Full Text
Sociological Perspectives 2019, Vol. 62(3) 383–401 © The Author(s) 2018 Article reuse guidelines: sagepub.com/journals-permissions DOI: 10.1177/0731121418800271 journals.sagepub.com/home/spx Urban Sociology Correctional Facility Establishments and Neighborhood Housing Characteristics Kelly McGeever1 Abstract In the 1990s, the United States experienced unprecedented correctional population growth and accommodated the increase by building new facilities. Worries about the negative effect of such facilities on property values were a primary concern and complicated the siting of facilities. The aim of this study was to examine whether establishing correctional group quarters in urban neighborhoods changed housing characteristics. Propensity score matching was used to estimate the effect of establishing a correctional group quarter between 1990–2000 on 2000 median property values, median rent, home ownership rates, and vacancy rates for 12,790 neighborhoods in 124 large U.S. cities. Housing outcomes in neighborhoods with correctional facility sitings did not differ from what would be expected if such establishments were not created. This finding held regardless of the correctional facility type. The concern of declining property values as a result of the introduction of correctional group quarters populations in neighborhoods is generally unwarranted. Keywords correctional facilities, property values, NIMBY Introduction Prior work on the impacts of correctional facility placements has focused on perceptions and reactions to correctional facilities (Garland, Wodahl, and Saxon 2017; Kilburn et al. 2014; Takahashi and Dear 1997; Takahashi and Gaber 1998). This line of research has found that com- munities fear the addition of a correctional facility to the neighborhood, citing worries about declining property values (Hawes 1985; Kilburn et al. 2014; Maxim and Plecas 1983; Shichor 1992; Thies 2000), increased crime (Farrington and Parcells 1991; Hawes 1985; Maxim and Plecas 1983; Shichor 1992; Thies 2000), and changes to the community character and reputation (Hawes 1985; Kilburn et al. 2014; Shichor 1992). Some communities, particularly those with high levels of social capital, can leverage their not-in-my-backyard (NIMBY) protests in pre- venting establishments, whereas other neighborhoods, often impoverished, become the home for the undesirable facilities and stigmatized residents. 1University of Hartford, West Hartford, CT, USA Corresponding Author: Kelly McGeever, Department of Sociology and Criminal Justice, University of Hartford, 200 Bloomfield Ave., 416 Hillyer Hall, West Hartford, CT 06117, USA. Email: [EMAIL REDACTED] 800271 PerspectivesMcGeever research-article2018 ---PAGE BREAK--- 384 Sociological Perspectives 62(3) The establishment of a correctional facility may signify that the neighborhood has become—or is entrenched—as a stigmatized place, with a loss of status for those in it. As a consequence, many residents living near proposed or operational correctional facilities report that their behaviors or intended behaviors have been altered as a result of the facility placement. Recent investigations (Martin 2000; Martin and Myers 2005; Myers and Martin 2004) of community attitudes regarding effects of prison placement suggest that while worries about safety were not overwhelmingly voiced, concern about property values “yielded the highest rate of negative responses” (Martin 2000:286) and 8 percent of the sample indicated they wanted to move farther away from the prison or out of the county altogether. Similar sentiments were voiced in regard to a correctional halfway house in a northeastern town, with researchers finding that the majority of interviewed residents believed that property values had declined as a result of the correctional placement and many respondents were willing to sell their homes below market value (Kilburn et al. 2014). Despite the evidence that community members have pessimistic outlooks about correctional facility sitings in their neighborhood, especially regarding property values, there have been few rigorous evaluations of the realities of the worries and little effort to theorize the causal pathway that links sitings to deleterious outcomes. In addition, there are a wide range of correctional facilities, from large state prisons to small halfway houses, with differing levels of stigmatiza- tion. This suggests that establishing correctional facilities are unlikely to have uniform effects on neighborhood housing characteristics. The present study considers the placement of a wide vari- ety of residential correctional facilities on property and rent values, vacancy rates, and levels of owner occupancy in urban neighborhoods. Theory and Empirical Background Stigma refers to an “attribute that is deeply discrediting” that propels an individual to go “from a whole and usual person to a tainted, discounted one” (Goffman 1963:3). Individuals or groups with stigmatized identities experience status loss and changes to life chances (Link and Phelan 2001). In addition to studies of stigmatized individuals and groups, there has been growing attention to how neighborhoods and places can become stigmatized (Keene and Padilla 2014; Sampson and Raudenbush 2004; Strike, Myers, and Millson 2004; Takahashi 1997; Wutich et al. 2014). Just as individuals’ life chances may detrimentally change because of stigma, neighborhoods’ reputations, quality of life, and housing characteristics may also deteriorate due to the “spoiled identity.” There are two key ways that correctional facilities can introduce stigma to a neighborhood or exacerbate preexisting degrees of stigma. First, correctional facilities and their inhabitants serve as very observable marks of disgrace for the neighborhood. Correctional facilities evoke the specter of criminality and associate their host communities with the imagery of crime. They can signify to outsiders looking for housing that the neighborhood is undesirable, resulting in declines in housing values and occupancy. Second, residents who protest facility establishments and fail may interpret their failure as evidence of their community’s declining political or social capital, or even confirma- tion of their community’s political and social impotence. This may spur capitalized residents to move away, further diminishing the commitment to community of the left behind residents. Correctional Facilities as Stigmatizing As a whole, correctional facilities are unwelcome additions to neighborhoods. Prisons have con- sistently been ranked as one of the human service facilities least accepted in residential neighbor- hoods (Takahashi 1997; Takahashi and Dear 1997), rating on par to factories and landfills and below drug treatment centers and homeless shelters (Takahashi and Gaber 1998). Residential support for transitional correctional housing acceptance depends on the type of ex-offenders housed. Christopher Dum, Kelly M. Socia, and Jason Rydberg (2017) found respondents were ---PAGE BREAK--- McGeever 385 less likely to favor emergency housing policies supporting ex-offenders when compared with less stigmatized groups homeless families with children). However, while emergency hous- ing for violent sex offenders received the least amount of support from respondents, it was not statistically different than the level of support provided for nonviolent drug offenders (Dum et al. 2017). In contrast, Garland and colleagues (Garland et al. 2017; Garland, Wodahl, and Schuhmann 2013) found that housing for those convicted of drug violations was preferred over violent offenders and sex offenders. The findings reveal that residents do not perceive potential harms of all correctional facilities as uniform. It is also unlikely that correctional facilities will have a uniform effect on neighborhood housing. Dear and colleagues (Dear 1992; Dear et al. 1997; Dear and Gleeson 1991; Dear and Taylor 1982; Takahashi and Dear 1997) have identified multiple facility and client