- What are the central arguments in the reading?
- What data sources and/or concepts does the author use to support the argument?
- What other lines of reasoning or thinking occur to you as a result of reading this selection? What is it about the chapters that interest you?
- What are the strengths and weaknesses of the articles or book chapters? If you were studying this issue, what would you have done similarly or differently?
As you write your critical essay, you should select one or two concepts, issues, themes, problems, questions to orient your essay.
Articl
e
Crime & Delinquency
57(1) 102 –12
9
© 2011 SAGE Publications
DOI: 10.1177/0011128708322856
http://cad.sagepub.com
Parolees’ Physical
Closeness to Social
Services: A Study of
California Parolees
John R. Hipp1, Jesse Jannetta1,
Rita Shah1, and Susan Turner1
Abstract
This study examines the proximity of service providers to recently released
parolees in California over a 2-year period (2005-2006). The addresses of
parolee residences and service providers are geocoded, and the number of
various types of service providers within 2 miles (3.2 km) of a parolee are
measured. “Potential demand” is measured as the number of parolees within
2 miles of a provider. Although racial and ethnic minority parolees have more
service providers nearby, these providers appear to be particularly impacted
based on potential demand. It is also found that the parolees arguably most
in need of social services—those who have spent more time in correctional
institutions, have been convicted of more serious or violent crimes in their
careers, or are sex offenders—live near fewer social services, or the provid-
ers near them appear impacted.
Keywords
parolees, social services, neighborhoods, propinquity
Over the past 25 years, imprisonment rates in the United States have increased
dramatically, from 330,000 in prison in 1980 to more than 1.5 million in 2005,
a 450% increase (Harrison & Beck, 2006; Lynch & Sabol, 2001). One important
1University of California, Irvine
Corresponding Author:
Dr. John R. Hipp, Department of Criminology, Law and Society, University of California,
Irvine, 2367 Social Ecology II, Irvine, CA 92697
Email: john.hipp@UCI.edu
http://crossmark.crossref.org/dialog/?doi=10.1177%2F0011128708322856&domain=pdf&date_stamp=2008-08-14
Hipp et al. 103
implication of this massive increase in incarceration has been the corresponding
increase in the number of offenders returning to communities from prison. The
number of offenders returning annually from prison to U.S. neighborhood
s
increased from 170,000 in 1980 to about 700,000 in 2005 (Lynch & Sabol,
2001; Sabol & Harrison, 2007). As a consequence, the number of ex-offenders
residing in communities has risen from 1.8 million in 1980 to 4.3 million in
2000 (Raphael & Stoll, 2004). These large numbers of ex-offenders make it all
the more important that they have access to the types of social services that
may be crucial in preventing recidivism. If the geographic location of offenders
returning to communities and the social service providers in those communities
are not coterminous, the ability of these offenders to reintegrate into the com-
munity may be seriously hampered. This is particularly relevant to reentry
models that incorporate the idea of a continuum of care from institutions into
the community (Byrne, Taxman, & Young, 2002; Lin & Turner, 2007).
Prisoners returning to their communities often have serious problems with
substance abuse, financial problems, family conflict, low educational attain-
ment, and lack strong social networks of support (Petersilia, 2003), resulting
in difficulties obtaining employment and stable housing and desisting from
criminal behavior. If offenders are returning to neighborhoods that do not
provide access to the range of services that are important for reintegrating
them into the broader community, it stands to reason that they will be less
likely to succeed in their postrelease transition and more likely to recidivate.
Given the evidence of the importance of services for reintegrating offenders
(Zhang, Roberts, & Callanan, 2006) and evidence from the public health and
workforce literature that proximity to social services is an important facilitator
of accessing them (Allard, Tolman, & Rosen, 2003; Brameld & Holman, 2006;
Gregory et al., 2000; Piette & Moos, 1996; Weiss & Greenlick, 2007), inves-
tigating the proximity of released offenders to services in the community
represents a key first step in understanding successful reentry of offenders.
Nonetheless, evidence regarding the proximity of returned offenders and ser-
vices is sparse and somewhat contradictory. Watson et al. (2004) found that few
of the organizations providing services to ex-prisoners in Houston were located
in the neighborhoods with the greatest concentration of returning offenders. By
contrast, Fleming et al. (2005) found that the locations of substance abuse and
mental health services in Allegheny County, Pennsylvania, mapped quite closely
with the residences of ex-prisoners. However, both of these studies addressed these
issues using bivariate analyses only, leaving open the possibility that spurious
effects could be driving the results.
We have even less systematic evidence regarding whether there are differ-
ences in access to service providers by the characteristics and service needs
104 Crime & Delinquency 57(1
)
of offenders. That is, do all offenders have equal access to various types of
services? Or does this access differ based on demographic characteristics of
the offender (e.g., race or age) or the criminal history of the offender (e.g.,
long-term criminals or violent criminals)? Given their particular need for such
social services, do racial/ethnic minorities indeed have reasonable access to
such services? And do those with a long history of criminal offending or those
who engage in more serious or violent crime—and thus are in need of such
services—have reasonable access? Despite the importance of these questions,
given the large increase in incarceration of the past 20 years, answers based
on empirical evidence are lacking.
We addressed this void by constructing and analyzing a unique data set to
test the relative availability of social services to parolees in the state of California
in 2 recent years: 2005 and 2006. Because of California’s determinant sentenc-
ing laws, parolees account for virtually all releases from prison. In 2006, only
1,994 of 129,811 felons (1.5%) released from state prison were not released to
parole supervision (California Department of Corrections and Rehabilitation,
2007). We tested whether the number of social service providers near parolees
differs systematically based on the demographic characteristics of parolees or
their criminal history. We also tested whether the potential demand of these
service providers differs based on these parolee characteristics.
Literature Review
Why Might Closeness of Social Services Matter for Offenders?
