-existing with interpersonal violence in lesbian relationships. Journal of Family Violence, 23, 697–705.

26.

Eberhardt

M. S.

Pamuk

E. R.

(2004). The importance of place of residence: Examining health in rural and nonrural areas. American Journal of Public Health, 94, 1682–1686.

27.

Edwards

K. M.

(2015). Intimate partner violence and the rural–urban–suburban divide myth or reality? A critical review of the literature. Trauma, Violence, and Abuse, 16, 359–373.

28.

Elo

I. T.

Hendi

Preston

(2018). Trends in U.S. mortality by region and metropolitan-nonmetropolitan continuum between 1990 and 2015 [Paper presentation]. Population Association of America, Denver, CO, United States. https://paa.confex.com/paa/2018/meetingapp.cgi/Paper/23470

29.

Ennis

P. H.

(1967). Criminal victimization in the United States: A report of a national survey. National Opinion Research Center.

30.

Fawcett

C. B.

(1922). British conurbations in 1921. The Sociological Review, 14, 111–122.

31.

Felson

Cohen

L. E.

(1980). Human ecology and crime: A routine activity approach. Human Ecology, 8, 389–406.

32.

Fisher

Weber

B. A.

(2004). Does economic vulnerability depend on place of residence? Asset poverty across metropolitan and nonmetropolitan areas. Review of Regional Studies, 34, 137–155.

33.

Foulkes

Schafft

K. A.

(2010). The impact of migration on poverty concentrations in the United States, 1995–2000. Rural Sociology, 75, 90–110.

34.

Gagné

P. L.

(1992). Appalachian women: Violence and social control. Journal of Contemporary Ethnography, 20, 387–415.

35.

Greenfeld

L. A.

Rand

M. R.

Craven

Klaus

P. A.

Perkins

C. A.

Ringel

Warchol

Maston

Fox

J. A.

(1998). Violence by intimates: Analysis of data on crimes by current or former spouses, boyfriends, and girlfriends. Bureau of Justice Statistics.

36.

Guerrero

Walsh

M. C.

Malecki

K. C.

Nieto

F. J.

(2014). Urban-rural and regional variability in the prevalence of food insecurity: The survey of the health of Wisconsin. Wisconsin Medical Journal, 113, 133–138.

37.

Harrell

Langton

Berzofsky

Couzens

Smiley-McDonald

(2014). Household poverty and nonfatal violent victimization, 2008–2012. Bureau of Justice Statistics.

38.

Hertz

Kusmin

Marré

Parker

(2014). Rural employment trends in recession and recovery. U.S. Department of Agriculture, Economic Research Service.

39.

Horton

N. J.

Lipsitz

S. R.

(1999). Review of software to fit generalized estimating equation regression models. The American Statistician, 53, 160–169.

40.

Jordan

T. G.

(1966). On the nature of settlement geography. The Professional Geographer, 18, 26–28.

41.

Kantor

G. K.

Jasinski

J. L.

Aldarondo

(1994). Sociocultural status and incidence of marital violence in Hispanic families. Violence and Victims, 9, 207–222.

42.

Klein

P. J.

Gerberich

S. G.

Gibson

R. W.

Maldonado

Kruttschnitt

Larntz

Renier

(1997). Risk factors for work-related violent victimization. Epidemiology, 8, 408–413.

43.

Koss

M. P.

Gidycz

C. A.

Wisniewski

(1987). The scope of rape: Incidence and prevalence of sexual aggression and victimization in a national sample of higher education students. Journal of Consulting and Clinical Psychology, 55, 162–170.

44.

Lauritsen

J. L.

Heimer

Lang

J. B.

(2018). The enduring significance of racial and ethnic disparities in male violent victimization, 1973–2010. Du Bois Review, 15, 69–87.

45.

Lauritsen

J. L.

Rezey

M. L.

(2013). Measuring the prevalence of crime with the National Crime Victimization Survey. Bureau of Justice Statistics.

46.

Lawson Clark

. (2012). In search of housing: Urban families in rural contexts. Rural Sociology, 77, 110–134.

47.

Lenhart

Anderson

Smith

(2015). Teens, technology and romantic relationships. Pew Research Center.

48.

Michniewicz

K. S.

Vandello

J. A.

Bosson

J. K.

(2014). Men’s (mis)perceptions of the gender threatening consequences of unemployment. Sex Roles, 70, 88–97.

49.

Miller

Votruba-Drzal

(2013). Early academic skills and childhood experiences across the urban–rural continuum. Early Childhood Research Quarterly, 28, 234–248.

50.

Mulvaney

(2018). OMB Bulletin No. 18-04: Revised delineations of metropolitan statistical areas, micropolitan statistical areas, and combined statistical areas, and guidance on uses of the delineations of these areas. Office of Management and Budget.

51.

Noordzij

Dekker

F. W.

Zoccali

Jager

K. J.

(2010). Measures of disease frequency: Prevalence and incidence. Nephron Clinical Practice, 115, c17–c20.

52.

Office for National Statistics. (2018). User guide to crime statistics for England and Wales.

53.

Oudekerk

B. A.

Truman

J. L.

(2017). Repeat violent victimization, 2005-14. Bureau of Justice Statistics.

54.

Parks

S. E.

Housemann

R. A.

Brownson

R. C.

(2003). Differential correlates of physical activity in urban and rural adults of various socioeconomic backgrounds in the United States. Journal of Epidemiology and Community Health, 57, 29–35.