characteristics that explain the varying levels of stigma attached to human service facilities. Facility characteristics such as size (smaller preferred over larger) and type (residential preferred over outpatient facili- ties; Dear 1992; Dear et al. 1997; Dear and Gleeson 1991; Dear and Taylor 1982) suggest that community facilities are more desirable and less toxic to housing characteristics than prisons and other confinement facilities. In addition, a number of the identified client characteristics functionality, unpredictability, dangerousness, and treatability) support the idea that confinement facilities have a stronger detrimental effect on housing characteristics (Dear et al. 1997; Goldstein, Brown, and Goodrich 1989). Stigma and Siting of Correctional Facilities The concept of NIMBY provides an avenue to examine resistance toward correctional facilities and their residents. NIMBY refers to the opposition voiced by residents of proposed develop- ments—such as landfills, airports, and prisons—because they fear not only potential threats to the environment and personal health but also to the associated blight and quality of life issues that may accompany the project. Failed NIMBY campaigns may signify to residents that their neigh- borhood is one in decline, lacking capital to prevent the establishment. As such, residents with the means may relocate, while others may be forced to accept the “ecological contamination” (Sampson and Raudenbush 2004) of the facility siting. The NIMBY literature provides guidance on the demographic, economic, and social charac- teristics of community group members who engage in NIMBY protests of facility sitings and also acknowledges the political capital and organizing capacity of communities. Generally, it is believed that economically advantaged and whiter neighborhoods are better positioned to influ- ential politicians and able to prevent facility sitings within their communities (Costanza, Kilburn, and Vendetti-Koski 2013; Kilburn et al. 2014). Neighborhoods that have this political capital, in conjunction with strong social ties between residents, are better able to rally against potential undesired facilities (Aldrich and Crook 2008; Grimes and Esaiasson 2014), and it has been sug- gested that planners anticipate the mobilization capacities and avoid such locations (Hamilton 1993). Residential facilities, such as Federal Emergency Management Agency (FEMA) trailers, homeless shelters, and criminal justice facilities, have been sited in places associated with weaker social and political capacities (Aldrich and Crook 2008; Grimes and Esaiasson 2014). Targeting stigmatized neighborhoods that may not have the social, financial, or political capi- tal to wage successful NIMBY campaigns is one potential remedy for facility siting committees. Neighborhoods that have “bad” reputations due to the level of poverty, concentrations of racial and ethnic minorities, or perceived dangerousness might be deemed suitable and easy placement sites for undesirable facilities. There is evidence of disproportionately siting hazardous facilities hazardous waste, storage, and disposal facilities) in places with greater racial and ethnic minority presence (Mohai and Saha 2006, 2007; Pastor, Sadd, and Hipp 2001; Pinderhughes 1996; Saha and Mohai 2005). The limited work linking human service agency siting ---PAGE BREAK--- 386 Sociological Perspectives 62(3) and environmental racism has found that correctional halfway houses are located in primarily non-white neighborhoods (Costanza et al. 2013; Goldstein et al. 1989). The facility can perpetu- ate and solidify the marginalized status of the neighborhood, creating a spatial stigma, where individuals in “vilified and degraded locales may come to embody the perceived negative char- acteristics of their environment” (Keene and Padilla 2014:393). Conversely, residents may try to distance themselves from this marked territory by moving away from the “degraded place” (Broto et al. 2010) and the housing values and occupancy rates may drop. Stigmatization and Urban Location The current study departs from previous work (Abrams and Lyons 1987; et al. 1984; Stanley 1978) by focusing on correctional facilities in urban neighborhoods. While the prior research helps inform this study, there are several drawbacks to concentrating solely on rural places. First, the findings from rural studies may not hold in urban areas. Many rural communi- ties have actively recruited state and federal prisons to their area to stimulate the local economy (Courtright et al. 2010; Gibbons and Pierce 1995; Hooks et al. 2004; Swanson 1993). Therefore, many rural residents perceive correctional facilities as a sign of hope rather than stigmatization. John M. Eason (2017:16) also suggests that rural towns that established prisons were already suffering from “the quadruple stigma of rurality, race, region and poverty.” The combination of already being a stigmatized place with the active recruitment of the correctional facility suggests that feelings of disinvestment, residential withdrawal, and changes to neighborhood housing are unlikely to occur in rural areas. The second drawback of prior studies is the level of geographic aggregation. The prison stud- ies have used counties et al. 1984) and cities (Hawes 1985) as the units of analysis, but neighborhood effects of prisons have not been fully assessed. The impact of correctional facili- ties, if any, is expected to occur at a localized level. A city resident is likely to be indifferent to a proposed correctional facility siting across town, but may be adamantly against a facility pro- posed for next door. Indeed, Brett Garland and others (2013) found that 50 percent of respondents supported transitional housing in one’s city compared with just 25 percent indicating support for ex-offender housing in one’s neighborhood. One would not expect the impact of correctional facilities on property values to extend uniformly across a city. Such small, localized effects would likely be absorbed or masked at a more aggregate ecological level. Third, a nontrivial number of people live in correctional facilities in urbanized areas. According to the 2000 decennial census, over 30 percent of the total correctional population are located in urbanized areas, with large variations depending on the type of correctional facility. For instance, the majority (71%) of the halfway house correctional population lives in urbanized areas com- pared with about a third of the federal prison population. There is also evidence that the growth of correctional facilities has been predominantly experienced in urban areas. In the 10 states that experienced the largest growth in the number of prisons, metropolitan counties (241% increase) were much more likely than nonmetro counties (62% increase) to receive a state or federal prison (Lawrence and Travis 2004). Correctional Facilities and Property Values While prior research has consistently suggested that residents expect lower property values in places with correctional facilities, the research on whether the perceptions represent reality is inconsistent. Several studies (Abrams and Lyons 1987; Hawes 1985; Stanley 1978) found that there were no negative effects of prisons to housing prices in the surrounding community, and one study suggested communities with prisons had higher property values et al. 1984). These studies, however, are rather dated and have a number of methodological drawbacks that minimize the generalizability and reliability of their findings (see McShane, Williams, and ---PAGE BREAK--- McGeever 387 Wagoner 1992 for full critique). For instance, lack of appropriate control groups, failing to ade- quately describe measures or matching techniques, and misuse of statistics lead the authors (McShane et al. 1992:116) to conclude that most of these studies are poorly designed and are simply unable to determine whether measured outcomes are a product of the prison or of other factors . . . These studies