Examining the Concentration of Registered Sex Offenders in Upstate New York Census Tracts

Abstract

This study examined the census tract characteristics associated with the spatial concentration of registered sex offender (RSO) residences in 1,823 census tracts across 53 counties in upstate New York. The concentration of RSOs for each tract was measured using excess risk scores that essentially measure disproportionate concentrations of RSOs based on the resident population of the county and tract. The tract characteristics examined included structural characteristics from the 2010 Census, such as indicators of social disorganization, housing availability and affordability, and population density, legal characteristics describing the presence of residence restrictions, and controls for spatial autocorrelation and regional differences. Results indicate that registered sex offenders (RSOs) are disproportionately more likely to be found in tracts exhibiting high levels of concentrated disadvantage, available housing, and affordable housing, and disproportionately less likely to be found in tracts with high levels of ethnic heterogeneity. Controlling for spatial autocorrelation (lag) did not change overall results but was significantly and positively associated with excess risk. Implications for future policy and research practices are discussed.

Keywords

registered sex offenders residence restrictions social disorganization housing geographic information systems

For the past few decades, Americans have become increasingly interested in knowing about, and attempting to control, where convicted sex offenders live once they are released from prison. This interest has been fueled by widespread media coverage of heinous sex crimes involving children (e.g., Megan Kanka, Jacob Wetterling, Adam Walsh, Jessica Lunsford), as well as fears about a “missing child” epidemic (Best, 1990; Garland, 2001; Socia & Stamatel, 2010). This has led to a number of new and revised criminal justice policies, including the registration of sex offender addresses with law enforcement officials, the release of this information to the community, increased electronic monitoring provisions, and residence restriction policies. The last of these are policies that prohibit RSOs from living near child congregation locations (e.g., schools, daycares, parks).

This issue is important given the research suggesting that living near RSOs has tangible effects on community members. For instance, as noted by Socia (2013a), increased presence of RSOs in a neighborhood has been associated with higher levels of fear among community members (e.g., Beck & Travis, 2004; Kernsmith, Craun, & Foster, 2009; Zevitz, 2004) and decreased housing values (Larsen, Lowrey, & Coleman, 2003; Linden & Rockoff, 2008; Pope, 2008). Some research suggests this can also result in increased victimization risks for adults (Socia, 2013b; Tewksbury, Mustaine, & Covington, 2010; but see Tewksbury, Mustaine, & Stengel, 2008).¹ However, the majority of sex crimes involve individuals without prior sexual convictions (Sandler, Freeman, & Socia, 2008; Walker & Ervin-McLarty, 2000), and as a group, sex offenders have relatively low rates of sexual recidivism (Hanson & Bussiere, 1998; Harris & Hanson, 2004; Zimring & Leon, 2008). Thus, although proximity to large populations of RSOs may increase the chances of sexual victimization by an RSO, the majority of sexual victimization risk comes from non-RSOs.

In any case, if any of these actual or perceived consequences lead to a migration of some residents out of an area, or reduce levels of informal social control among the remaining residents, deviant place theory suggests that it could result in a continuing cycle of negative consequences for that community (see Stark, 1987). This could be further exacerbated if large concentrations of RSOs overwhelm the resources of social control in a neighborhood, such as halfway houses and parole offices, which in turn could have implications for the successful reentry and rehabilitation of such RSOs. Thus, examining the areas where RSOs are living is an important issue, and findings can help inform on how policies such as residence restrictions, the placement of halfway houses or treatment facilities, resources devoted to parole and reentry, and access to public transportation options can influence successful reentry and promote community safety.

The present study addresses these issues by using multivariate regression models to determine which socio-economic, legal, and spatial characteristics of census tracts are associated with disproportionate concentrations of RSOs. In doing so, this study draws on the prior theory and research regarding RSO residences and social disorganization, housing, residence restrictions, and spatial clustering. It also addresses some of the limitations of previous research, including limited geographic coverage and issues with spatial autocorrelation by focusing on a large geographic area that includes 1,823 census tracts in 53 counties in upstate New York.

Results of this study can be useful both to policymakers and communities in terms of identifying the neighborhoods and neighborhood characteristics associated with the “burdens of sex offenders” (see, Leon, 2011). Furthermore, results can help generate recommendations regarding RSO-related housing policies such as residence restrictions, as well as the placement of resources that can help promote successful reentry and community involvement with the criminal justice system.

Literature Review

Prior studies have focused on a few distinct but related issues concerning the placement of RSO residences. The first issue involves examining the association between structural characteristics and concentrations of RSOs in the community. The second issue involves examining how residence restriction policies have (or might have) affected the types of locations and neighborhood where RSOs are likely to find unrestricted housing (and therefore move to). The third issue involves the influences that neighborhoods can, in turn, have on successful RSO reentry and rehabilitation. Although some studies have examined multiple issues at once (e.g., Hughes & Burchfield, 2008), most have focused only on one.

Structural Characteristics and RSO Housing

Studies focusing on the first issue have identified several structural characteristics associated with RSO residences. These typically include the characteristics of social disorganization (e.g., Hipp, Turner, & Jannetta, 2010; Hughes & Burchfield, 2008; Mustaine & Tewksbury, 2011; Mustaine, Tewksbury, & Stengel, 2006; Socia & Stamatel, 2012; Tewksbury & Mustaine, 2008), available and affordable housing options (Mustaine et al., 2006; Red-Bird, 2009), and high population densities (e.g., Zgoba, Levenson, & McKee, 2009).

These findings suggest that RSOs tend to live in poor, urban neighborhoods that offer housing that is easily obtainable and affordable. This is unsurprising, given the economic and social realities of released felons (see La Vigne, Mamalian, Travis, & Visher, 2003; Solomon, Johnson, Travis, & McBride, 2004; Travis, 2005; Visher, La Vigne, & Travis, 2004). Furthermore, as noted by Socia and Stamatel (2012), disorganized areas may be less able to exert informal social control measures either to keep RSOs from moving in or to force them to move out of the neighborhood.² Thus, for the present study, it seems reasonable to assume that indicators of social disorganization, housing availability and affordability, and population density will be associated with disproportionate concentrations of RSOs.

Residence Restrictions and RSO Housing

Studies focusing on the second issue have frequently found that the presence of residence restriction laws affect where RSOs are able to live in a community. In fact, many studies have explored the spatial implications of residence restrictions as they relate to hypothetical or actual RSO residences (e.g., Barnes, Dukes, Tewksbury, & De Troye, 2009; Grubesic, Mack, & Murray, 2007; Grubesic, Murray, & Mack, 2008; Levenson, 2008; Walker, Golden, & VanHouton, 2001). This research generally finds that residence restrictions disproportionately restrict housing options in urban areas, which may force RSOs to seek housing in more rural areas (see Berenson & Appelbaum, 2011; Casady, 2009; Morgan, 2008; Socia, 2011a; Youstin & Nobles, 2009; Zandbergen & Hart, 2006, 2009a, 2009b). Furthermore, interviews with RSOs have indicated that these policies have forced them out of their previous residences (Levenson & Hern, 2007; Tewksbury & Mustaine, 2009) and into housing that is farther from family members, employment options, and treatment facilities (Levenson & Cotter, 2005; Levenson & Hern, 2007; Mercado, Alvarez, & Levenson, 2008). This is an important issue, because if residence restrictions result in RSOs clustering in certain areas and not others, then these laws may be hindering RSOs’ ability to find housing or access public transportation or treatment facilities (Socia, 2011a), thus affecting their chances of successful rehabilitation.