Accessing services after release from prison is necessary for the successful
integration of most, if not all, offenders released from prison. Parolees face
numerous challenges during the reentry process, and social services can assist
them in meeting those challenges (Petersilia, 2003). For instance, employment
services can provide information on job openings, job training, and assistance
with job search techniques such as interviewing and résumé writing. Housing
services can help parolees secure a stable residence, a necessary first step in
community reintegration. Parolees may also need substance abuse treatment,
legal assistance, family services, transportation help, and other services.
There is considerable evidence that the utilization of various social services
has positive consequences for ex-offenders. For instance, postrelease attendance at
community-based substance abuse programs is associated with less substance use
and reduced recidivism (Anglin, Prendergast, Farabee, & Cartier, 2002; Visher &
Courtney, 2007; Wexler, DeLeon, Thomas, Kressel, & Peters, 1999). Program
evaluation evidence suggests that community employment programs reduce
Hipp et al. 10
5
recidivism (Bouffard, Mackenzie, & Hickman, 2000). Zhang et al. (2006) found
that meeting the service goal of one of the constituent programs of California’s
Preventing Parolee Crime Program (PPCP) was associated with about 15% lower
recidivism rates, and that parolees who participated in multiple programs had even
better outcomes. The fact that only about 40% of the parolees in this same study
met at least one of these program goals highlights the importance of actual utiliza-
tion of available services.
We suggest that one important characteristic that might increase the extent
to which ex-offenders access services is physical proximity to the provider.
We build on the behavioral model of health care access from the public health
literature, which posits that services located near populations in need is an
important enabling resource and that the interaction of this closeness with
predisposing service-seeking characteristics and need (both perceived and
assesse
d)
on the part of individuals will increase the likelihood of accessing
these services (Anderson, 1995). Although this model was based on nonof-
fender populations, there seems little reason to suppose that it would not
operate similarly for the ex-offender population. For our population of parol-
ees, accessing services might increase simply because the presence of proxi-
mate services makes ex-offenders more aware of them. Beyond simple
awareness of services, nearby services might encourage utilization because
they require the expenditure of less time and fewer resources on the part of
ex-offenders. That is, traveling longer distances takes more time and thus can
be perceived as burdensome for some parolees who experience other time
demands in their lives. Furthermore, simply obtaining transportation for
traveling the longer distances can pose an additional burden for parolees:
Those relying on public transportation may find that longer distances result
in a nonlinear increase in travel time due to the peculiarities of negotiating
public transportation routes.
Nonetheless, there is little evidence regarding the extent to which proximity
to social services contributes to service utilization by ex-offenders. Qualitative
studies of the dynamics of prisoner reentry have found that lack of access to
transportation (La Vigne, Wolf, & Jannetta, 2004; Visher, Palmer, & Gouvis
Roman, 2007) and lack of information regarding the existence of service
providers (Visher & Farrell, 2005; Visher et al., 2007) deter ex-offenders from
accessing services. Issues of difficulty with transportation as a barrier to
employment for residents of low-income urban communities, to which a large
proportion of offenders return, are well-established (Blumenberg & Manville,
2004). This is particularly salient given evidence of racial disparities in access
to transportation (Hess, 2005).
106 Crime & Delinquency 57(1)
On the other hand, there is a much larger literature in the public health field
showing that physical closeness increases access to various types of services.
For instance, multiple studies have provided support for the proposition that
proximity to health care services results in increased service utilization (Brameld
& Holman, 2006; Gregory et al., 2000; Piette & Moos, 1996; Weiss &
Greenlick, 2007). Proximity to social services has been shown to increase the
likelihood that welfare recipients will access those services (Allard et al.,
2003). Likewise, proximity to employment opportunities increases the likeli-
hood that welfare recipients will be employed and transition off welfare rolls
(Allard & Danziger, 2003; Blumenberg & Ong, 1998). Welfare recipients are
a useful reference group because they frequently lack job skills, have low
levels of educational attainment, and suffer from mental health problems and
substance abuse (Allard et al., 2003), difficulties that are also common among
offenders. Given this evidence, it seems plausible to presume that physical
closeness to providers enables access to these services for offenders as well.
Is the Presence of Services More Important
for Some Types of Ex-Offenders?
Although the question has been explored less, it is important to ask whether
there are systematic differences in the types of offenders who have greater
access to social services. We consider three characteristics that often indicate
parolees in particular need of such services: (a) their race or ethnicity, (b) their
criminal history, and (c) whether they are a sex offender.
First, it is likely that accessing various social services is particularly impor-
tant for racial/ethnic minorities. For instance, given the considerable prior
evidence that African Americans experience discrimination in the labor force
regardless of whether they are ex-offenders (Pager, 2003; Pager & Quillian,
2005), it may be particularly important that minorities have adequate access
to employment services. A study finding that employers felt “soft” skills—such
as motivation and good customer relations—are the most important and that
African American men are frequently lacking such skills suggests that employ-
ment services that address such skills may pay significant dividends for
minorities (Moss & Tilly, 1996). Access to housing services may also be
particularly important for racial/ethnic minorities. Several studies have shown
that Whites are more likely to own a home than are Blacks or Latinos, and
there is evidence that the gap in homeownership is growing (for a review,
see Painter, Gabriel, & Myers, 2001). Zhang et al. (2006) found that African
Americans were more likely than other racial and ethnic groups to access
Hipp et al. 107
services in the Preventing Parolee Crime Program relative to their representa-
tion in the parolee population, again suggestive of particular needs for such
services. In general, research indicates that racial, ethnic, and cultural factors
affect whether people access services in general (Allard et al., 2003; Williams,
Pierce, Young, & Van Dorn, 2001).
While racial- or ethnic-minority parolees may have a particularly acute need
for various social services, there is also reason to expect that they may live near
many such services. For instance, research has shown that urban census tracts
with high poverty rates are located in closer proximity to social service provid-
ers than are census tracts with lower poverty rates (Allard, 2004), and there
have been similar findings on a national scale in New Zealand (Pearce, Witten,
Hiscock, & Blakely, 2007). Thus, offenders living in low-income, urban areas
are likely to be proximate to more social services than are offenders living
elsewhere. These neighborhood differences, combined with patterns of resi-
dential segregation in which minorities often reside in such low-income neigh-
borhoods, can result in differential proximity to service providers along racial
and ethnic lines (Allard, 2004).