are almost always case studies and the external validity of their findings is essentially nonexistent. To fully assess the role of correctional facilities’ impact on housing characteristics, it is important to not only have a generalizable sample with accurate baseline measures prior to construction/operation but also have a well-selected comparative control to determine whether rates of change differ. Data and Method Sample This study focuses on the 1990–2000 decade, which experienced large increases in the number of correctional facilities, as well as the number of people incarcerated. Between 1990 and 2000, 381 new correctional facilities were established (United States Department of Justice. Office of Justice Programs. Bureau of Justice Statistics 2005) and the inmate population sentenced to state and federal facilities increased by approximately 80 percent (Beck and Harrison 2001). The area of interest is urban neighborhoods, and cities with populations of 100,000 people or more in the year 2000 provided the sampling universe. There were 239 cities that met this popu- lation requirement in 2000 with a corresponding 18,912 census tracts. The analyses were restricted to census tracts that had no established adult or juvenile correctional population in 1990 (n = 17,541), that had at least one house in the census tract in both 1990 and 2000 (n = 17,260), and that were in a city that established at least one correctional facility between 1990 and 2000 (n = 12,790 tracts in 124 cities). As it was likely that census tract boundaries were redrawn between 1990 and 2000, I used data from the Neighborhood Change Database (NCDB) (GeoLytics 2000) that normalizes 1990 tract-level census data to 2000 boundaries. Outcome Measures There were four main outcome variables related to housing characteristics of a neighborhood and all were derived from the 2000 U.S. decennial census. First, the median property value of owner- occupied homes for the tract was used. The U.S. Census collects information on the median value of one-family houses on less than 10 acres without a business or medical office on the property. As the demand for housing may be lower in neighborhoods that receive correctional facilities, the aver- age rent for specified renter-occupied homes (excludes one-family houses on 10 acres or more) was used as a second outcome measure. Third, the percentage of occupied houses that are owner occu- pied was considered. According to NIMBYists, it would be expected that homeowners would flee their neighborhood out of fear of correctional facilities and their residents. Finally, the percentage of total houses that are vacant was also evaluated. If owners do flee their homes and neighborhoods because of fear, and have a difficult time finding a new buyer, housing may sit vacant while it is on the market. In addition, if the neighborhood is mostly rentals and the area is deemed undesirable, more apartments may be vacant because potential renters are avoiding the area. Treatment Variables Group quarters is the term that the U.S. Census uses to describe places where nonrelated indi- viduals live or stay, in a group living arrangement, which is owned or managed by an entity or ---PAGE BREAK--- 388 Sociological Perspectives 62(3) organization providing housing and/or services for the residents. The census collects informa- tion on a variety of adult and juvenile, public and private, correctional group quarters popula- tions. Correctional group quarters indicate the presence of at least one correctional facility. Census data do not, however, indicate the number of correctional facilities in the census tract. Rather, they provide the total number of persons living in a correctional group quarter(s). In theory, this count can refer to just one facility containing the entire specified population or numerous small facilities within a tract. While it is not possible to determine whether there are multiple facilities of the same type, it is possible to determine whether there are different types of correctional facilities within a neighborhood. Several treatment variables were derived from the census correctional group quarters data. The treatment variables were all dichotomous, with 1 signifying treatment a correctional population is founded by 2000) and 0 denoting nontreatment no correctional population in 1990 or 2000). I used the count of people living in federal prison and detention centers, halfway houses, state prisons, other types of adult correctional facilities (the census indicates such facili- ties house alcohol/drug dependents), training schools for juvenile delinquents, and short-term care, detention or diagnostic centers for delinquent youth. I did not include military disciplinary barracks, local jails, or police lockups in the measures. While it would be ideal to include local jails as a correctional group quarter type, the census groups together jails with police lockups— temporary, nonresidential holding facilities for suspects awaiting interrogation, processing, and transfer (42 USCS § 15609, 2003). If a census tract had a nonzero population in 2000 for any of the above listed correctional group quarters, the census tract was considered a treated tract; oth- erwise, it was considered a control tract. The first analysis evaluated whether the introduction of any correctional group quarter popu- lation in the tract impacted neighborhood property characteristics. To determine whether the type of correctional facility population exert a different effect on property characteristics, two succes- sive models evaluated whether a neighborhood received a community correctional facility population between 1990 and 2000 and a confinement correctional facility population between 1990 and 2000. I included halfway houses and the facilities for alcohol/drug dependents as community correctional populations and populations living in juvenile facilities, federal pris- ons, and state prisons as confinement populations. Treatment is determined by the absence of any correctional quarters populations in 1990, but a presence by type in 2000, according to the U.S. Census data. The analyses were then further disaggregated by correctional type with separate treatments for establishments of halfway houses, other correctional facilities, juvenile facilities, and a combined federal/state prison measure. Covariates I include numerous covariates to try to account for the neighborhood selection for correctional facility placement. Sociodemographic indicators attempt to capture the characteristics of the typical NIMBYist and control for gender, age, income, education, family composition, and income level. Measures of neighborhood racial and ethnic composition, as well as the percent- age of the neighborhood living in noncorrectional group quarters, are included to capture the potential effects of environmental racism and clustering of facilities. To account for social and political resources and involvement, I use electoral participation and socioeconomic indica- tors. Housing stock and land use indicators are also considered and measures of population, density, and region are included. Apart from some of the city, county, and state variables, all the covariates that are used are census tract measures from 1990 available from the NCDB. Table 1 lists the covariates used to create a matched data set of treated and control groups; provides the source for census tract, and city, county, and state variables; and reports descrip- tive statistics for all variables used in the analyses. ---PAGE BREAK--- McGeever 389 Table 1. Means and Standard Deviations for Dependent and Independent Variables by Presences of Correctional Population. Variable names All neighborhoods N = 12,790 Neighborhoods without correctional populations n = 12,461 Neighborhoods that establish a correctional population between 1990 and 2000 n = 329 t-test significance M SD M SD M SD Outcome variables Property values 2000 $153,392 $144,285 $154,289 $144,911 $119,433 $113,220 Median