In fact, the two studies that examined residence restrictions and RSO residences over time found that the implementation of a local residence restriction had decreased the number of RSOs in the area (Morgan, 2008; Youstin & Nobles, 2009). Specifically, these studies, both set in Florida, found that the more rural areas without local-level residence restrictions had experienced increases in the number of RSO residences, whereas the more urban areas with local-level residence restrictions experienced decreases. However, because neither study controlled for structural characteristics related to RSO housing, it remains unclear whether the rural areas contained higher concentrations of RSOs than would otherwise have been expected given their underlying structural characteristics and populations. Furthermore, it is unclear whether and how Florida’s preexisting state-level residence restriction had influenced findings.

The existing research on residence restrictions suggests that these policies may lead RSOs to disperse into more rural and less restricted areas in search of housing. However, a number of studies have found high non-compliance rates with residence restrictions (e.g., Berenson & Appelbaum, 2011; Hughes & Burchfield, 2008; Tewksbury & Mustaine, 2006; Youstin & Nobles, 2009). If non-compliance is the norm, then although residence restrictions can restrict what housing options are legally available to RSOs, they may not be associated with where RSOs are actually living in the community. Furthermore, it may be that the effects of a residence restriction increases (or decreases) the longer it is in place, as compliance or enforcement builds (or wanes). In any case, measuring the presence of residence restrictions, while controlling for structural characteristics influential to RSO housing, can help inform on whether (and how) these policies are actually affecting RSOs.

The Effects of RSO Concentrations

In addition to how policies and structural characteristics can influence RSO housing options, RSO concentrations can in turn influence neighborhood-level processes and capabilities. For example, large RSO populations can create burdens on law enforcement, social services, and treatment providers in certain neighborhoods. That is, law enforcement officials tasked with enforcing such policies may be overburdened with both helping to find unrestricted housing for returning RSOs and monitoring large populations of RSOs already in the community. This would be especially important in areas that have disproportionate concentrations of RSOs compared with the concentration of law enforcement, probation, and/or parole officials who serve this population (Barnes, 2011; Casady, 2009). Furthermore, if residence restrictions result in an increase of homeless RSOs in an area (Levenson, Ackerman, Socia, & Harris, 2013; Socia, Levenson, Ackerman, & Harris, 2014), this could create further burdens on those officials tasked with monitoring these transient populations, as well as inhibiting successful reentry chances.

Large concentrations of RSOs, homeless or otherwise, may also overwhelm the social services available in an area. For example, halfway houses may fill up and be forced to turn some RSOs away. Other informal providers of aid, such as religious institutions, may similarly become overwhelmed with the influx of individuals requesting assistance. Perhaps most importantly, RSO residences can have negative impacts on community members and the social integration of a neighborhood (Zevitz, 2004). As such, it is important to determine where RSOs are concentrating, not only to explain the influence of policies and structural characteristics but also to better understand how RSO concentrations may, in turn, be influencing neighborhoods.

Limitations of Existing Research

The existing literature, although fairly extensive, offers two main areas that could be improved on by future studies. These areas include the limited geographic coverage of individual studies and potential issues with spatial autocorrelation.

Limited geographic coverage of individual studies

Regarding the first area, the majority of studies only examined a small number of counties or cities (e.g., Berenson & Appelbaum, 2011; Chajewski & Mercado, 2009; Levenson & Cotter, 2005; Mack & Grubesic, 2010; Mulford, Wilson, & Parmley, 2009; Mustaine & Tewksbury, 2008). This limitation has led researchers to question the generalizability of their findings (e.g., Berenson & Appelbaum, 2011; Tewksbury & Mustaine, 2008; Zgoba et al., 2009). In fact, only a few studies examined more than six counties (Hipp et al., 2010; Socia, 2011a, 2013a; Tewksbury & Mustaine, 2009; Tewksbury, Mustaine, & Stengel, 2007), and only a single study examined structural characteristics, residence restriction indicators, and RSO residences (Socia, 2013a). Unfortunately, the research of Socia (2013a) only examined the proximity of RSOs to each other, rather than their concentration in the community, and measured structural characteristics using 2000 Census data, which was 10 years prior to the measurement of RSO residences. Thus, more research is needed that includes multiple counties, more recent Census data, as well information on structural characteristics, residence restrictions, and RSO residences in relation to the underlying resident population. This can help generate conclusions that are generalizable to large areas, can supplement the findings of the existing, largely isolated studies, and can identify findings in the existing literature that have widespread applicability.

Issues of spatial autocorrelation

A second area stems from the frequently overlooked issue of spatial autocorrelation. Spatial autocorrelation occurs when observations are correlated with each other based on spatial location. This can be a problem, as it suggests that the concentration of RSOs in one area may be influenced in part by the characteristics of surrounding areas (Socia & Stamatel, 2012). To isolate the association between internal area characteristics and RSO concentrations, one must examine the influence of spatially proximate (i.e., surrounding) areas and control for this influence if it is significant. This has been accomplished in prior studies by the inclusion of a spatial lag measure of the dependent variable for each area (e.g., Hughes & Kadleck, 2008; Socia & Stamatel, 2012).

The Present Study

The present study addresses these limitations and builds on the prior literature in its examination of the census tract characteristics that are associated with the spatial distribution of RSOs in upstate New York. It utilizes data on RSO residences as well as structural indicators of disorganization, housing affordability and availability, and population density taken from recent Census data. It also considers residence restrictions implemented at the county and local level, the influence of spatial autocorrelation, and the regional differences of upstate New York.

Sample

The final sample consisted of 1,823 Census tracts across 53 counties in upstate New York.³ Upstate New York was considered to consist of all counties outside the New York Metropolitan Statistical Area (NYMSA), which is essentially the New York City area (including Long Island). However, Putnam County, although technically in the NYMSA, was included in the sample as it is more typical of upstate New York in terms of population and parcel density than the other NYMSA counties (Socia, 2011a).

Although upstate New York has been examined in prior research (e.g., Socia, 2011a, 2011b, 2012a, 2012b, 2013a, 2013b), continuing to study this area is still useful for a number of reasons. First, the upstate New York area offers a wide geographic coverage and diverse demographic, social, and geographic characteristics, as well as a mixture of counties and local-level jurisdictions with and without residence restrictions (Socia, 2013a). Second, excluding the vast number of extremely dense tracts in NYMSA counties that lie in and around New York City means that the results of the analyses will be more applicable to the “average” tract found throughout most of the state (and throughout much of the United States; Socia, 2011a). As such, although direct generalizability to other states is always subject to debate, the similarity between the density of upstate New York counties and those scattered throughout the United States makes such a comparison more tenable than if the study focused on New York City or a similarly densely packed urban area. Finally, although prior studies have examined upstate New York in terms of the effects of potential residence restrictions on RSO housing options (e.g., Socia, 2011a), or the location of RSOs in relation to existing residence restrictions (e.g., Berenson & Appelbaum, 2011) or other RSOs (e.g., Socia, 2013a), no research has examined this area using both structural characteristics and the concentrations of actual RSOs compared with the distribution of residents. Thus, this study extends the prior research conducted by Berenson and Appelbaum (2011) and Socia (2011a, 2013a) by examining the association between structural and legal characteristics and actual RSO distributions across Census tracts in the upstate New York area.