Second, it is crucial to understand whether the criminal history of an ex-offender
is related to the ex-offender’s proximity to services. This is an important factor
for two reasons. First, criminal history, in terms of both severity of the most recent
offense and the extent of prior criminal activity, is a key determinant of recidivism
risk for offenders (Gottfredson & Gottfredson, 1986). Therefore, access to services
that might address the needs of serious, violent, or habitual offenders and make
them less likely to reoffend is very important from a public safety standpoint.
Second, offenders with more-serious convictions or with multiple convictions
will have served longer prison sentences on average. This greater time of incar-
ceration may affect both their service needs (e.g., employment, housing) and their
willingness to access necessary services. Nonetheless, despite this presumably
greater need for services, one study found negligible differences between ex-
offenders who did and did not access services in terms of number of prior incar-
cerations and commitment offense (Zhang et al., 2006).
Third, a special category of offender type to consider is sex offenders, given
that they have been the subject of particular public concern and interest in recent
years, as demonstrated by the passage in 2006 of Proposition 83 in California
(popularly known as “Jessica’s Law”). Sex offenders have treatment needs
specific to their offenses and are subject to residency restrictions that are likely
to affect their proximity to services. This is particularly the case in California
after the passage of Jessica’s Law, the residency restriction component of which
makes large portions of many California urban areas off limits to sex offenders
108 Crime & Delinquency 57(1)
for residence purposes (California Coalition Against Sexual Assault, 2006). To
the extent that such residency restrictions limit the access to services of this
particular population, such laws may produce a rather undesirable unintended
consequence.
Are All Service Providers Alike?
An analysis of the proximity of ex-offenders to social services that considers
only the attributes of the ex-offenders illustrates only half the picture. Equally
important is the capacity of the service providers to furnish those services. Spatial
proximity to social services is greatest for poor populations in urban areas, rela-
tive to suburban areas, but social service providers in urban areas are also proxi-
mate to many more low-income households (Allard, 2004). As a result, service
provision in low-income urban neighborhoods may fall short of demand despite
greater proximity to low-income populations that are likely to need those services.
Returning ex-offenders tend to cluster in a few urban areas, and even within a
few neighborhoods within those urban areas (La Vigne, Kachnowski, Travis,
Naser, & Visher, 2003; Solomon, Thomson, & Keegan, 2004; Watson et al.,
2004). Service providers may be concentrated in those areas as well, but service
providers in those areas may also be proximate to more ex-offenders who need
their services than are service providers in other areas.
This likely occurs because of a dynamic process. On one hand, many service
providers may choose to locate in neighborhoods with large numbers of ex-
offenders (potential clients). On the other hand, ex-offenders returning from
prison may choose to reside in neighborhoods with large numbers of available
services. In addition, when ex-offenders choose to change residences, they may
select a neighborhood on the basis of the number of services there. This dynamic
process suggests an equilibrium solution of neighborhoods with both many
service providers and many ex-offenders. Of course, other social processes can
also drive the system toward such a result, including the income level and
housing costs of neighborhoods, as well as the racial/ethnic composition. Our
interest here is simply in observing the equilibrium solution of these processes
in this study, not in attempting to tease out causal relationships.
Unfortunately, we do not have information on the capacity or utilization
levels of the service providers in our study. Allard (2004) adopted a strategy
of estimating what he termed “potential demand”: the number of persons living
near each service provider. We employed this strategy in this study. Although
it provides only a rough estimate of the impact of parolee clustering on service
access, as service providers may differ in the number of persons to whom they
Hipp et al. 109
can provide services at any given time, it does allow for a rough analysis of
the differential burden on the service provision environment of ex-offenders
returning to California communities.
Data
We addressed our research questions of the number of social service providers
near different types of parolees, and their potential demand, by using a unique
data set with information on parolees in the state of California in 2005 and
2006, their addresses during this time, and information on social services geared
toward these returning parolees. The data on parolees were obtained from the
California Department of Corrections and Rehabilitation (CDCR, 2007). We
defined our sample as those who began parole at some point between January
1, 2005, and December 31, 2006. These data provide information on all parolees
during the period, the dates of entry to and exit from a CDCR institution, and
certain characteristics of the parolees. We geocoded all addresses of a parolee
during this 2-year period and placed them at a specific latitude–longitude point.
Addresses were geocoded with a success rate of 81% for the parolees and 89%
for the service providers, and analyses were performed on these parolees.
Outcome Measures
The data on social services available to parolees comes from CDCR provider
database. Although this data set does not include all service providers avail-
able in California, it was constructed for parole agents to guide parolees
toward service providers, which suggests that it captures the most important
service providers. That is, it is these providers that parolees will be made
aware of; in addition, should parolees discover additional service providers,
this information can be added to the database by the parole agent. We geo-
coded these organizations on the basis of the address provided and placed
them at a specific latitude–longitude point. We initially created a taxonomy
of 13 types of services of importance to parolee reintegration and classified
each organization on the basis of the type of services it provides. Because
we are theoretically interested in the availability of services to parolees, and
not the existence of providers, we allowed a service provider to be counted
for each type of service it provided. Given that the initial analyses using these
13 categories showed considerable similarity over the different types of
services offered, we collapsed the 13 categories into 4 broad categories:
(a) social services, (b) self-sufficiency (financial, transportation, employment,
education, identification, and legal services), (c) family and housing, and
(d) linking with the community (community and networking services).1 These
110 Crime & Delinquency 57(1)
broader categories eased the interpretation of the results (the full results for
the 13 categories are provided in the Appendix).