rent 2000 $656.98 $198.52 $659.58 $259.33 $558.71 $198.52 % Owner occupied 2000 47.98 24.91 48.16 24.95 40.92 21.97 % Vacant 2000 7.61 6.73 7.57 6.71 9.48 7.15 Covariates Population characteristics Tract population 3,619.8 1,862.5 3,612.99 1,861.37 3,879.22 1,889.40 * Population density 14,731.1 21,295.6 14,829.45 21,403.47 11,472.35 16,397.97 % Non-Hispanic White 52.99 35.33 53.31 35.33 40.93 32.86 % Non-Hispanic black 26.55 34.13 26.30 34.07 36.12 35.06 % Hispanic 14.91 21.39 14.83 21.31 17.83 24.09 * % Male 47.99 4.53 47.98 4.48 48.32 6.05 NS % Elderly (65+) 12.26 7.12 12.25 7.08 12.40 8.39 NS % Kids (<18) 9.08 3.64 9.06 3.62 9.93 4.05 Adult child ratio 4.55 12.79 4.56 12.91 4.19 7.13 NS % Foreign born 14.26 15.87 14.32 15.90 11.90 14.61 Family structures % female-headed households 33.21 21.08 32.96 21.00 42.78 21.59 % married 41.85 15.28 42.04 15.24 34.41 14.71 % divorced 9.72 4.33 9.68 4.25 11.16 6.44 Socioeconomic characteristics % high school dropout 14.40 13.17 14.31 13.11 17.61 15.02 % college graduate 20.85 17.32 20.99 17.38 15.64 14.34 % poverty 19.36 16.22 19.16 16.16 27.10 16.36 % of householdss on public assistance 12.50 12.50 12.35 12.44 18.21 13.54 % unemployed 9.50 7.76 9.44 7.72 11.89 8.89 Median household incomea $34,409 $19,986 $34,585 $20,052 $27,743 $15,983 % professional jobs 29.16 15.09 29.28 15.09 24.77 14.26 Housing characteristics Property valuesa $134,549 $129,896 $135,313 $130,516 $105,620 $99,623 % owner occupied 47.72 25.13 47.92 25.17 40.04 22.60 % vacant 8.48 6.93 8.44 6.88 10.07 8.31 % moved 48.34 15.28 48.27 15.32 51.03 13.62 Median renta $582 $249 $583 $250 $518 $204 % other group quarters 2.42 8.33 2.37 8.27 4.20 10.19 (continued) ---PAGE BREAK--- 390 Sociological Perspectives 62(3) Analytic Approach Neighborhoods that established a correctional facility between 1990 and 2000 may be systematically different from neighborhoods that did not establish a facility, and the difference between these neigh- borhoods may have existed well before the facility. These potential differences could lead to biased estimates of the effect of a correctional facility on neighborhood housing characteristics. In attempt Variable names All neighborhoods N = 12,790 Neighborhoods without correctional populations n = 12,461 Neighborhoods that establish a correctional population between 1990 and 2000 n = 329 t-test significance M SD M SD M SD % multiunit houses 47.51 31.21 47.42 31.26 50.94 29.08 * Average # of rooms 4.92 1.04 4.93 1.04 4.65 0.94 % of older houses 28.53 27.02 28.43 27.02 32.62 26.92 % of houses built since 1980 6.76 12.28 6.78 12.31 5.92 11.08 NS Average # of people per household 2.66 0.60 2.66 0.60 2.64 0.64 NS State, county, and city characteristics Violent crime rateb 1,550.17 689.33 1,554.58 687.78 1,383.17 727.36 Property crime ratesb 8,166.00 2,034.81 8,167.20 2,031.07 8,128.03 2,174.86 NS % Republicanc 46.40 12.10 46.39 12.08 46.55 12.65 NS Police per capitad 2.71 1.00 2.72 0.99 2.55 1.17 Imprisonment ratee 268.89 112.29 268.67 110.61 277.21 163.82 NS % of state budget that is correctionalf 5.71 1.91 5.72 1.90 5.48 2.05 * Form of governmentg 1: mayor council 0.66 0.47 0.66 0.47 0.56 0.50 2: council-manager 0.31 0.46 0.31 0.46 0.41 0.49 3: commission 0.03 0.16 0.03 0.16 0.03 0.16 NS Region South 0.26 0.48 0.26 0.44 0.29 0.46 NS West 0.25 0.43 0.24 0.43 0.28 0.45 Northeast 0.26 0.44 0.26 0.44 0.18 0.39 Midwest 0.24 0.43 0.24 0.43 0.24 0.43 NS aAdjusted to 2000 dollars using appropriate factor from Consumer Price Index Research Series Using Current Methods (CPI-U-RS). bAverage city crime rates per 100,000 population during 1988 to 1990 from Uniform Crime Reports (http://www. ucrdatatool.gov/Search/Crime/Local/LocalCrime.cfm). cPercent of county who voted for Republican presidential candidate in 1988 from American Votes 1988. dAverage number of police per 1,000 population during 1988 to 1990 in city from Uniform Crime Reports: Police Employee Data. eAverage state imprisonment rate during 1988 to 1990 per 100,000 population from Bureau of Justice Statistics (http://www.bjs.gov/index.cfm?ty=nps). fAverage percent of the 1989–1990 state budget that is correctional from National Association of State Budget Officers. gCity form of government from City Databook 1986 (http://ccdb.lib.virginia.edu/). *p < .05. < .010. < .000. (Two-tailed test). Table 1. (continued) ---PAGE BREAK--- McGeever 391 to control for the selection effects, propensity score matching was used, essentially creating a data set of treated observations matched to control neighborhoods that are alike in every way except treat- ment status. The propensity scores, or the predicted probability of establishing a correctional facility, were obtained from probit regression using the identified covariates. Following Marco Caliendo and Sabine Kopeinig (2008), I attempted a variety of matching techniques nearest neighbor, kernel matching) to reduce the most amount of bias. While all techniques resulted in similar findings sug- gesting robustness, using nearest neighbor that matched the propensity score of one treated observa- tion to the 10 best control observations with replacement resulted in the largest decrease in overall standardized bias and best overall balance between the treatment and control groups. More explicitly, the difference in means between the treated and control groups were reduced to nonsignificance in t tests for all covariates and all standardized biases were reduced to 10 percent or less. Graphical evi- dence of the improvement of the standardized bias is provided in Figure 1. After the matching procedure, I calculated the average treatment effect on the treated (ATT). This is the average effect of the treatment among neighborhoods that established a correctional facility and is the difference between the means of treated and control neighborhoods and esti- mates the causal effect of establishing correctional facilities on housing indicators. Rather than relying solely on p values when evaluating the ATT, I follow Carlisle Rainey’s (2014) emphasis on confidence intervals for “arguing for a negligible effect” when test statistics fail to reject the null. This procedure allows one to assess whether meaningful, but not necessarily statistically significant, effects are present. All analyses were completed using STATA 14 and implemented programs teffects, psmatch2, and pstest (StataCorp 2014). Missing Values While the majority of the covariates used to match come from the 1990 decennial census and therefore do not have missing values, the inclusion of city, county, and state variables introduced some missingness. Measures for the number of police officers per capita and city violent crime rates both contained missing values. There are many techniques to deal with missing data, but I used the missing indicator approach because it has a theoretical justification (Rosenbaum and Rubin 1984) and is easy to implement and effective in propensity score analyses (Harder, Stuart, and Anthony 2010; Haviland, Nagin, and Rosenbaum 2007). I used mean imputation for the above variables and included a missing indicator to represent the pattern of missingness, as miss- ing patterns may also reveal something about the propensity for particular neighborhoods to establish correctional facilities (Haviland et al. 2007). Following Amelia Haviland and col- leagues (2007), the missing indicators were not used in the balancing estimates. Finally, I included a flag variable to indicate whether the census tract had a zero tract population in 1990. Results Descriptive Statistics As Table 1 indicates, the establishment of a correctional facility in urban neighborhoods between 1990 and 2000 was a rare event, as only 2.6 percent (n = 329) of the census tracts were affected. Significant differences for unadjusted means exist for many covariates, including those related to neighborhood socioeconomic and housing characteristics. Neighborhoods with correctional populations established by 2000 have higher percentages of non-Hispanic