Dependent Variable

The dependent variable measured the concentration of RSO residences across Census tracts, and was calculated as an excess risk score (see Grubesic, 2010).⁴ Excess risk compares the actual rate of RSOs per resident in a tract to the expected rate of RSOs per resident.⁵ The expected rate of RSOs assumes a homogeneous distribution of RSOs per resident within a county and is calculated using the actual distribution of residents across tracts in a county.⁶ Thus, excess risk indicates which tracts in a county have disproportionately more (or less) RSOs per resident than would otherwise be expected in the county, given how many residents actually live in those tracts.⁷

Excess risk was calculated for each tract using the following formula:

\begin{array}{l} ER = \frac{RS O_{n}}{PO P_{n}} - \frac{RS O_{c}}{PO P_{c}}, \end{array}

Where $ER$ is the excess risk score, $RS O_{n}$ is the number of RSOs in a tract, $RS O_{c}$ is the number of RSOs in the county, $PO P_{n}$ is the resident population of the tract (in 10,000s), and $PO P_{c}$ is the resident population of the county (in 10,000s). Values of excess risk above zero indicate a tract that has disproportionately more RSOs per capita than expected, whereas values below zero indicate a tract that has disproportionately less.⁸

Excess risk has some important qualities that make it more attractive than other measures of RSO concentrations. First, it is geographically specific, particularly compared with more general measurements of clustering such as Moran’s I. That is, excess risk identifies the level of clustering for an individual tract rather than for a group of tracts. Second, excess risk accounts for differences in the underlying distribution of residents across tracts in a county, regardless of spatial location. This differs from measures of clustering like Local Indicators of Spatial Association (LISA) that considers clustering in terms of similarities and differences between spatially proximate areas. This measure also helps control for unmeasured tract characteristics that are associated with residents’ housing decisions but are not specific to RSOs’ housing decisions. These qualities are important, as they allow for tract-level comparisons between different counties, despite the potential for dramatically different geographic or resident characteristics from county to county.⁹

As the original distribution of the excess risk variable was highly skewed, a log transformation was used to ensure a more normal distribution. This involved offsetting all excess risk scores by 2 (to ensure all positive values) and then logging the result. This transformation resulted in a more normal distribution. Furthermore, because this study uses census tracts, which are situated within towns and cities, and within counties, it is possible that errors be geographically clustered. As such, robust models errors were estimated using a cluster option based on the county (Rogers, 1993; Wooldridge, 2001).

Independent Variables

Independent variables were selected for inclusion based on their theoretical connections to RSO housing. As noted earlier, research has found that RSO residences are frequently found in disorganized urban areas offering available and affordable housing options. Furthermore, RSO residences may also be influenced by residence restrictions. Although not as commonly examined, RSO residences have also been found to be spatially clustered, suggesting spatial autocorrelation may be important. Finally, due to differences between different regions in such a large area (upstate New York), regional indicators were included. The measurements of these characteristics and the expected relationship to excess risk are examined below.

Structural characteristics

Social disorganization

The structural characteristics of social disorganization typically include measures of concentrated disadvantage, residential instability, and ethnic heterogeneity. This study constructed these structural disorganization measures using factor analyses of U.S. Census Bureau (2010) data in a manner largely consistent with other neighborhood-level studies (e.g., Sampson, Raudenbush, & Earls, 1997; Socia, 2011a, 2013a; Socia & Stamatel, 2012). Concentrated disadvantage was estimated using a factor score based on percent living in poverty, percent unemployed, percent female heads of household with children, and percent non-Hispanic Black residents. Residential instability was estimated using a factor score based on percent of owner-occupied homes and percent of residents 5 years and older who have lived in the same house for at least 5 years. This original factor score was reverse coded so that a higher value indicates more instability. Ethnic heterogeneity was estimated using a factor score based on percent Hispanic residents and percent foreign-born residents.¹⁰ Given the connection between disorganization and RSO residences, it is expected that all three measures will be positively associated with excess risk.

Housing availability

Housing availability was measured as the percent of vacant housing units for rent or sale in a tract, as reported by the U.S. Census Bureau (2010).¹¹ Given the housing difficulties faced by ex-felons, it is expected that housing availability will be positively associated to excess risk.¹²

Housing affordability

Housing affordability was measured as the percent difference between the tract’s median gross rent in 2010, as reported by the U.S. Census Bureau (2010), and the fair market rent for a two-bedroom apartment in 2010 in the associated housing market, as reported by the Office of Housing and Urban Development (HUD; 2010). This measurement allowed for comparisons of housing affordability across different housing markets and has precedence in the existing literature (see Socia, 2011a). For intuitive purposes, the measure was reverse coded so that a higher value indicates a more affordable tract.¹³ After reverse coding, and given the aforementioned housing and employment difficulties faced by ex-felons, it is expected that housing affordability will be positively associated to excess risk.

Population density

Population density was measured as the number of residents (in 10,000s) per square mile, as reported by the U.S. Census Bureau (2010). As RSOs are frequently found in densely packed urban areas, it is expected that population density will be positively associated to excess risk.

Residence restriction indicators

A number of residence restriction measures were included in the model to test the association between differences in these policies and excess risk. Each of these measures is explained in more detail below.

Residence restriction months

The length of time a residence restriction had applied to the tract, as of September 2010, was measured in months. Although it is assumed that the presence of these policies will be associated with excess risk due to the reduction of legal housing options, high non-compliance rates may result in no overall association.¹⁴

Residence restriction size and scope

Three dichotomous variables were included to control for different types of residence restrictions. The first two variables indicated whether the size of the largest residence restriction applying to the tract was small (1,000 feet or less) or large (greater than 1,000 feet). The third variable indicated whether the tract was subject to a residence restriction with a comprehensive scope (i.e., one that included child congregation locations other than schools, daycares, parks, and playgrounds). Due to the restriction of more housing options, it is expected that the presence of either a large or a comprehensive residence restriction will be negatively associated with excess risk.

Nearby residence restriction

A dichotomous variable measured, for tracts that did not have residence restriction of their own, whether there was a spatially proximate tract that was subject to a residence restriction. If RSOs are being pushed out of areas with such restrictions and into less restricted areas (Nobles, Levenson, & Youstin, 2012), then a nearby residence restriction should be positively associated with excess risk.

Spatial autocorrelation

Spatial lag

As expected, an exploratory analysis of the spatial distribution of excess risk found the presence of positive spatial autocorrelation, as indicated by a Moran’s I value of .42 (p < .001). This spatial autocorrelation must be accounted for to isolate the relationship between an individual tract’s characteristics and its excess risk value. Therefore, a measure of the spatial lag of the dependent variable was included in the analysis. Spatial lag was calculated on the untransformed excess risk value using a first-order Queen’s contiguity weight. In essence, this measured, for each tract, the average tract excess risk value for all of the spatially proximate tracts. The spatial lag measure was transformed in a similar manner as the excess risk measure.