For each parolee in our sample, we calculated the number of social service
organizations offering each type of service within 2 miles (3.2 km) of the
parolee’s current address. Although 2 miles is a somewhat arbitrary figure, it
does comport with the distance used in prior work and has been suggested as
an important distance by county social service administrators (Allard, 2004;
Allard et al., 2003).2 We measured distance from parolee address to service
provider “as the crow flies,” based on the latitude and longitude of the parolees
and the services. This approach provides a more precise assessment of the
presence of services near parolees than do approaches that simply count the
number of service providers in a parolee’s census tract. Our outcome measures
are the number of social service organizations providing a particular type of
service within 2 miles of the parolee. Table 1 provides the summary statistics
for our analysis variables. The average number of service providers within
Table 1. Summary Statistics for Measures Used in Analyses, California Parolees
in 2005-2006
Independent Variables Number %
African American 63,185 28.3
Latino 63,109 28.3
Asian 1,409 0.6
Other race/ethnicity 7,578 3.4
Female 30,290 13.6
Registered sex offender 22,066 9.9
M SD
Age 35.6 9.8
Property offenses 0.350 0.714
Violent offenses 0.329 0.789
Total violations on record 3.468 2.892
Days spent in CDCR institutions 1,184.9 1,252.2
Outcome measures, types of service providers
Self-sufficiency 9.982 12.853
Family and housing 7.568 11.41
0
Community networking 5.867 7.568
Social services 4.592 7.226
Average number of parolees within 403.62 387.42
2 miles (3.2 km) of service
N = 223,129 person observations.
Hipp et al. 111
2 miles of parolees ranged from 4.59 (for social services) to 9.98 (for self-
sufficiency services).
As another outcome measure, we computed a proxy for server capacity by
calculating the “potential demand” for services. We accomplished this by first
calculating the number of parolees within 2 miles of each service provider on
the initial date of our study period (January 1, 2005)—this is the potential
demand—and then calculating for each parolee the average potential demand for
all service providers within 2 miles of the specific parolee. This gives a sense of
how affected the service providers near the parolee are. In our sample, service
providers near the average parolee had approximately 400 other parolees residing
within 2 miles. We log transformed this outcome to reduce the possibility of
extreme cases as well as to ease interpretation of the results.3
Given that parolees are able to change residences, the unit of analysis for
our study is parolee address spells. Whereas about half the parolees had a single
address during the study period, others moved about. We included information
for all the addresses of particular parolees (and the number of service providers
each address placed them near) and accounted for this nonindependence in the
analyses by computing the standard errors using a Huber/White correction. In
this sample, 50% did not change residences, 26% moved just once, 12% moved
twice, and 12% moved more than twice.
Characteristics of Parolees
We took into account several parolee characteristics to determine their impact
on the number of services near a parolee. For all parolees, we had information
from their criminal record on their total number of prior offenses, number of
prior property offenses, number of prior violent offenses, and total number of
days spent in a CDCR institution. By California statute, violent offenses include
all murders, about 80% of rapes, 50% of assaults, and 40% of robberies. Serious
offenses include all of the above four violent offenses as a subset, as well as
property crimes as defined in California Penal Code Sections 667.5(c), 1192.7(c),
and 1192.8. Consequently, 60% of burglaries and about 95% of arsons are
included as serious crimes (for a complete description of these categories, see
Greenwood et al., 1994, pp. 44-47). For each parolee, we also computed the
total number of days spent in CDCR institutions during the parolee’s lifetime.
This measure revealed parolees with a long record of institutionalization and
hence perhaps a particular need for services. We also created an indicator of
whether the parolee was a sex offender.
We also accounted for demographic characteristics of the parolees. To
account for racial/ethnic differences, we created measures indicating whether
112 Crime & Delinquency 57(1)
a parolee was African American, Latino, Asian, White, or Other. Given that
ex-offenders likely have different service needs as they age (for example,
Uggen, 2000, suggested that employment services may be particularly effec-
tive for older ex-offenders as opposed to younger ones), we created a measure
of the age of a parolee at the first date at an address. To take into account pos-
sible nonlinear effects of age, we also included measures of age squared and
age cubed to test their relationship to relative closeness of services.4 Given
the evidence that female ex-offenders have service needs that differ signifi-
cantly from those of male offenders (Bloom, Owen, & Covington, 2002), we
created an indicator of women parolees to account for possible gender differ-
ences in access to services.
Method
Given that our outcome measures are counts of the number of social service
providers within a 2-mile radius of a parolee, we estimated fixed effects negative
binomial regression models. The negative binomial model treats the outcome
measure as a Poisson distribution with an additional parameter with an assumed
gamma distribution to account for the overdispersion created by the noninde-
pendence of events. Whereas a simplistic approach would simply compare all
parolees, it is arguably not appropriate to compare the number of service provid-
ers near parolees living in urban areas with the number near parolees living in
more suburban or rural areas. One strategy is to account for these differences
by including county-level variables capturing important differences over coun-
ties and to estimate a multilevel model. A risk with such an approach is that
failing to include all relevant county-level covariates will result in biased coef-
ficients at the parolee level. Given this, and the fact that we were not interested
in explaining differences between counties in the current research project, a safer
approach was to simply condition out all differences between counties through
a fixed effects approach. We adopted the fixed effects approach advocated by
Allison and Waterman (2002) because it appropriately conditions out differences
across counties.5 In this approach we are estimating the following model:
y = α + Pβ + COUNTYδ (1)
where y is the number of social services within 2 miles of the parolee, α is an
intercept, P is the particular characteristic of interest of the parolee that has β
effect on the outcome, COUNTY is a matrix of K − 1 indicators for the K
counties in California, and δ is a vector of the effects of each of these counties.
Note that whereas this strategy of accounting for differences across tracts by
Hipp et al. 113
including indicator variables results in the ‘incidental parameters’ problem for
logistic regression models, Allison and Waterman (2002) highlighted that such
is not the case in the negative binomial regression model. In this model, we
were effectively comparing parolees only with other parolees living in the same
county. For the model using the potential demand for the providers near a
particular parolee as the outcome measure, we estimated Equation 1 with a
normally distributed error term (an ordinary least squares model), given that
this is a continuous measure. All analyses were estimated in Stata 9.2. We
tested for and found no evidence of multicollinearity problems or outliers in
any of these models.