blacks, Hispanics, female-headed households, poverty, and unemployment in 1990 than neighborhoods that do not introduce correctional groups. Neighborhoods that establish correctional facilities are also more likely to have higher rates of other group quarters populations, multiunit housing, and vacant housing, as well as lower property values, median rents, and rates of owner occupancy than other neighborhoods in 1990. State, county and city covariates had few significant differences between the treated and control groups. ---PAGE BREAK--- 392 Sociological Perspectives 62(3) Figure 1. Standardized bias before and after matching. ---PAGE BREAK--- McGeever 393 In 2000, the mean property values and average rents for neighborhoods with a correctional facility were nearly $35,000 and $100, respectively, lower than in neighborhoods without cor- rectional facilities. In addition, control neighborhoods had higher rates of owner occupancy and lower rates of vacancy than the treated neighborhoods in 2000. Matching Results: Neighborhood Mean Property and Rent Values Table 2 reports the balance statistics and the causal estimates for the effect of establishing correc- tional facility between 1990–2000 on 2000 neighborhood property and rent values. The mean stan- dardized bias was much improved after matching, with reductions in bias ranging from 75.8 to 91.5 percent. The standardized bias for each covariate after matching was greatly reduced (see Figure and indicates that after matching, the treated neighborhoods are matched to control neighborhoods that are nearly identical in socioeconomic, demographic, and other measured characteristics. In all cases, the average treatment effect of establishing a correctional facility of any kind between 1990 and 2000 did not result in substantively meaningful property value differences. The introduction of a correctional facility in a neighborhood between 1990 and 2000 was asso- ciated with property values that were approximately $1500 less than the matched control neigh- borhoods in 2000, but this difference was a negligible change. The establishment of a confinement facility (federal prison, state prison, and juvenile institutions) resulted in property values that were, on average, $6,347 less than neighborhoods without correctional facility establishments, but this was not a substantively meaningful change. The estimate of the treatment effect of hav- ing a community correctional facility (halfway houses and transitional homes for drug/alcohol rehabilitation) establishment was in the hypothesized direction (–$4,391, n.s.) but was not sta- tistically different than noncommunity facility tract property values. Disaggregating by the dif- ferent types of community facilities shows variation in changes to property values. The establishment of drug/alcohol facilities negligibly increased property values by nearly $3,900 more than control neighborhoods, but halfway home establishments were associated with a reduction of over $4,000 compared with neighborhoods without halfway home establishments. The average rent in 2000 for neighborhoods with correctional group facility establishment n.s.) was no different than the rents that would be expected if the neighborhood did not have the correctional population. Similar to the findings for neighborhood property values, neighborhoods with confinement facilities had larger differences from control neighborhoods compared with the difference experienced in community facility neighborhoods (–$22 vs. but in both cases, the differences are not substantively meaningful changes. Matching Results: Neighborhood Owner Occupied and Vacancy Similar null results for the establishment of correctional facility populations on the percent owner occupied in the neighborhood was found (see Table The founding of any type of correctional facil- ity between 1990 and 2000 resulted in less than a 1 percent decrease in the owner occupancy rates (ATT= n.s.) than would be expected if the correctional group population was not established. This null finding persisted for all varying forms of treatment confinement, community, etc.). The establishment of correctional facilities did not result in decreased occupancy rates by owners. Last, the results from the propensity score matching and percent vacant in the neighborhood are reported in Table 3. There was no evidence that establishing correctional facilities resulted in increased vacancy rates compared with control neighborhoods. In fact, the introduction of state and federal prison populations into neighborhoods was significantly related to vacancy rates that were lower than the control neighborhoods. That is, neighborhoods that established federal or state prisons between 1990 and 2000 had neighborhood vacancy rates that were 3 percent lower (p = .045) than similar control neighborhoods. This negative effect is contrary to NIMBY expec- tations but may point to the building of state and federal prisons to replace vacant housing. ---PAGE BREAK--- 394 Table 2. Causal Estimates for Neighborhood Property and Rent Values Using 10:1 Nearest Neighbor Propensity Matching. ATT SE p value 90% confidence interval (LCL/UCL) Treated observations Potential controls Mean bias before/after matching Before (raw) After (matched) % reduction in bias Property values Any correctional facility −$1,495.80 $6,557.89 .820 [–$12,284, $9,292] 329 12,461 20.1 1.7 91.5 Confinement facilities: Federal prison, state prisons, and juvenile facilities −$6,347.06 $10,640.03 .551 [–$23,850, $11,156] 123 12,667 18.2 4.4 75.8 Juvenile facilities only −$3,561.74 $9,888.56 .719 [–$19,828, $12,705] 86 12,704 17.0 2.9 82.9 Federal and state prisonsa $2,338.97 $25,737.45 .928 [–$39,999, $44,677] 38 12,751 24.5 3.4 86.1 Community facilities: Halfway houses and drug/alcohol facilities −$4,391.46 $8,363.84 .600 [–$18,150, $9,367] 212 12,578 24.2 2.5 89.7 Halfway house only −$4,188.19 $8,991.21 .641 [–$18,979, $10,602] 138 12,652 24.4 3.3 86.5 Drug/alcohol facilities only $3,897.45 $15,262.92 .798 [–$21,210, $29,005] 78 12,712 24.8 3.2 87.1 Rent Any correctional facility −$9.79 $10.89 .368 [–$27.70, $8.12] 329 12,461 20.1 1.7 91.5 Confinement facilities: Federal prison, state prisons, and juvenile facilities −$21.50 $16.95 .205 [–$49.38, $6.38] 123 12,667 18.2 4.4 75.8 Juvenile facilities only −$30.19 $18.96 .111 [–$61.38, $1.00] 86 12,704 17.0 2.9 82.9 Federal and state prisons* $2.75 $28.76 .924 [–$44.56, $50.06] 38 12,751 24.5 3.4 86.1 Community facilities: Halfway houses and drug/alcohol facilities −$12.29 $14.29 .389 [–$35.80, $11.22] 212 12,578 24.2 2.5 89.7 Halfway house only −$5.07 $18.91 .789 [–$36.18, $26.04] 138 12,652 24.4 3.3 86.5 Drug/alcohol facilities only −$14.93 $21.07 .478 [–$49.59, $19.73] 78 12,712 24.8 3.2 87.1 Note. ATT = average treatment effect on the treated; LCL = lower confidence limit; SE = Abadie–Imbens robust standard error; UCL = upper confidence limit. *p < .05. < .010. < .000. aOne treated observation dropped due to imposition of common support. ---PAGE BREAK--- 395 Table 3. Causal Estimates for Owner Occupied and Vacancy Rates Using 10:1 Nearest Neighbor Propensity Matching. ATT SE p value 90% confidence interval (LCL/ UCL) Treated observations Potential controls Mean bias before/after matching Before (raw) After (matched) % reduction in bias Percent owner occupied Any correctional facility −0.26% 1.20 .829 [–1.98, 1.97] 329 12,461 20.1 1.7 91.5 Confinement facilities: Federal prison, state prisons, and juvenile −1.49% 1.99 .455 [–3.29, 3.26] 123 12,667 18.2 4.4 75.8 Juvenile facilities only 0.67% 2.35 .775 [–3.86, 3.87] 86 12,704 17.0 2.9 82.9 Federal and state prisonsa 2.98% 4.34 .493 [–7.11, 7.17] 38 12,751 24.5 3.4 86.1 Community facilities: Halfway houses and drug/alcohol 0.22% 1.42 .878 [–2.33, 2.34] 212 12,578 24.2 2.5 89.7 Halfway house only 0.13% 1.86 .945 [–3.06, 3.06] 138 12,652 24.4 3.3 86.5 Drug/alcohol only −1.01% 2.61 .699 [–4.31, 4.28] 78 12,712 24.8 3.2 87.1 Percent vacant Any correctional facility −0.24% 0.43 .577 [–0.71, 0.70] 329 12,461 20.1 1.7 91.5 Confinement facilities: Federal prison, state prisons, and juvenile facilities −0.03% 0.62 .960 [–1.02, 1.02] 123 12,667 18.2 4.4 75.8 Juvenile facilities only −0.28% 0.82 .731 [–1.35, 1.35] 86 12,704 17.0 2.9 82.9 Federal and state prisons* −3.17% 1.58 .045* [–2.63, 2.57] 38 12,751 24.5 3.4 86.1 Community facilities: Halfway houses and drug/alcohol 0.30% 0.53 .576 [–0.87, 0.87] 212 12,578 24.2 2.5 89.7 Halfway house only 0.16% 0.65 .810 [–1.07, 1.07] 138 12,652 24.4 3.3 86.5 Drug/alcohol only 0.54% 0.82 .509 [–1.34, 1.35] 78 12,712 24.8 3.2 87.1 Note. ATT = average treatment effect on the treated; LCL = lower confidence limit; UCL = upper confidence limit. *p < .05. < .010. < .000. aOne treated observation dropped due to imposition of common support. ---PAGE BREAK--- 396 Sociological Perspectives 62(3) Additional Analyses It is worth noting that the standard errors for the property value and rent causal estimates are large, indicating large variabilities in the way neighborhood housing valuation changed after correctional facility establishments. For example, neighborhoods that have confinement facilities in 2000 have property values that are more than $6,000 less than neighborhoods without such establishments, but the standard error is more than $10,000. The means for the treated ($113,860) and control ($120,207) neighborhoods in this specification suggest, especially in the context of a 30-year mortgage, the difference is negligible. In addition, city housing markets can vary considerably across the United States. To try to account for these variations, I completed additional analyses that stratified the sample based on baseline 1990 property values. The propensity scores were recalculated and matched using a nearest neighbor 10:1 algorithm. In each strata, the difference between property values of the treated and control neighborhoods again resulted in negligible change. This provides additional evidence that the effect of correctional facility placements is negligible. I also stratified the sample based on the percentage of the tract that was composed of non- Hispanic blacks in 1990 and then ran the propensity score matching algorithms to tap into facility siting in potentially stigmatized neighborhoods. All tracts were stratified into quartiles and pro- pensity score matching (10:1 nearest neighbor) was employed. In all cases, save for one, the effect of establishing any type of correctional facility did not affect neighborhood property char- acteristics (see Table The lone exception occurred in the third quartile (neighborhoods with percent non-Hispanic black was between 7.5% and 45%) for median rent. Rent in these tracts that established correctional facilities had $34 lower rents (p = .019) than would be expected if the tracts did not have correctional facilities. It was also evident that as the percentage of non-His- panic blacks increased in the neighborhood, the neighborhood was more likely to have correc- tional facilities sited. In the first quartile, only 41 tracts were treated compared with 62, 105, and 117 in the second, third, and fourth quartiles. Discussion Prior studies have found that worries about declining neighborhood property values is a pri- mary fear articulated by those who live near proposed or newly established noxious facilities, such as correctional facilities (Hawes 1985; Maxim and Plecas 1983; Thies 2000). The cur- rent study specifically tested whether such a concern is warranted and focused on the relation- ship between establishing correctional facilities and the resulting property characteristics in urban neighborhoods. The null findings that many previous prison impact studies (Abrams and Lyons 1987; Hawes 1985) reported were echoed in this study that considered a variety of correctional facilities. There was little evidence that the introduction of correctional popula- tions between 1990 and 2000 in a sample of neighborhoods in large U.S. cities resulted in substantively meaningful changed property values in 2000 than would be expected if such facilities were not established. I argued that the attitudinal studies about correctional facility concerns and NIMBY protests that attempt to block facility sitings in residential neighborhoods were evidence that neighborhood resi- dents view correctional facilities as neighborhood stigma. I further hypothesized that confinement facilities would be the most stigmatizing facility establishment. These hypotheses were not sup- ported. It may be that the beneficial individual housing characteristics (design, amenities, etc.) or community attributes (school district) play a larger role in buyers’ and sellers’ decision-making pro- cesses than the potential negative externalities produced by the presence of a correctional facility. The “stickiness” of neighborhood reputations also may explain why the introduction of correc- tional populations has little effect on the property characteristics (Sampson 2012, 2013). It may take more than one halfway house to shift the reputation and stigmatize “good” or affluent neighbor- hoods that site correctional facilities. Likewise, a disadvantaged neighborhood that established a ---PAGE BREAK--- 397 Table 4. Causal Estimates for Neighborhood Characteristics Using 10:1 Nearest Neighbor Propensity Matching for Sample Stratified by Percent Non-Hispanic Black. ATT SE p value 90% confidence interval (LCL/UCL) Treated observations Potential controls Mean bias before/after matching Before (raw) After (matched) % Reduction in bias Property values Any correctional facility Quartile 1a: 0%–1.80% black −$21,414.37 $24,441.7 0.381 [–$61,621, $18,792] 41 3,154 32.6 5.0 84.7 Quartile 2a: 1.81%–7.46% black −$10,423.55 $15,038.79 0.488 [–$35,162, $14,315] 62 3,134 24.7 3.2 87.0 Quartile 3: 7.47%–45.07% black −$5,302.19 $13,101.56 0.686 [–$26,854, $16,250] 105 3,092 12.6 3.0 76.2 Quartile 4: 45.08%–100% black $683.25 $6,739.03 0.919 [–$10,402, $11,769] 117 3,081 16.1 2.6 83.9 Rent Any correctional facility Quartile 1: 0%–1.80% black −$0.48 $44.74 0.991 [–$74.08, $73.12] 41 3,154 32.6 5.0 84.7 Quartile 2: 1.81%–7.46% black −$34.45 $29.93 0.250 [–$83.69, $14.78] 62 3,134 24.7 3.2 87.0 Quartile 3: 7.47%–45.07% black −$34.05 $14.49 0.019* [–$56.89, $10.21] 105 3,092 12.6 3.0 76.2 Quartile 4: 45.08%–100% black $8.95 $14.13 0.527 [–$14.29, $32.19] 117 3,081 16.1 2.6 83.9 Percent owner occupied Any correctional facility Quartile 1: 0%–1.80% black 0.41 3.28 0.900 [–4.99, 5.81] 41 3,154 32.6 5.0 84.7 Quartile 2: 1.81%–7.46% black −3.09 2.79 0.267 [–7.68, 1.50] 62 3,134 24.7 3.2 87.0 Quartile 3: 7.47%–45.07% black 0.15 2.23 0.946 [–3.52, 3.82] 105 3,092 12.6 3.0 76.2 Quartile 4: 45.08%–100% black 0.80 1.85 0.665 [–2.24, 3.84] 117 3,081 16.1 2.6 83.9 Percent vacant Any correctional facility Quartile 1: 0%–1.80% black 0.34 0.76 0.658 [–0.91, 1.59] 41 3,154 32.6 5.0 84.7 Quartile 2: 1.81%–7.46% black −0.62 0.49 0.202 [–1.43, 0.19] 62 3,134 24.7 3.2 87.0 Quartile 3: 7.47%–45.07% black 0.23 0.55 0.680 [–0.67, 1.13] 105 3,092 12.6 3.0 76.2 Quartile 4: 45.08%–100% black 0.53 0.84 0.530 [–0.85, 1.91] 117 3,081 16.1 2.6 83.9 Note. ATT = average treatment effect on the treated; LCL = lower confidence limit; UCL = upper confidence limit. *p < .05. < .010. < .000. aTwo treated observations dropped due to imposition of common support for all housing measures. ---PAGE BREAK--- 398 Sociological Perspectives 62(3) correctional population may not view the facility as stigmatizing, but the introduction of the facility does little to shift the reputation or housing stock of the neighborhood for the better. Assessing whether there is a tipping point in the neighborhood in regard to the number of correctional facili- ties or the size of the correctional population in the tract is needed. Unfortunately, due to data limita- tions, this could not be explored in the current study. Neighborhood reputation can help explain one of the statistically significant relationships found in this study: lower