Regional controls

Upstate New York covers a large geographic area that has important regional differences, differences that might bias conclusions if they are not accounted for. For instance, counties in some regions of the state are extremely rural (e.g., The Adirondacks), whereas others, particularly closer to the New York City area, are much more urban (e.g., Hudson Valley). To control for any underlying unmeasured differences between tracts in different regions of the state, nine unique regions were identified from New York tourism web sites. The regions identified were as follows: Chautauqua-Allegheny, Niagara Frontier, Finger Lakes, Thousand Islands-Seaway, The Adirondacks, Central Leather-Stocking, Capital-Saratoga, The Catskills, and Hudson Valley.¹⁵ A list of the counties contained in each region is available from the author. Each region had a unique indicator variable assigned to it, and eight of these variables were included in the final model. The ninth indicator (Hudson Valley) was excluded for comparison purposes, and as it was the region closest to New York City, it was thought to be the region most different from the other regions in the upstate area.

Analytical Models

If ordinary least squares (OLS) was used to estimate the models, the inclusion of a spatially lagged dependent variable would have resulted in endogeneity (i.e., the lag term being correlated with the error term), and that could have biased estimates (Anselin & Kelejian, 1997; Fingleton & Le Gallo, 2008; Franzese & Hays, 2007). Thus, this study utilized a two stage least squares (2SLS) model to predict excess risk, with instrumental variables included to predict the (endogenous) spatial lag of the dependent variable (see Anselin & Kelejian, 1997). The instrumental variables were measured as the spatial lag of the three social disorganization factors in the model (e.g., lagged concentrated disadvantage, lagged residential instability, and lagged ethnic heterogeneity). This technique should help control for the endogeneity stemming from the inclusion of a spatial lag of the dependent variable.¹⁶

As such, the analyses used three iterative 2SLS regression models with instrumental variables.¹⁷ The first model included the structural characteristics of tracts as well as measures of residence restriction differences. The second model added the spatial autocorrelation (lag) measure (and the associated instrumental variables), and the third model added the regional indicators. In all models, clustered standard errors (based on county) were estimated to account for any potential problems with model assumptions or specification.

Results

Descriptive statistics are presented in Table 1. Analysis of the correlation matrix (available from the author) indicated that some of the variables had strong bivariate correlations (r > .60), but none had correlations above .80, which is the general rule of thumb for potential issues with multicollinearity (Allison, 1998; Knoke, Bohrnstedt, & Mee, 2002). The variance inflation factor (VIF) scores for each model indicated that multicollinearity was not a concern. After the transformation of the dependent variable, residuals from all three models were normally distributed and were uncorrelated with both the predicted values and the independent variables.

Table 1.

Descriptive Statistics.

Variable	Observations	M	SD	Minimum	Maximum
Excess risk^a	1,823	2.30	1.04	0.69	6.39
Structural characteristics
Concentrated disadvantage	1,823	0.01	0.89	−1.14	4.91
Residential instability	1,823	0.00	0.89	−1.80	3.90
Ethnic heterogeneity	1,823	0.00	0.68	−0.68	5.31
Housing availability	1,823	3.59	2.49	0.00	21.06
Housing affordability	1,823	7.75	23.30	−151.06	100.00
Population density	1,823	0.27	0.36	0.00	3.29
RR indicators
Months of RR	1,823	14.86	20.75	0	57
RR size 1,000’ or less	1,823	0.08	0.27	0	1
RR size more than 1,000’	1,823	0.22	0.42	0	1
RR comprehensive scope	1,823	0.10	0.30	0	1
Nearby RR	1,823	0.09	0.29	0	1
Spatial autocorrelation
Spatial lag^a	1,823	2.54	0.75	0.69	5.38
Regional controls
Chautauqua-Allegheny	1,823	0.04	0.19	0	1
Niagara Frontier	1,823	0.18	0.38	0	1
Finger Lakes	1,823	0.29	0.45	0	1
Thousand Islands-Seaway	1,823	0.04	0.20	0	1
The Adirondacks	1,823	0.06	0.24	0	1
Central Leather-Stocking	1,823	0.11	0.31	0	1
Capital-Saratoga	1,823	0.12	0.33	0	1
The Catskills	1,823	0.05	0.23	0	1
Hudson Valley	1,823	0.11	0.31	0	1

Note. RR = residence restriction.

Variable has been transformed by increasing the original values by 2 and taking the log of the result.

Results are presented as unstandardized coefficients in Table 2, with standard deviations given in parentheses. Given the log transformation of the dependent variable, the unstandardized coefficients, when multiplied by 100, can be interpreted as the percent change in the dependent variable given a one-unit change in the independent variable. Results are discussed in more detail below, with an emphasis on the full model (Model 3, Table 2).

Table 2.

Characteristics Associated with Excess Risk.

	Excess risk^a
	Model 1	Model 2	Model 3
Variable	b	b	b
Structural characteristics
Concentrated disadvantage	0.51***	0.33***	0.34***
	(0.08)	(0.05)	(0.05)
Residential instability	0.04	−0.03	−0.04
	(0.05)	(0.06)	(0.06)
Ethnic heterogeneity	−0.27***	−0.15**	−0.11*
	(0.08)	(0.05)	(0.05)
Housing availability	0.07**	0.05**	0.06**
	(0.02)	(0.02)	(0.02)
Housing affordability	<0.01***	<0.01**	<0.01**
	(<0.01)	(<0.01)	(<0.01)
Population density	−0.14	−0.15	−0.13
	(0.14)	(0.11)	(0.11)
RR indicators
Months of RR	<0.01	<0.01	>−0.01
	(<0.01)	(<0.01)	(<0.01)
RR size 1,000’ or less	<0.01	0.03	0.15
	(0.16)	(0.08)	(0.10)
RR size more than 1,000’	−0.11	−0.07	−0.09
	(0.09)	(0.05)	(0.09)
RR comprehensive scope	−0.32**	−0.13	−0.14
	(0.12)	(0.10)	(0.09)
Only Nearby RR	0.18	0.10	0.08
	(0.11)	(0.09)	(0.08)
Spatial autocorrelation
Spatial lag^a		0.50***	0.44***
		(0.08)	(0.08)
Regional controls^b
Chautauqua-Allegheny			0.39*
			(0.17)
Niagara Frontier			0.19
			(0.14)
Finger Lakes			0.15
			(0.14)
Thousand Islands-Seaway			0.27
			(0.19)
The Adirondacks			0.32
			(0.17)
Central Leather-Stocking			0.34*
			(0.15)
Capital-Saratoga			0.27
			(0.15)
The Catskills			0.12
			(0.12)
Constant	2.10***	0.83***	0.81***
	(0.15)	(0.20)	(0.17)
N	1,823	1,823	1,823
R ²	.30	.41	.42

Note. Models present robust standard errors (clustered by County) in parentheses. RR = Residence restriction.

Variable has been transformed by increasing the original values by 2 and taking the log of the result.

Reference category is Hudson Valley.

p < .05. **p < .01. ***p < .001 (two-tailed).

Structural Characteristics

In the full model, a one-unit increase in concentrated disadvantage was associated with a 34% increase in excess risk and was significant at an alpha level of .001 (Model 3, Table 2).¹⁸ That is, a one-unit increase in disadvantage was associated with 34% more RSOs per capita than was otherwise expected given the population of the tract and the number of RSOs and residents in the county. Residential instability was negative but did not achieve significance in the full model. A one-unit increase in ethnic heterogeneity was associated with an 11% decrease in excess risk and was significant at an alpha level of .05.

In the full model, both housing availability and affordability were significantly (p < .01 for both) associated with excess risk (Model 3, Table 2). A 1% increase in the amount of available housing was associated with a 6% increase in excess risk, whereas a 1% increase in housing affordability was associated with a 0.3% increase in excess risk.¹⁹ Population density did not achieve significance in any model.