Results
Relationship Between Returning Parolees
and Proximity to Services
We begin by focusing on the relative closeness of various social service pro-
viders to our sample of parolees. An advantage of the negative binomial regres-
sion model is that the exponentiated coefficients are easily interpreted as
percentage effects on the outcome measure. For instance, the model with the
number of self-sufficiency service providers near parolees as the outcome in
Table 2 (column 1) shows that an African American parolee has 26.0% more
such providers within 2 miles, on average, than a White parolee has. A Latino
parolee has 7.5% more self-sufficiency service providers within 2 miles than
a White parolee has.
It is also clear that the general pattern for race/ethnicity is similar across these
different types of service providers. African Americans have more providers of
all types nearby, ranging from 20.3% more social service providers on average
than Whites have to 29.4% more family and housing providers. Latinos also have
more providers nearby than Whites have, though considerably fewer than African
Americans have. Service providers near Latinos range from, on average, 5.6%
more social service providers than Whites have to 7.5% more self-sufficiency
service providers. The rates are similar for Other race/ethnicity parolees. Although
these results may seem somewhat surprising, recall that prior studies have sug-
gested that poverty areas—the types of neighborhoods minorities in general, and
minority parolees in particular, tend to live in—have more services nearby (Allard,
2004). Although minorities have more social service providers nearby, these pro-
viders may be particularly impacted. We explore this possibility next.
Whereas the number of nearby service providers implicitly assumes that all
service providers are equally impacted—and therefore their services are equally
114 Crime & Delinquency 57(1)
available to the nearby parolees—it may be that the providers near some parolees
are overburdened in terms of capacity relative to need. If this is the case, obtain-
ing the needed services from such providers may be difficult. We addressed
this question in the final column of Table 2, in which the outcome is our measure
Table 2. Number of Service Providers Within 2 Miles (3.2 km) of Parolee, by
Characteristics of Parolee
Number of Service Providers Within 2 Miles (3.2 km)
Number of
Family Parolees Near
Independent Self- and Community Social Services
Variables Sufficiency Housing Networking Services (Logged)
Age (× 1,000) 7.630** 8.516** 7.092** 6.834** 8.562**
(18.24) (18.93) (16.63) (14.63) (11.02)
Age squared (× 1,000) 0.126** 0.139** 0.118** 0.125** 0.180**
(4.70) (4.94) (4.36) (4.19) (3.6
9)
Age cubed (× 1,000) −0.007** −0.008** −0.007** −0.006** −0.010**
−(5.75) −(6.30) −(5.47) −(4.28) −(4.12)
African American 0.231** 0.258** 0.235** 0.185** 0.652**
(34.07) (36.29) (34.64) (25.10) (53.60)
Latino 0.072** 0.057** 0.062** 0.055** 0.277**
(10.41) (7.76) (8.84) (7.10) (21.44)
Asian 0.037 0.040 0.049 0.018 0.157
(1.13) (1.11) (1.47) (0.47) (2.43)
Other race/ethnicity 0.077** 0.070** 0.080** 0.068** 0.197**
(5.18) (4.43) (5.39) (4.03) (6.95)
Female 0.014 0.011 0.009 0.025** 0.092**
(1.82) (1.33) (1.20) (2.86) (6.65)
Years in prison 0.004** 0.005** 0.004** 0.003** 0.014**
(4.13) (4.42) (4.17) (2.78) (7.62)
Violent offenses −0.022** −0.027** −0.020** −0.021** −0.016**
−(5.34) −(6.43) −(4.85) −(4.67) −(2.25)
Property offenses −0.024** −0.023** −0.023** −0.022** −0.027**
−(5.90) −(5.54) −(5.61) −(4.94) −(3.69)
Sex offender 0.033** 0.047** 0.036** 0.036** 0.149**
(3.59) (4.81) (3.89) (3.54) (8.69)
Just released 0.056** 0.062** 0.058** 0.054** −0.028**
from prison (8.43) (8.77) (8.66) (7.30) −(2.70)
Intercept 1.968** 1.350** 1.479** 0.700** 5.266**
(177.35) (127.37) (140.50) (50.08) (267.63)
R2 0.023 0.035 0.025 0.045 0.209
Note: Fixed effects (by county) negative binomial regression models. Standard errors corrected for
clustering by parolee. N = 223,129; t values are in parentheses.
**p < .01 (two-tailed test).
Hipp et al. 115
of average potential demand for the providers near a parolee. These analyses
show considerable evidence that the service providers near minorities are indeed
overburdened. Given that the outcome measure in this linear regression model
is natural log transformed, we once again can interpret the coefficients in terms
of percentages. Thus, the service providers near African American parolees
have 65% more parolees within 2 miles than do the providers near White
parolees. Given that, on average, an African American parolee has about 23%
more service providers nearby than does a White parolee, these combined
results suggest that minority parolees in general, and African American parolees
in particular, live clustered in neighborhoods with many service providers but
also many more parolees. The net result may well be overburdened service
providers in these neighborhoods. The pattern is similar for other minority
parolees: Latinos, for instance, have about 6% more service providers nearby,
on average, than Whites have, but these service providers have 28% more
potential demand in their catchment areas. For Other race/ethnicity parolees,
these percentages are 7% and 20%.
Turning to the other demographic measures, we see that the effect for women
is much more modest than the race/ethnicity effects: Whereas females have
2.5% more social service providers nearby than males have, the differences for
the other types of services are not significantly different when comparing males
and females. On the other hand, there is evidence here that the providers near
women are more impacted. The service providers near female parolees have,
on average, about 9% more parolees nearby than do service providers near
male parolees.
The effects of age are somewhat stronger. To get a sense of the magnitude
of these effects for age, we plotted in Figure 1 the marginal effect on the number
of nearby self-sufficiency service providers as age increases, holding all other
variables in the model constant. This figure highlights the nonlinear effect of
age, in which older parolees tend to be located nearer to more self-sufficiency
providers than are younger parolees. If younger parolees have more need for
such services (given their greater propensity to recidivate), this suggests that
those most in need are not necessarily the ones with the most service providers
nearby. The effect of age on the other types of service providers is generally
similar to that depicted in Figure 1. Whereas older parolees have more service
providers nearby, these providers appear to be more impacted, based on our
measure of potential demand in the catchment areas. Figure 2 illustrates this
nonlinear effect by plotting the effect of age using the coefficients from the
model in the last column of Table 2. Note that Figure 2 largely follows the
pattern in Figure 1 for the number of nearby services by age of parolee. This
suggests that older parolees are clustered in neighborhoods that have more
providers, but also more parolees.