rent in neighborhoods that establish a correctional facility (ATT= –$34.05, p = .019) and that are in the third strata of percent non-Hispanic black composition. Positive or negative reputations are likely entrenched in neighborhoods at the top and bottom strata. Neighborhoods that are in the middle are in a precarious position and may fear that the introduction of correctional facilities could be the tipping point to being labeled undesirable. One way that community stakeholders attempt to control neighborhood conditions is to ensure that houses remain occupied and lowering rent is one way to ensure such status. While fewer than 3 percent of all sampled urban neighborhoods established correctional facilities, it was clear that neighborhoods in 1990 that had higher concentrations of non-Hispanic blacks were more likely to have a correctional facility established in the following decade, consistent with the idea of environmental racism. The neighborhood quartile that represented the highest concentration of blacks had nearly 3 times the number of correctional facility sitings compared to neighborhoods with the lowest concentration of blacks. It is unclear whether correctional planning committees targeted urban black communities assuming there was less social and political capital to protest the establish- ment, whether sites in primarily black neighborhoods were advantageous to corrections’ agencies because of available space and lower property values, or whether they were places of last resort after whiter neighborhoods thwarted establishments. Regardless of the reason, the coupling of mass incar- ceration and race is visible and may further marginalize and link neighborhood residents to national incarceration trends (Schept 2015). Future research should focus on whether neighborhoods have become overburdened by hosting undesirable human service facilities and whether the establishment of correctional facilities perpetuate and/or exacerbate negative neighborhood reputations. The findings suggest that residential concerns about declining property values may be unwar- ranted in urban neighborhoods. This has positive implications for facility siting committees, which may have an easier process to overcome NIMBY protests. More importantly, these find- ings may help stabilize impoverished urban neighborhoods that have been disproportionately affected by mass incarceration (Clear 2007). Survey results suggest that as distance from home increased, the number of in-person prison visits from family members decreased (Rabuy and Kopf 2015). Keeping the incarcerated within urban neighborhoods may help retain family and social networks and allow for smoother reentry. In addition, as political representation is often determined by the population sizes reported in the decennial censuses, keeping (and counting) the incarcerated population in urban neighborhoods will provide more resources and capital that otherwise have been delegated to rural prison communities. There were data limitations that restricted detailed temporal and spatial analysis and that may have obscured stronger causal relationship between correctional facilities and property character- istics. First, this project focused on the effects of correctional facilities using census tracts as a proxy for neighborhoods. While this is a marked improvement from prior studies that used county- or city-level data, this still may be too large of a geographic entity to precisely measure the effect that is often thought of as a “not in my backyard” problem. Ideally, property and hous- ing characteristics data would be available at the block level or by address to capture more local- ized processes. I decided to use data with a broader geographic scope to include more cities and increase generalizability, particularly as so many of the prior studies were case studies. In addition to the spatial limitations, there were also temporal limitations of the data. Identifying the exact date of correctional facility establishment would have been desirable; unfortunately, the time between censuses is long and the exact date of when the correctional facilities are established in neighborhoods was unknown. Correctional facilities may have been sited immediately after the ---PAGE BREAK--- McGeever 399 1990 census was taken, immediately before the 2000 census was taken, or anytime in-between. It is possible that there was a detrimental effect of correctional facility establishments on property values in the short-term, but stabilized at some point within the decade. Conclusion Contrary to NIMBY expectations, lower property values and other unfavorable housing charac- teristics were not statistically related to establishing a correctional facility in urban neighbor- hoods. This finding held, regardless of the correctional type. This is an important finding because much of the prior work has focused on neighbors’ fears but have failed to test the empirical link- age between correctional facilities and neighborhood characteristics. The association between correctional facilities and neighborhood quality of life is ripe for studying, especially when considering the multiple types, sizes, and functions of correctional facilities. The current landscape of states closing prisons at a rapid rate and turning to halfway houses and other alternatives to prison provides another test of neighborhood reputation and environmental racism. Closing facilities in whiter neighborhoods may point to more support for environmental racism, and evaluating changes to the community after closure may help to pro- vide further evidence of the durability of neighborhood reputations. Declaration of Conflicting Interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding The author(s) received no financial support for the research, authorship, and/or publication of this article. References Abrams, Kathleen Shea and William Lyons. 1987. “Impact of Correctional Facilities on Land Values and Public Policy.” North Miami, FL: FAU-FIU Joint Center for Environmental and Urban Problems, Florida International University. Aldrich, Daniel and Kevin Crook. 2008. “Strong Civil Society as a Double-edged Sword: Siting Trailers in Post-Katrina New Orleans.” Political Research Quarterly 61(3):379–89. Beck, Allen J. and Paige M. Harrison. 2001. “Prisoners in 2000.” Bureau of Justice Statistics Bulletin, NCJ 188207. Washington, DC: U.S. Department of Justice, Office of Justice Programs. Broto, Vanesa Castán, Kate Burningham, Claudia Carter, and Lucia Elghali. 2010. “Stigma and Attachment: Performance of Identity in an Environmentally Degraded Place.” Society & Natural Resources 23(10):952–68. Caliendo, Marco and Sabine Kopeinig. 2008. “Some Practical Guidance for the Implementation of Propensity Score Matching.” Journal of Economic Surveys 22(1):31–72. Clear, Todd R. 2007. Imprisoning Communities: How Mass Incarceration Makes Disadvantaged Neighborhoods Worse. New York: Oxford University Press. Costanza, S. John C. Kilburn, Jr., and Susan Vendetti-Koski. 2013. “Are Minority Areas Disproportionately Targeted for Halfway House Placement?” Journal of Ethnicity in Criminal Justice 11:256–76. Courtright, Kevin Susan H. Packard, Michael J. Hannan, and Edward T. Brennan.2010. “Prisons and Rural Communities: Exploring the Health of the Relationship and the Possibility of Improvement.” The Prison Journal 90(1):69–93. Dear, Michael. 1992. “Understanding and Overcoming the NIMBY Journal of the American Planning Association 58(3):288–300. Dear, Michael, Sharon Lord Gaber, Lois M. Takahashi, and Robert D. Wilton. 1997. “Seeing People Differently: The Sociospatial Construction of Disability.” Environment and Planning D: Society and Space 15:455–80. ---PAGE BREAK--- 400 Sociological Perspectives 62(3) Dear, Michael and Brendan Gleeson. 1991. “Community Attitudes toward the Homeless.” Urban Geography 12(2):155–76. Dear, Michael and S. Martin Taylor. 1982. Not on Our Street: Community Attitudes to Mental Health Care. London, England: Pion Ltd. Dum, Christopher, Kelly M. Socia, and Jason Rydberg. 