Residence Restriction Indicators

In the full model, none of the measures of residence restrictions was significant. Furthermore, the only coefficient to achieve significance was that of a comprehensive residence restriction scope, and this was only significant in the first model. Once spatial lag was included, scope became non-significant.

Spatial Autocorrelation

Spatial autocorrelation (lag) was significant (p < .001) and positively associated with excess risk in both the second and third models (Table 2). The only coefficient that changed substantially when spatial lag was added was the presence of a comprehensive residence restriction, which became non-significant, as noted earlier. This suggests that the inclusion of spatial lag may have produced somewhat better estimates overall, but its exclusion, at least in the present study, would probably not have influenced overall conclusions regarding most other tract characteristics.

Regional Controls

As noted earlier, Hudson Valley was excluded from the model to serve as a comparison with the other regions of upstate New York. The regional controls were non-significant when added as a group (F = 1.17, p = .33). Although all region coefficients were negative (suggesting tracts had lower excess risk scores compared with the Hudson Valley tracts), only the Chautauqua-Allegheny region reached significance at an alpha level of .05 (Model 3, Table 2). The addition of the region coefficients did not substantially change other coefficient estimates.

Discussion

This study found that the tracts exhibiting more concentrated disadvantage or containing more available or affordable housing had higher excess risk scores (i.e., disproportionately more RSOs than was otherwise expected within the county). These findings support much of the prior literature (e.g., Hughes & Kadleck, 2008; Mustaine & Tewksbury, 2011; Tewksbury & Mustaine, 2008) and suggest that RSOs end up disproportionately living in disadvantaged urban areas, likely because of economic and social hardships. It also underscores that access to viable housing options is a key concern for ex-felons returning to the community (La Vigne et al., 2003; Solomon et al., 2004; Travis, 2005; Visher et al., 2004). Furthermore, it suggests that the areas least likely to exert informal social control over their residents may be burdened with the highest concentrations of RSOs. This can, in turn, have negative influences on the social cohesion of such areas (see Zevitz, 2004). Finally, these areas may also lack the formal social control and treatment resources to monitor and serve the disproportionate RSO populations.

One interesting finding was that tracts that were more ethnically heterogeneous had lower excess risk scores and thus fewer RSOs per capita. This supports the findings of Socia and Stamatel (2012), and may suggest that RSOs are not drawn to ethnically heterogeneous areas in search of housing, at least in upstate New York. More research on individual RSO decision making, or research that examines RSOs separately by race or ethnicity, could be useful in determining the unique influence of ethnic heterogeneity.

In contrast to prior research, population density was not associated with excess risk. This may be because where RSOs tend to live depends less on density and more on disorganization and housing conditions. That is, the most densely packed areas in upstate New York are typically those that are the most disorganized and offer the most available and affordable housing (Socia, 2011a). As a result, these findings suggest that RSOs do not live in densely packed urban neighborhoods because of this density but rather live in neighborhoods that exhibit characteristics that are frequently found in densely packed neighborhoods, namely, disadvantage and housing availability and affordability. This is important, because it indicates that the characteristics that draw (or allow) RSOs into certain neighborhood may be similar to those that draw (or allow) other disadvantaged populations into those same neighborhoods.

Interestingly, none of the residence restriction variables was significant in the final model. One reason for this may be that RSOs are frequently grandfathered (formally or informally) from compliance until their next residential move, and thus the effect of residence restrictions may be both relatively small and take a long time to present itself. Furthermore, when one considers that high non-compliance rates are common (Berenson & Appelbaum, 2011; Hughes & Burchfield, 2008; Tewksbury & Mustaine, 2006; Youstin & Nobles, 2009), this finding is easier to understand. This finding is also in contrast to other research that finds RSOs move out from areas with newer residence restrictions (Morgan, 2008; Youstin & Nobles, 2009) but does not contradict research showing any effects on RSO housing may be time limited (Socia, 2013a).

Because the size and scope of a residence restriction was not associated with excess risk, RSOs did not appear to be moving away from areas with more severe residence restrictions, at least when examined at the tract level. Given the long time it may take for a residence restriction to sizably affect RSO populations in a tract (or county), the difference in the size or scope of a residence restriction may be of little importance compared with other influences such as housing options.

Furthermore, the presence of a nearby residence restriction was not associated with excess risk in tracts that were not subject to their own policies. This may suggest that RSOs are not simply moving out of areas with any residence restrictions to the nearest “unrestricted” tract they can find. This means that one of the reasons used to justify the passage of residence restrictions, that communities without such policies would become “dumping groups” for sex offenders, may not necessarily be accurate (see Levenson, 2009; Socia, 2012b; Yung, 2007). That is, even with cross-sectional data, if RSOs migrated to areas without residence restrictions, there would at least be a negative association between the presence of these policies and clusters of RSOs. That was not the case in the present study. However, longitudinal data would be required to confirm the lack of RSO movement into non-residence restriction areas over time.

The regional controls indicated that tracts in the Hudson Valley region (the comparison group) consistently had higher levels of excess risk than tracts in other regions of upstate New York. However, only the Chautauqua-Allegheny region was significantly different at an alpha level of .05. Given that the Hudson Valley region is the closest region to the New York City area, and likely looks more like that area than any other region in the state, it suggests that RSOs may be disproportionately living closer to New York City, but this effect was not enough to reach significance for most regions in New York. This also suggests that there may be some unique mechanisms for how the underlying, unmeasured regional characteristics could be influencing RSO residences, and controlling for these regional differences may be of some use in certain cases. However, given the non-significance of most of the region coefficients, and their non-significance as a group, this seems to be of limited benefit for the current study.

Future Research

There are a number of future research ideas that stem from this study. First, research is needed that combines data on the compliance rates of residence restrictions (e.g., Berenson & Appelbaum, 2011) with data on RSO movement over time (e.g., Youstin & Nobles, 2009). This would help determine whether and how residence restrictions affect the spatial distribution of RSOs, particularly in light of the contrasting findings of the present study. Second, future research on RSO residences should consider measuring and controlling for spatial autocorrelation and/or regional differences, particularly when examining very large geographic areas. Including these controls may help better identify relationships between internal tract characteristics and RSO residences by controlling for external influences. Finally, more research is needed that examines the individual-level decision-making process of RSOs when they seek out housing, particularly in relation to structural and housing characteristics, or that examines RSOs separately by race and ethnicity.

Limitations

Perhaps the most important limitation of this study is its cross-sectional design. Although methods were used to help control for different exposure times to residence restrictions, RSO residences were still measured at a single point in time. Thus, although this study can and has identified significant (and non-significant) associations between tract characteristics and RSO residences, and has suggested potential causes for these associations, it cannot confirm the causality of these associations per se.

Another limitation concerns the generalizability of these findings to areas outside upstate New York. Although upstate New York provides a broad geographic area for consideration, the extent that these findings are generalizable to much more urban areas like New York City is unclear. This is particularly true given the consistent (albeit frequently non-significant) difference between the Hudson Valley region and the rest of upstate New York. Yet, the findings of the present study are very similar to those of prior studies set in other areas of the country, such as Chicago (Socia & Stamatel, 2012) and California (Hipp et al., 2010). Thus, although these findings may not necessarily be generalizable per se, they suggest consistent associations are present between certain neighborhood characteristics and RSO residences, and add to the existing body of research exploring RSO residences.