116 Crime & Delinquency 57(1)
We next turn to the results for our more direct measures of parolees most
in need of such services. The general pattern for parolees most in need of
services is that they either live near fewer providers or else live near more
potentially impacted providers. First, we see consistent evidence that parolees
who have spent long periods behind bars have more service providers nearby.
For instance, an additional 3.5 years in prison (a 1-standard-deviation increase
in this sample) increases the number of social service providers nearby 1.1%,
the number of self-sufficiency and community networking services 1.5%, and
the number of family and housing service providers 1.7%. Although significant,
–
0.1
–
0.05
0
0.05
0.1
0.15
0.2
18 22 26 30 34 38 42 46 50 54 58
Age
P
ro
po
rt
io
n
ch
an
ge
in
pr
ov
id
er
s
Figure 1. Marginal Effect of Age on Number of Self-Sufficiency Service Providers
Nearby
Figure 2. Marginal Effect of Age on Potential Demand for Service Providers
–0.1
–0.05
0
0.05
0.1
0.15
0.2
18 22 26 30 34 38 42 46 50 54 58
Age
P
ro
po
rt
io
n
ch
an
ge
in
pr
ov
id
er
d
em
an
d
Hipp et al. 117
these are rather modest effects. Furthermore, it appears that the providers near
these long-term offenders may be more overburdened than other providers.
Each additional 3.5 years behind bars results in 5% more parolees near these
providers.
On the other hand, the most troubled parolees—those with more property and
violent crimes on their permanent record—actually have fewer service providers
nearby. Each additional property crime and each additional violent crime on a
parolee’s record reduce the number of various nearby service providers by between
2% and 2.7%. This effect is observed for all these types of service providers.
The story for the special case of sex offenders is a similar one: living near
providers who may well be overburdened. Although there is variability in the
closeness of sex offenders to different types of service providers, these ex-
offenders have, on average, nearly 4% more service providers nearby than
ex-offenders who are not sex offenders have. Nonetheless, it should be high-
lighted that these service providers have about 15% more parolees nearby than
do the service providers near ex-offenders who are not sex offenders. Again,
to the extent that such providers are limited in their ability to serve nearby
parolees, sex offenders may have difficulty in accessing these services.6
Finally, an unexpected finding was the evidence here that those just released
from prison tend to have more service providers nearby. In general, a parolee
just released from prison has between 5% and 6% more providers nearby than
does a parolee who has moved to a new address since reentering the community.
However, there is no evidence that these providers near parolees who have just
reentered the community are more overburdened. This may suggest that later
residential mobility decisions by parolees are taking them to worse neighbor-
hoods vis-à-vis service providers.
Conclusion
Although recent scholarship has noted the large increase in prison incarceration
during the past 20 years and the potentially important role that service providers
play in reintegrating these parolees into society, little systematic evidence exists
regarding whether parolees live near these service providers. Given the literature
on other populations, suggesting that physical distance plays a large role in
whether those in need actually utilize services, there is a crucial need to know
whether parolees live near service providers. This is an important consideration
for reentry models that stress the continuity of services from institutions into
the community (Byrne et al., 2002; Lin & Turner, 2007). Our study has utilized
a unique data set to address these questions, as well as the question of whether
certain types of parolees live near more service providers.
118 Crime & Delinquency 57(1)
By also asking whether physical closeness to service providers differs sys-
tematically based on the characteristics of parolees, our findings have illumi-
nated important differences. One key finding is that the parolees arguably most
in need of services—specifically, those who have spent more time in CDCR
institutions, have committed more property or violent offenses in their careers,
or are sex offenders—tend either to live near fewer service providers or to be
near providers that may well be overburdened. Although we could not directly
measure how overburdened providers were but instead measured the number
of parolees living near such providers, a key question is the accuracy of this
proxy. Note that our approach assumes that the capacity of these providers is
the same. For this proxy to be flawed requires that the service providers located
closest to these more long-term and serious offenders systematically have
greater capacity levels. We know of no evidence supporting this conjecture.
The question of overburdened service providers also played a large role
in the story regarding minority parolees. Although we found that Latino and
African American parolees actually have more service providers near them
than do White parolees, these providers have far more potential demand in
their catchment areas. In general, it appeared that the level to which these
providers were overburdened was about 2 to 4 times greater than the advan-
tage minorities obtained from living near more providers. Paralleling the
discussion regarding long-term and serious-crime parolees above, it would
be necessary for service providers near minority parolees to have systemati-
cally larger capacity levels than the providers near White parolees for these
findings to be overturned. Although we are aware of no such evidence, this
does suggest an avenue for future research.
A somewhat unexpected finding was that the first address after release from
prison appears to place parolees near more service providers. Given that we
know of no systematic programs by the CDCR to place parolees in advanta-
geous locations—instead, parolees are generally allowed to locate at their own
discretion, and often into the same neighborhoods they left prior to imprison-
ment (Visher & Farrell, 2005)—this may suggest that in their subsequent resi-
dential moves, parolees are moving into neighborhoods that are subpar vis-à-vis
the nearby presence of service providers—an additional direction for future
research.
The fact that residential moves appear to be taking parolees to areas where
there are fewer service providers indicates a potential problem of which policy
makers must be made aware. The factors driving this process need to be
understood and suggest another avenue for future research. Understanding
these factors would provide insight for policy makers: If parolees are moving
out of service-rich neighborhoods, then steps would need to be taken to assist
Hipp et al. 119
parolees in staying in residences proximate to services. On the other hand, if
parolees are simply unaware of the lack of services in their new neighborhoods,
then a policy geared toward providing information on the existence of such
services would be called for. This suggests a need for collaboration between
providers of housing service to parolees and agencies with an awareness of
the physical location of such services.