2017. “Public Support for Emergency Shelter Housing Interventions Concerning Stigmatized Populations: Results from a Factorial Survey.” Criminology & Public Policy 16(3):833–74. Eason, John M. 2017. Big House on the Prairie: Rise of the Rural Ghetto and Prison Proliferation. Chicago, IL: University of Chicago Press. Farrington, Keith and R. Pete Parcells. 1991. “Correctional Facilities and Community Crime Rates: Alternative Hypotheses and Competing Explanations.” Humboldt Journal of Social Relations 17(1/2):17–127. Garland, Brett, Eric Wodahl, and Caryn Saxon. 2017. “What Influences Public Support of Transitional Housing Facilities for Offenders during Recovery?” Criminal Justice Policy Review 28(1):18–40. Garland, Brett, Eric Wodahl, and Robert Schuhmann. 2013. “Value Conflict and Public Opinion toward Prisoner Reentry Initiatives.” Criminal Justice Police Review 24:27–48. GeoLytics. 2000. Neighborhood Change Database. East Brunswick, NJ: GeoLytics. Gibbons, Stephen G. and Gregory L. Pierce. 1995. “Politics and Prison Development in a Rural Area.” The Prison Journal 75(3):380–89. Goffman, Erving. 1963. Stigma: Notes on the Management of Spoiled Identity. Englewood, NJ: Prentice Hall. Goldstein, Marc Cheryl H. Brown, and Elizabeth J. Goodrich. 1989. “Public Preferences and Site Location of Residential Treatment Facilities.” Journal of Community 17:186–93. Grimes, Marcia and Peter Esaiasson. 2014. “Government Responsiveness: A Democratic Value with Negative Externalities?” Political Research Quarterly 67(4):758–68. Hamilton, James. 1993. “Political and Social Costs: Estimating the Impact of Collective Action on Hazardous Waste Facilities.” RAND Journal of Economics 24(1):101–25. Harder, Valerie Elizabeth A. Stuart, and James C. Anthony. 2010. “Propensity Score Techniques and the Assessment of Measured Covariate Balance to Test Causal Associations in Research.” Methods 15(3):234–49. Haviland, Amelia, Daniel S. Nagin, and Paul R. Rosenbaum. 2007. “Combining Propensity Score Matching and Group-based Trajectory Analysis in an Observational Study.” Methods 12(3):247–67. Hawes, Jerry A. 1985. “Cities with Prisons: Do They Have Higher or Lower Crime Rates?” in Special Report to Senator Robert Presley, Chair Joint Committee on Prison Construction and Operations. Sacramento, CA: California State Senate Office of Research. tions/Abstract.aspx?id=100945). Hooks, Gregory, Clayton Mosher, Thomas Rotolo, and Linda Lobao. 2004. “The Prison Industry: Carceral Expansion and Employment in U.S. Counties, 1969–1994.” Social Science Quarterly 85(1):37–57. Keene, Danya E. and Mark B. Padilla. 2014. “Spatial Stigma and Health Inequality.” Critical Public Health 24(4):392–404. Kilburn, John Stephen E. Costanza, Kelly Frailing, and Stephanie Diaz. 2014. “A Paper Tiger on Chestnut Lane: The Significance of NIMBY Battles in Decaying Communities.” Urbanities 4(2):3-20. Lawrence, Sarah and Jeremy Travis. 2004. “The New Landscape of Imprisonment: Mapping America’s Prison Expansion.” Washington, DC: Urban Institute. Link, Bruce G. and Jo C. Phelan. 2001. “Conceptualizing Stigma.” Annual Review of Sociology 27:363–85. Martin, Randy. 2000. “Community Perceptions about Prison Construction: Why Not in My Backyard?” The Prison Journal 80(3):265–94. Martin, Randy and David L. Myers. 2005. “Public Response to Prison Siting: Perceptions of Impact on Crime and Safety.” Criminal Justice and Behavior 32(2):143–71. Maxim, Paul and Darryl Plecas. 1983. “Prisons and Their Perceived Impact on the Local Community: A Case Study.” Social Indicators Research 13(1):39–58. McShane, Marilyn Frank P. Williams, and Carl P. Wagoner. 1992. “Prison Impact Studies: Some Comments on Methodological Rigor.” Crime & Delinquency 38(1):105–20. Mohai, Paul and Robin Saha. 2006. “Reassessing Racial and Socioeconomic Disparities in Environmental Justice Research.” Demography 43(2):383–99. ---PAGE BREAK--- McGeever 401 Mohai, Paul and Robin Saha. 2007. “Racial Inequality in the Distribution of Hazardous Waste: A National- Level Reassessment.” Social Problems 54(3): 343-70. Myers, David L. and Randy Martin. 2004. “Community Member Reactions to Prison Siting: Perceptions of Prison Impact on Economic Factors.” Criminal Justice Review 29(1):115–44. Pastor, Manuel, Jim Sadd, and John Hipp. 2001. “Which Came First? Toxic Facilities, Minority Move Up, and Environmental Justice.” Journal of Urban Affairs 23(1):1–21. Pinderhughes, Raquel. 1996. “The Impact of Race on Environmental Quality: An Empirical and Theoretical Discussion.” Sociological Perspectives 39(2):231–48. Rabuy, Bernadette and Daniel Kopf. 2015. Separation by Bars & Miles: Visitation in State Prisons. Northampton, MA: Prison Policy Initiative. Retrieved January 22, 2018 org/factsheets/visitation2015.pdf). Rainey, Carlisle. 2014. “Arguing for a Negligible Effect.” American Journal of Political Science 58(4):1083–91. Rosenbaum, Paul R. and Donald B. Rubin. 1984. “Reducing Bias in Observational Studies Using Subclassification on the Propensity Score.” Journal of the American Statistical Association 79:516–24. Saha, Robin and Paul Mohai. 2005. “Historical Context and Hazardous Waste Facility Siting: Understanding Temporal Patterns in Michigan.” Social Problems 52:618–48. Sampson, Robert J. 2012. Great American City: Chicago and the Enduring Neighborhood Effect. Chicago, IL: University of Chicago Press. Sampson, Robert J. 2013. “Thinking about Context.” City & Community 12:28–34. Sampson, Robert J. and Stephen W. Raudenbush. 2004. “Seeing Disorder: Neighborhood Stigma and the Social Construction of ‘Broken Windows.’” Social Quarterly 67(4):319–42. Schept, Judah. 2015. Progressive Punishment: Job Loss, Jail Growth, and the Neoliberal Logic of Carceral Expansion. New York: New York University Press. Shichor, David. 1992. and Realities in Prison Siting.” Crime & Delinquency 38(1):70–87. John Ortiz, Carl E. Ferguson, Jr., David C. Cheng, Carolyn Trent, Barbara French, and Annette Waters. 1984. “Effects of a Prison Facility on the Regional Economy.” Journal of Criminal Justice 12(6):521–39. Stanley, Craig E. 1978. “Impact of Prison Proximity on Property Values in Green Bay and Waupun, Wisconsin.” Milwaukee, WI: Wisconsin Division of Corrections and Bureau of Facilities Management. StataCorp. 2014. Stata: Release 14. Statistical Software. College Station, TX: StataCorp. Strike, Carol Ted Myers, and Margaret Millson. 2004. “Finding a Place for Needle Exchange Programs.” Critical Public Health 14(3):261–75. Swanson, Cheryl. 1993. “Citizens’ Perceptions of Prison Effects on Their Community.” State and Local Government Review 25(2):107–16. Takahashi, Lois M. 1997. “When Does ‘Race’ Matter? Exploring Variation in Attitudes toward Controversial Facilities.” Urban Geography 18(5):451–59. Takahashi, Lois M. and Michael J. Dear. 1997. “The Changing Dynamics of Community Opposition to Human Service Facilities.” Journal of the American Planning Association 63(1):79–93. Takahashi, Lois M. and Sharon Lord Gaber. 1998. “Controversial Facility Siting in the Urban Environment.” Environment and Behavior 30(2):184–215. Thies, Jeanie M. 2000. “Prisons and Host Communities: Debunking the and Building Community Relations.” Corrections Today 62(2):136–39. United States Department of Justice. Office of Justice Programs. Bureau of Justice Statistics. 2005. “Census of State and Federal Adult Correctional Facilities, 2000.” Ann Arbor, MI: Inter-University Consortium for Political and Social Research [distributer]. Wutich, Amber, Alissa Ruth, Alexandra Brewis, and Christopher Boone. 2014. “Stigmatized Neighborhoods, Social Bonding, and Health.” Medical Anthropology Quarterly 28(4):556–77. Author Biography Kelly McGeever is an assistant professor of sociology and criminal justice at the University of Hartford. Her research interests include corrections, sentencing, and communities and crime. Her work has been published in Criminology, Journal of Quantitative Criminology, and Law & Society Review.