A further limitation concerns the measurement of residence restrictions. That is, none of the residence restriction measurements achieved significance in the model. This could be for one of two reasons. First, it may be that residence restrictions simply are not as useful in determining RSO clustering compared with other neighborhood characteristics. However, it may instead (or also) be because the methods used to measure residence restrictions were not nuanced enough to detect significant effects. Due to the variation between the sizes and scopes of residence restrictions, it may be that more detailed measurements of residence restrictions’ coverage are needed. This should be considered for future studies when determining how to measure such policies.

Conclusion and Implications

This study provides a number of conclusions and implications regarding RSO residences that may be of interest to policymakers and communities. In confirmation with much of the existing literature, RSOs disproportionately lived in tracts that exhibited concentrated disadvantage and that offered available and affordable housing options. This suggests that RSOs may be drawn into disadvantaged areas due to social and/or economic concerns, particularly as they relate to housing. This finding is particularly salient when one considers the influence that RSOs’ concentrations can have on fear levels, housing prices, and potentially overburdened formal and informal social control mechanisms in neighborhoods. Perhaps the most interesting finding was that the presence, size, or scope of a residence restriction was not significantly related to rates of RSOs in the community. Furthermore, nearby residence restrictions did not appear to be associated with RSOs living in nearby, unrestricted areas. Thus, there appear to be other, more important mechanisms at play that influence where RSOs are most likely to locate in a community.

Although the finding that RSOs disproportionately live in disadvantaged areas with available and affordable housing is not unique, its value lies in supporting the conclusions from other studies set in other geographic areas. Thus, just because a study on RSO residences examines a particular state, county, or city does not mean its findings are irrelevant to policymakers in other areas, particularly when studies from multiple locations come to the same general conclusions.

However, the findings of this (and related) studies are only useful to the extent that jurisdictions, policymakers, correctional organizations, and treatment providers utilize this information to make future policy decisions. For example, this knowledge can be used to help target policy interventions, facilities, and reentry resources in the areas expected to host disproportionate rates of RSOs. Thus, the placement of social service offices, halfway houses, treatment facilities, and public transportation options in the most disadvantaged neighborhoods that offer available and affordable housing options for ex-offenders may lead to a greater utilization by RSOs and, thus, to greater chances of reentry success.

Although policymakers may want to alleviate the burden or inequity involved in having disproportionately more RSOs in certain areas, they also need to ensure that any policy interventions are having their intended effects. That is, interventions that help to more equitably relocate RSOs in a community based on available resources may also help alleviate levels of fear, increase housing values, and/or promote more successful reentry (see Grubesic & Murray, 2008; Socia, 2013b). Unfortunately, this goal may be undermined if local communities continue to pass policies that are meant only to keep RSOs from living within their jurisdictions and that do not address the reentry needs of those RSOs who do live within their jurisdictions. This problem is compounded when policies treat RSOs as a one homogeneous “high risk” group, and when questioning RSO policies is seen as political suicide (see Leon, 2011; Socia, 2014). In the end, policymakers must be willing to use the steadily growing body of research to inform future policy decisions that balance pressure from their constituents with the benefits of the larger community.

Footnotes

Acknowledgements

The author would like to thank the following individuals for their helpful comments on earlier versions of the article: Kirstin Morgan, Christopher Dum, Russell Lundberg, Andrew Wheeler, too many anonymous reviewers, and Drs. Elizabeth Brown, Robert Apel, Jill Levenson, Alan Lizotte, Steven Messner, Greg Pogarsky, Aki Roberts, Janet Stamatel, and Paul Tracy.

Author’s Note

Data for the New York State Sex Offender Registry were supplied by the New York State Division of Criminal Justice Services (DCJS). The opinions, findings, and conclusions or recommendations expressed in this publication are those of the author and do not necessarily reflect those of either the Department of Justice or the New York State DCJS. Neither New York State nor DCJS assumes liability for its contents or use thereof.

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 disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was supported by Award No. 2010-IJ-CX-0004, awarded by the National Institute of Justice, Office of Justice Programs, U.S. Department of Justice.

Notes

Author Biography

Kelly M. Socia is an assistant professor in the School of Criminology and Justice Studies at the University of Massachusetts, Lowell. He received his PhD in criminal justice from the School of Criminal Justice at the University at Albany, State University of New York. His research interests include offender reentry and recidivism, registered sex offenders (RSO), public policymaking, geographic information systems, and spatial analyses.

References

Allison

P. D.

(1998). Multiple regression: A primer (1st ed.). Thousand Oaks, CA: Pine Forge Press.

Anselin

Kelejian

H. H.

(1997). Testing for spatial error autocorrelation in the presence of endogenous regressors. International Regional Science Review, 20, 153-182.

Barnes

J. C.

(2011). Place a moratorium on the passage of sex offender residence restriction laws. Criminology & Public Policy, 10, 401-409. doi:10.1111/j.1745-9133.2011.00715.x

Barnes

J. C.

Dukes

Tewksbury

De Troye

T. M.

(2009). Analyzing the impact of a statewide residence restriction law on South Carolina sex offenders. Criminal Justice Policy Review, 20, 21-43. doi:10.1177/0887403408320842

Beck

V. S.

Travis

L. F.

(2004). Sex offender notification and fear of victimization. Journal of Criminal Justice, 32, 455-463. doi:10.1016/j.jcrimjus.2004.06.007

Bell

(1954). A probability model for the measurement of ecological segregation. Social Forces, 32, 357-364. doi:10.2307/2574118

Berenson

J. A.

Appelbaum

P. S.

(2011). A geospatial analysis of the impact of sex offender residency restrictions in two New York Counties. Law and Human Behavior, 35, 235-246. doi:10.1007/s10979-010-9235-3

Best

(1990). Threatened children. Chicago, IL: University of Chicago Press.

Brantingham

P. J.

Brantingham

P. L.

(1981). Environmental criminology. Beverly Hills, CA: Sage.

10.

Casady

(2009). A police chief’s viewpoint: Geographic aspects of sex offender residency restrictions. Criminal Justice Policy Review, 20, 16-20. doi:10.1177/0887403408327692

11.

Chajewski

Mercado

C. C.

(2009). An evaluation of sex offender residency restriction functioning in town, county, and city-wide jurisdictions. Criminal Justice Policy Review, 20, 44-61. doi:10.1177/0887403408320845

12.

Fingleton

Le Gallo

(2008). Estimating spatial models with endogenous variables, a spatial lag and spatially dependent disturbances: Finite sample properties. Papers in Regional Science, 87, 319-339.

13.

Franzese

R. J.

Jr. Hays

J. C.

(2007). Spatial econometric models of cross-sectional interdependence in political science panel and time-series-cross-section data. Political Analysis, 15, 140-164.

14.

Garland

(2001). The culture of control: Crime and social order in contemporary society. Chicago, IL: University of Chicago Press.

15.

Greenfield

L. A.

(1997). Sex offenses and offenders: An analysis of data on rape and sexual assault (Bureau of Justice Statistics, Trans.). Washington, DC: National Institute of Justice.

16.

Grubesic

T. H.

(2010). Sex offender clusters. Applied Geography, 30, 2-18. doi:10.1016/j.apgeog.2009.06.002

17.