Despite the uniqueness of our data and the importance of our findings, certain
limitations should be acknowledged. First, our data contained information only
on parolees and service availability in one state. Although these data allowed
us to carefully explore the predictors of parolee proximity to services, the gen-
eralizability of our findings hinges on the extent to which this state is representa-
tive of other states. Confidence in the findings will therefore be increased by
replications on other states. Second, our data were limited to 2 recent years.
The generalizability of our findings to other times should thus be viewed with
caution. Nonetheless, the recency of the data at least provides important evidence
on the current status of parolees’ access to social services designed to serve
them. Third, we did not have information on all service providers available in
California, suggesting that we may have missed some potential providers.
However, our data set was constructed for parole agents to guide parolees
toward such providers, which suggests that it captures most service providers
of which parolees will be aware.
Fourth, we had limited information on the characteristics of these parolees.
Although we took into account a few key demographic characteristics, as well
as some information on their criminal history, we lacked information regarding
their marital status, children, income or education level, whether they own their
home, and their social support resources. Although this is a limitation common
to nearly all data sources on parolees, the importance of these constructs for the
reintegration of parolees suggests an important avenue for future research. These
characteristics have important implications for the type of services these parolees
need, and future work should match the closeness of service providers to the
specific needs of parolees. Indeed, the CDCR has recently implemented a survey
instrument (Correctional Offender Management Profiling for Alternative Sanc-
tions) that is geared toward assessing such needs among California parolees and
that may be useful for future assessments. In addition, the combination of the
informal social resources a parolee has, on one hand, and access to the formal
resources provided by these service providers, on the other, is likely important
for understanding the successful reintegration of parolees into the community.
Finally, we did not have information on the actual capacity level or current
demand of these service providers. Following other studies, we employed a
proxy of the number of parolees living nearby the providers. This is, of course,
an approximation given that these providers serve other populations as well,
120 Crime & Delinquency 57(1)
and measuring the actual capacity levels of these providers, as well as their
current demand, is a crucial next step for future analyses. It is possible that if
parolees reside in neighborhoods with many nonparolee residents who also
utilize such services, these providers are even more overburdened than we have
estimated here—another avenue for future research. Furthermore, although we
focused simply on the closeness to service providers under the assumption
that this will affect the degree to which parolees access such services, future
researchers will want to explicitly test the effect this physical closeness has on
the actual use of these providers. Although the public health literature suggests
that there are likely important effects, there is still a crucial need for research
on this issue for the specific population of parolees.
That the service provider database does not include information about the
capacity of these service providers is itself telling and suggests policy implica-
tions. Correctional and parole agencies need to know not only what service
providers exist that might be of use to parolees, but the capacity of both indi-
vidual providers and the service provision environment as a whole to serve
ex-offenders. To that end, it would be valuable to conduct an analysis of exist-
ing service capacity in the communities to which large numbers of parolees
return in order to determine which provider services are most appropriate for
which types of parolees, where service capacity is insufficient, and which
capacity gaps are most significant (in terms of the most lacking or overburdened
services and the services with the greatest impact on offender recidivism and
reintegration). With the information from that analysis, corrections and parole
agencies can work to link parolees with services that can meet their needs and
strategically allocate resources to fill service capacity gaps.
Despite these limitations, it should be highlighted that the uniqueness
of our data allowed us to explore important questions that have heretofore
not been addressed. With more parolees returning to neighborhoods after a
long period of mass incarceration, understanding the relative access to
social services of these returning parolees is absolutely crucial, especially
as California corrections incorporates a logic model that considers linkages
with the community as part of the reentry phase (National Research Council,
2007). Our finding that minority residents live in neighborhoods in which
the service providers may well be more overburdened based on our measure
of potential demand suggests the possibility of inequality in access to services
and also suggests a policy need to address this inequality. And our finding
that the parolees who have spent more time in prison or been convicted of
more serious offenses live near fewer service providers or near providers
that may be overburdened because of larger potential demand suggests an
important policy implication for officials in guiding service providers to
areas most in need of such services.
121
T
a
b
le
A
1
.
N
um
be
r
o
f
So
ci
a
l
a
nd
E
co
n
o
m
ic
S
er
vi
ce
P
r
o
vi
de
rs
W
it
hi
n
2
M
ile
s
(3
.2
k
m
)
o
f
Pa
ro
le
e,
B
y
C
ha
ra
ct
er
is
ti
cs
o
f
Pa
ro
le
e
N
um
be
r
o
f
Se
rv
ic
e
Pr
o
vi
de
rs
W
it
hi
n
2
M
ile
s
(3
.2
k
m
)
In
de
pe
nd
en
t
So
ci
al
S
er
vi
ce
s:
So
ci
al
S
er
vi
ce
s:
So
ci
al
S
er
vi
ce
s:
V
ar
ia
bl
es
G
en
er
al
C
lo
th
in
g
F
o
o
d
Ed
uc
at
io
n
La
bo
r
T
ra
ns
po
rt
at
io
n
Fi
na
nc
ia
l
A
ge
(
×
1,
00
0)
6.
81
7*
*
7.
50
2*
*
7.
72
0*
*
7.
32
8*
*
8.
08
6*
*
9.
94
1*
*
7.
32
0*
*
(1
4.
20
)
(1
3.
90
)
(1
4.
42
)
(1
7.
24
)
(1
7.
11
)
(1
4.
18
)
(1
6.
13
)
A
ge
s
qu
ar
ed
(
×
1,
00
0)
0.
12
7*
*
0.
14
4*
*
0.
10
1*
*
0.
11
8*
*
0.
11
4*
*
0.
14
5*
*
0.
12
6*
*
(4
.1
1)
(4
.1
2)
(2
.9
1)
(4
.4
1)
(3
.7
9)
(3
.2
1)
(4
.3
1)
A
ge
c
ub
ed
(
×
1,
00
0)
−0
.0
06
**
−0
.0
07
**
−0
.0
06
**
−0
.0
07
**
−0
.0
07
**
−0
.0
09
**
−0
.0
06
**
−(
4.