55.

Ramsey

Glenn

L. L.

(2002). Obesity and health status in rural, urban, and suburban Southern women. Southern Medical Journal, 95, 666–671.

56.

Rand

M. R.

Rennison

C. M.

(2005). Bigger is not Necessarily Better: An analysis of violence against women estimates from the National Crime Victimization Survey and the National Violence Against Women Survey. Journal of Quantitative Criminology, 21, 267–291.

57.

Ratcliffe

Burd

Holder

Fields

(2016). Defining rural at the U.S. Census Bureau. U.S. Census Bureau.

58.

Reidy

D. E.

Berke

D. S.

Gentile

Zeichner

(2014). Man enough? Masculine discrepancy stress and intimate partner violence. Personality and Individual Differences, 68, 160–164.

59.

Rennison

C. M.

(2001). Intimate partner violence and age of victim, 1993-99. Bureau of Justice Statistics.

60.

Rennison

C. M.

DeKeseredy

W. S.

Dragiewicz

(2013). Intimate relationship status variations in violence against women: Urban, suburban, and rural differences. Violence Against Women, 19, 1312–1330.

61.

Rennison

C. M.

Rand

M. R.

(2003). Nonlethal intimate partner violence against women: A comparison of three age cohorts. Violence Against Women, 9, 1417–1428.

62.

Rennison

C. M.

Welchans

(2000). Intimate partner violence. Bureau of Justice Statistics.

63.

Research Triangle Institute. (2012). SUDAAN language manual, volumes 1 and 2, release 11.

64.

Rigg

K. K.

Monnat

S. M.

Chavez

M. N.

(2018). Opioid-related mortality in rural America: Geographic heterogeneity and intervention strategies. International Journal of Drug Policy, 57, 119–129.

65.

Rivara

F. P.

Anderson

M. L.

Fishman

Reid

R. J.

Bonomi

A. E.

Carrell

Thompson

R. S.

(2009). Age, period, and cohort effects on intimate partner violence. Violence and Victims, 24, 627–638.

66.

Rosay

A. B.

(2016). Violence against American Indian and Alaska Native women and men: 2010 findings from the National Intimate Partner and Sexual Violence Survey. National Institute of Justice.

67.

Salamon

MacTavish

(2017). Singlewide: Chasing the American dream in a rural trailer park. Cornell University Press.

68.

Schneider

Harknett

McLanahan

(2016). Intimate partner violence in the great recession. Demography, 53, 471–505.

69.

Shepherd

(2001). Where do you go when it’s 40 below? Domestic violence among rural Alaska Native women. Affilia, 16, 488–510.

70.

Shook-Sa

Couzens

G. L.

Berzofsky

(2015). User’s guide to National Crime Victimization Survey (NCVS) direct variance estimation. Research Triangle Institute International.

71.

Sunstein

C. R.

(2010). 2010 standards for delineating metropolitan and micropolitan statistical areas. Federal Register, 75, 37246–37252.

72.

Taylor

Joseph

Mumford

(2017). Romantic relationship characteristics and adolescent relationship abuse in a probability-based sample of youth. Journal of Interpersonal Violence. Advanced online publication. https://doi.org/10.1177/0886260517730566

73.

Thiede

B. C.

Kim

Valasik

(2018). The spatial concentration of America’s rural poor population: A postrecession update. Rural Sociology, 83, 109–144.

74.

Tjaden

Thoennes

(1998). Violence and threats of violence against women and men in the United States: 1994-1996 [Computer file]. Inter-University Consortium for Political and Social Research.

75.

Tjaden

Thoennes

(1999). Violence and threats of violence against women in the United States: Questionnaire annotation. Inter-University Consortium for Political and Social Research.

76.

Tjaden

Thoennes

(2000). Extent, nature, and consequences of intimate partner violence: Findings from the National Violence Against Women Survey. National Institute of Justice.

77.

Truman

J. L.

Morgan

R. E.

(2014). Nonfatal domestic violence, 2003–2012. Bureau of Justice Statistics.

78.

Truman

J. L.

Morgan

R. E.

(2016). Criminal victimization, 2015. Bureau of Justice Statistics.

79.

U.S. Census Bureau. (2012a). 2010 Census summary file 1: Table G001, geographic identifiers.

80.

U.S. Census Bureau. (2012b). 2010 Census summary file 1: Table P2, urban and rural.

81.

U.S. Census Bureau. (2012c). 2010 Census summary file 1: Table P12, sex by age.

82.

U.S. Census Bureau. (2012d). 2010 Census summary file 1: Table PCT12L, sex by age (white alone, not Hispanic or Latino).

83.

Warrington

(2001). “I must get out”: The geographies of domestic violence. Transactions of the Institute of British Geographers, 26, 365–382.

84.

Websdale

(1998). Rural woman battering and the justice system: An ethnography. SAGE.

85.

Weisheit

R. A.

Wells

L. E.

(2005). Deadly violence in the heartland: Comparing homicide patterns in nonmetropolitan and metropolitan counties. Homicide Studies, 9, 55–80.

86.

Wolf Harlow

. (1991). Female victims of violent crime. Bureau of Justice Statistics.

87.

York Cornwell

Hall

. (2017). Neighborhood problems across the rural-urban continuum: Geographic trends and racial and ethnic disparities. Annals of the American Academy of Political and Social Science, 672, 238–256.