Grubesic

T. H.

Mack

Murray

A. T.

(2007). Geographic exclusion: Spatial analysis for evaluating the implications of Megan’s Law. Social Science Computer Review, 25, 143-162. doi:10.1177/0894439307298930

18.

Grubesic

T. H.

Murray

A. T.

(2008). Sex offender residency and spatial equity. Applied Spatial Analysis and Policy, 1, 175-192. doi:10.1007/s12061-008-9013-5

19.

Grubesic

T. H.

Murray

A. T.

Mack

(2008). Sex offenders, housing and spatial restriction zones. GeoJournal, 73, 255-269. doi:10.1007/s10708-008-9197-x

20.

Hanson

R. K.

Bussiere

M. T.

(1998). Predicting relapse: A meta-analysis of sexual offender recidivism studies. Journal of Consulting and Clinical Psychology, 66, 348-362. doi:10.1037/0022-006X.66.2.348

21.

Harris

A. J. R.

Hanson

R. K.

(2004). Sex offender recidivism: A simple question. Ottawa, Ontario: Solicitor General of Canada.

22.

Hipp

J. R.

Turner

Jannetta

(2010). Are sex offenders moving into social disorganization? Analyzing the residential mobility of California parolees. Journal of Research in Crime & Delinquency, 47, 558-590. doi:10.1177/0022427810381093

23.

Housing and Urban Development. (2010). Fair market rents. Retrieved from http://www.huduser.org/portal/datasets/fmr.html

24.

Hughes

L. A.

Burchfield

K. B.

(2008). Sex offender residence restrictions in Chicago: An environmental injustice? Justice Quarterly, 25, 647-673. doi:10.1080/07418820802119976

25.

Hughes

L. A.

Kadleck

(2008). Sex offender community notification and community stratification. Justice Quarterly, 25, 469-495. doi:10.1080/07418820701710941

26.

Kernsmith

P. D.

Craun

S. W.

Foster

(2009). Public attitudes toward sexual offenders and sex offender registration. Journal of Child Sexual Abuse, 18, 290-301. doi:10.1080/10538710902901663

27.

Knoke

Bohrnstedt

G. W.

Mee

A. P.

(2002). Statistics for social data analysis (4th ed.). Itasca, IL: F.E. Peacock Publishers.

28.

Larsen

J. E.

Lowrey

K. J.

Coleman

J. W.

(2003). The effect of proximity to a registered sex offender’s residence on single-family house selling price. Appraisal Journal, 71, 253-265.

29.

La Vigne

N. G.

Mamalian

C. A.

Travis

Visher

C. A.

(2003). A portrait of prisoner reentry in Illinois ( Center

J.P.

, Trans.). Washington, DC: Urban Institute.

30.

Leon

(2011). The contexts and politics of evidence-based sex offender policy. Criminology & Public Policy, 10, 421-430. doi:10.1111/j.1745-9133.2011.00717.x

31.

Levenson

J. S.

(2008). Collateral consequences of sex offender residence restrictions. Criminal Justice Studies, 21, 153-166. doi:10.1080/14786010802159822

32.

Levenson

J. S.

(2009). Sex offender residence restrictions. In Wright

R. G.

(Ed.), Sex offender laws: Failed policies, new directions (pp. 267-290). New York, NY: Springer.

33.

Levenson

J. S.

Ackerman

A. R.

Socia

K. M.

Harris

A. J.

(2013). Where for art thou? Transient sex offenders and residence restrictions. Criminal Justice Policy Review. doi:10.1177/0887403413512326

34.

Levenson

J. S.

Cotter

L. P.

(2005). The impact of sex offender residence restrictions: 1,000 feet from danger, or one step from absurd? International Journal of Offender Therapy and Comparative Criminology, 49, 168-178. doi:10.1177/0306624X04271304

35.

Levenson

J. S.

Hern

A. L.

(2007). Sex offender residence restrictions: Unintended consequences and community reentry. Justice Research and Policy, 9, 59-73. doi:10.3818/JRP.9.1.2007.59

36.

Lieberson

Carter

D. K.

(1982). Temporal changes and urban differences in residential segregation: A reconsideration. The American Journal of Sociology, 88, 296-310. doi:10.1086/227672

37.

Linden

L. L.

Rockoff

J. E.

(2008). Estimates of the impact of crime risk on property values from Megan’s Law. The American Economic Review, 98, 1103-1127. doi:10.1257/aer.98.3.1103

38.

Mack

Grubesic

T. H.

(2010). Sex offenders and residential location: a predictive–analytical framework. Environment and Planning A, 42, 1925-1942. doi:10.1068/a42370

39.

Maltz

M. D.

(1994). Deviating from the mean: The declining significance of significance. Journal of Research in Crime & Delinquency, 31, 434-463. doi:10.1177/0022427894031004005

40.

Mercado

C. C.

Alvarez

Levenson

J. S.

(2008). The impact of specialized sex offender legislation on community reentry. Sexual Abuse: A Journal of Research and Treatment, 20, 188-205. doi:10.1177/1079063208317540

41.

Morgan

J. K.

(2008). Spatial implications of municipal sex offender residency restrictions in Bay County, Florida, 2005-2007 (Master of Science). Northwest Missouri State University, Maryville. Retrieved from http://www.nwmissouri.edu/LIBRARY/theses/MorganJenniferK/jmorgan_thesis.pdf

42.

Mulford

C. F.

Wilson

R. E.

Parmley

A. M.

(2009). Geographic aspects of sex offender residency restrictions: Policy and research. Criminal Justice Policy Review, 20, 3-12. doi:10.1177/0887403408327683

43.

Mustaine

E. E.

Tewksbury

(2008). Registered sex offenders, residence, and the influence of race. Journal of Ethnicity in Criminal Justice, 6, 65-82. doi:10.1300/J222v06n01_05

44.

Mustaine

E. E.

Tewksbury

(2011). Residential relegation of registered sex offenders. American Journal of Criminal Justice, 36, 44-57. doi:10.1007/s12103-010-9102-2

45.

Mustaine

E. E.

Tewksbury

Stengel

K. M.

(2006). Social disorganization and residential locations of registered sex offenders: Is this a collateral consequence? Deviant Behavior, 27, 329-350. doi:10.1080/01639620600605606

46.

Nobles

M. R.

Levenson

J. S.

Youstin

T. J.

(2012). Effectiveness of residence restrictions in preventing sex offender recidivism. Crime & Delinquency, 58, 491-513.

47.

Pope

J. C.

(2008). Fear of crime and housing prices: Household reactions to sex offender registries. Journal of Urban Economics, 64, 601-614. doi:10.1016/j.jue.2008.07.001

48.

Poston

D. L.

Micklin

(2006). Handbook of population. New York, NY: Springer.

49.

Red-Bird

(2009). Assessing housing availability under Ohio’s sex offender residency restrictions. Columbus: The Ohio State University.

50.

Rengert

G. F.

Piquero

A. R.

Jones

P. R.

(1999). Distance decay reexamined. Criminology, 37, 427-446. doi:10.1111/j.1745-9125.1999.tb00492.x

51.

Rogers

(1993). Regression standard errors in clustered samples. Stata Technical Bulletin, 13, 19-23.

52.

Rossmo

D. K.

(1995). Overview: Multivariate spatial profiles as a tool in crime investigation. In Block

C. R.

Dabdoub

Fregly

(Eds.), Crime analysis through computer mapping (pp. 65-97). Washington, DC: Police Executive Research Forum.

53.

Sampson

R. J.

Raudenbush

S. W.

Earls

(1997). Neighborhoods and violent crime: A multilevel study of collective efficacy. Science, 227, 918-924. doi:10.1126/science.277.5328.918

54.

Sandler

J. C.

Freeman

N. J.

Socia

K. M.

(2008). Does a watched pot boil? A time-series analysis of New York State’s sex offender registration and notification law. Psychology, Public Policy, and Law, 14, 284-302. doi:10.1037/a0013881

55.

Socia

K. M.

(2011a). The policy implications of residence restrictions on sex offender housing in upstate, NY. Criminology & Public Policy, 10, 351-389. doi:10.1111/j.1745-9133.2011.00713.x

56.

Socia

K. M.

(2011b). Residence restriction legislation, sex crime rates, and the spatial distribution of sex offender residences (Doctoral dissertation, Available from ProQuest Dissertations and Theses database, UMI No. 3454528). University at Albany, State University of New York, Albany. Retrieved from http://proquest.umi.com/pqdweb?did=2365415911&;sid=1&Fmt=2&clientId=9718&RQT=309&VName=PQD.

57.

Socia

K. M.

(2012a). The efficacy of county-level sex offender residence restrictions in New York. Crime & Delinquency, 58, 612-642. doi:10.1177/0011128712441694

58.

Socia

K. M.

(2012b). The implementation of county residence restrictions in New York. Psychology, Public Policy, and Law, 18, 206-230. doi:10.1037/a0024993

59.

Socia

K. M.

(2013a). Residence restrictions and the association with registered sex offender clustering. Criminal Justice Policy Review, 24, 441-472. doi:10.1177/0887403412445613

60.

Socia

K. M.

(2013b). Too close for comfort: Registered sex offenders spatial clustering and recidivistic sex crime arrest rates. Sexual Abuse: A Journal of Research and Treatment, 25, 531-556. doi:10.1177/1079063212469061

61.

Socia

K. M.

(2014). Residence Restrictions are Ineffective, Inefficient, and Inadequate: So Now What? Criminology & Public Policy. Advance online publication. doi:10.1111/1745-9133.12071

62.

Socia

K. M.

Levenson

J. S.

Ackerman

A. R.

Harris

A. J.

(2014). “Brothers under the bridge”: Factors influencing the transience of registered sex offenders in Florida. Sexual Abuse: A Journal of Research & Treatment. doi:10.1177/1079063214521472

63.

Socia

K. M.

Stamatel

J. P.

(2010). Assumptions and evidence behind sex offender laws: Registration, community notification, and residency restrictions. Sociology Compass, 4, 1-20. doi:10.1111/j.1751-9020.2009.00251.x

64.

Socia

K. M.

Stamatel

J. P.

(2012). Neighborhood characteristics and the social control of registered sex offenders. Crime & Delinquency, 58, 565-587. doi:10.1177/001112871142011

65.

Solomon

A. L.

Johnson

K. D.

Travis

McBride

E. C.

(2004). From prison to work: The employment dimensions of prisoner reentry. Washington, DC: Urban Institute.

66.

Stark

(1987). Deviant places: A theory of the ecology of crime. Criminology, 25, 893-910. doi:10.1111/j.1745-9125.1987.tb00824.x

67.

Tewksbury

Mustaine

E. E.

(2006). Where to find sex offenders: An examination of residential locations and neighborhood conditions. Criminal Justice Studies, 19, 61-75. doi:10.1080/14786010600615991

68.

Tewksbury

Mustaine

E. E.

(2008). Where registered sex offenders live: Community characteristics and proximity to possible victims. Victims & Offenders, 3, 86-98. doi:10.1080/15564880701752371

69.

Tewksbury

Mustaine

E. E.

(2009). Stress and collateral consequences for registered sex offenders. Journal of Public Management and Social Policy, 15, 215-239.

70.

Tewksbury

Mustaine

E. E.

Covington

(2010). Offender presence, available victims, social disorganization, and sex offense rates. American Journal of Criminal Justice, 35, 1-14. doi:10.1007/s12103-010-9070-6

71.

Tewksbury

Mustaine

E. E.

Stengel

K. M.

(2007). Residential concentrations of rural registered sex offenders. American Journal of Criminal Justice, 31, 49-63. doi:10.1007/s12103-007-9004-0

72.

Tewksbury

Mustaine

E. E.

Stengel

K. M.

(2008). Examining rates of sexual offenses from a routine activities perspective. Victims & Offenders, 3, 75-85. doi:10.1080/15564880701752306

73.

Travis

(2005). But they all come back: Facing the challenges of prisoner reentry. Washington, DC: Urban Institute Press.

74.

U.S. Census Bureau. (2010). Census 2010. Retrieved from http://www.census.gov/2010census/

75.

Visher

C. A.

La Vigne

N. G.

Travis

(2004). Returning home: Understanding the challenges of prisoner reentry. Washington, DC: Urban Institute.

76.

Walker

J. T.

Ervin-McLarty

(2000). Sex offenders in Arkansas: Characteristics of offenders and enforcement of sex offender laws. Little Rock: Arkansas Crime Information Center.

77.

Walker

J. T.

Golden

J. W.

VanHouton

A. C.

(2001). The geographic link between sex offenders and potential victims: A routine activities approach. Justice Research and Policy, 3, 15-33. doi:10.3818/JRP.3.2.2001.15

78.

Wooldridge

J. M.

(2001). Econometric analysis of cross section and panel data (1st ed.). Cambridge, MA: MIT Press.

79.

Youstin

T. J.

Nobles

M. R.

(2009). Residency restrictions: A geospatial analysis of sex offender movement over time. Crime Mapping: A Journal of Research and Practice, 1, 55-76.

80.

Yung

C. R.

(2007). Banishment by a thousand laws: Residency restrictions on sex offenders. Washington University Law Review, 85, 101-160.

81.

Zandbergen

P. A.

Hart

T. C.

(2006). Reducing housing options for convicted sex offenders: Investigating the impact of residency restriction laws using GIS. Justice Research and Policy, 8, 1-24. doi:10.3818/JRP.8.2.2006.1

82.

Zandbergen

P. A.

Hart

T. C.

(2009a). Availability and spatial distribution of affordable housing in Miami-Dade County and implications of residency restriction zones for registered sex offenders. Miami, FL: Miami Coalition for the Homeless.

83.

Zandbergen

P. A.

Hart

T. C.

(2009b). Availability of affordable rental housing in Miami-Dade County under a 1,750 feet residency restriction zone (Addendum to Availability and Spatial Distribution of Affordable Housing in Miami-Dade County and Implications of Residency Restriction Zones for Registered Sex Offenders). Miami, FL: Miami Coalition for the Homeless.

84.

Zevitz

R. G.

(2004). Sex offender placement and neighborhood social integration: The making of a scarlet letter community. Criminal Justice Studies, 17, 203-222.

85.

Zgoba

K. M.

Levenson

J. S.

McKee

(2009). Examining the impact of sex offender residence restrictions on housing availability. Criminal Justice Policy Review, 20, 91-110. doi:10.1177/0887403408322119

86.

Zimring

F. E.

Leon

C. S.

(2008). A cite-checker’s guide to sexual dangerousness. Berkeley Journal of Criminal Law, 13, 65-76.