29
)
−(
3.
95
)
−(
3.
77
)
−(
5.
43
)
−(
5.
09
)
−(
4.
18
)
−(
4.
61
)
A
fr
ic
an
A
m
er
ic
an
0.
20
8*
*
0.
15
2*
*
0.
17
9*
*
0.
24
7*
*
0.
28
9*
*
0.
14
3*
*
0.
19
8*
*
(2
7.
38
)
(1
7.
99
)
(2
1.
04
)
(3
6.
55
)
(3
7.
77
)
(1
2.
85
)
(2
7.
12
)
La
ti
no
0.
07
3*
*
0.
03
5*
*
0.
04
9*
*
0.
06
6*
*
0.
08
9*
*
0.
02
3
0.
07
2*
*
(9
.1
3)
(4
.0
0)
(5
.4
8)
(9
.4
8)
(1
1.
21
)
(2
.0
5)
(9
.5
2)
A
si
an
0.
03
8
−0
.0
04
0.
01
3
0.
07
1
0.
09
2
0.
02
6
0.
02
5
(0
.9
9)
−(
0.
09
)
(0
.3
0)
(2
.1
8)
(2
.5
4)
(0
.4
7)
(0
.6
8)
O
th
er
r
ac
e/
et
hn
ic
it
y
0.
07
8*
*
0.
06
1*
*
0.
06
6*
*
0.
09
0*
*
0.
10
3*
*
0.
08
7*
*
0.
07
1*
*
(4
.5
4)
(3
.1
8)
(3
.4
1)
(6
.1
0)
(6
.2
4)
(3
.5
2)
(4
.3
7)
Fe
m
al
e
0.
02
4*
*
0.
03
7*
*
0.
02
2
0.
02
0*
*
0.
01
7
0.
03
7*
*
0.
01
1
(2
.6
9)
(3
.7
8)
(2
.1
8)
(2
.6
7)
(2
.0
1)
(2
.9
1)
(1
.2
9)
(c
on
tin
ue
d)
122
T
a
b
le
A
1
.
(c
o
n
ti
n
u
e
d
)
N
um
be
r
o
f
Se
rv
ic
e
Pr
o
vi
de
rs
W
it
hi
n
2
M
ile
s
(3
.2
k
m
)
In
de
pe
nd
en
t
So
ci
al
S
er
vi
ce
s:
So
ci
al
S
er
vi
ce
s:
So
ci
al
S
er
vi
ce
s:
V
ar
ia
bl
es
G
en
er
al
C
lo
th
in
g
F
o
o
d
Ed
uc
at
io
n
La
bo
r
T
ra
ns
po
rt
at
io
n
Fi
na
nc
ia
l
Y
ea
rs
in
p
ri
so
n
0.
00
4*
*
0.
00
2
0.
00
3
0.
00
5*
*
0.
00
5*
*
0.
00
1
0.
00
4*
*
(3
.0
6)
(1
.5
7)
(2
.3
5)
(4
.7
2)
(4
.2
5)
(0
.5
8)
(3
.6
0)
V
io
le
nt
o
ffe
ns
es
−0
.0
22
**
−0
.0
23
**
−0
.0
24
**
−0
.0
20
**
−0
.0
20
**
−0
.0
20
**
−0
.0
26
**
−(
4.
74
)
−(
4.
32
)
−(
4.
75
)
−(
4.
99
)
−(
4.
41
)
−(
2.
97
)
−(
5.
81
)
Pr
o
pe
rt
y
o
ffe
ns
es
−0
.0
23
**
−0
.0
17
**
−0
.0
24
**
−0
.0
25
**
−0
.0
22
**
−0
.0
19
**
−0
.0
24
**
−(
4.
98
)
−(
3.
41
)
−(
4.
79
)
−(
6.
18
)
−(
4.
86
)
−(
2.
81
)
−(
5.
62
)
Se
x
o
ffe
nd
er
0.
03
5*
*
0.
01
6
0.
05
3*
*
0.
01
4
0.
02
7*
*
0.
09
9*
*
0.
03
0*
*
(3
.2
7)
(1
.4
0)
(4
.4
8)
(1
.5
0)
(2
.6
4)
(6
.7
9)
(2
.9
8)
Ju
st
r
el
ea
se
d
0.
05
4*
*
0.
04
2*
*
0.
05
2*
*
0.
05
8*
*
0.
04
9*
*
0.
03
0
0.
05
5*
*
f
ro
m
p
ri
so
n
(7
.1
8)
(4
.7
1)
(6
.1
1)
(8
.8
6)
(6
.5
1)
(2
.5
2)
(7
.5
8)
In
te
rc
ep
t
0.
01
7
−0
.5
58
**
−0
.8
55
**
0.
89
3*
*
0.
28
0*
*
−1
.3
22
**
0.
64
5*
*
(1
.0
9)
−(
36
.7
3)
−(
41
.1
4)
(8
0.
42
)
(2
1.
51
)
−(
50
.7
8)
(4
8.
16
)
R
2
0.
05
8
0.
05
2
0.
08
8
0.
04
4
0.
05
8
0.
05
5
0.
03
5
N
o
te
: F
ix
ed
e
ffe
ct
s
(b
y
co
un
ty
)
ne
ga
ti
ve
b
in
o
m
ia
l r
eg
re
ss
io
n
m
o
de
ls
. S
ta
nd
ar
d
er
ro
rs
c
o
rr
ec
te
d
fo
r
cl
us
te
ri
ng
b
y
pa
ro
le
e.
N
=
2
23
,1
29
; t
v
al
ue
s
ar
e
in
p
ar
en
th
es
es
.
**
p
<
.0
1
(t
w
o
-t
ai
le
d
te
st
).
123
T
a
b
le
A
2
.
N
um
be
r
o
f
Fa
m
ily
a
nd
L
eg
al
S
er
vi
ce
P
ro
vi
de
rs
W
it
hi
n
2
M
ile
s
(3
.2
k
m
)
o
f
a
Pa
ro
le
e,
a
nd
P
o
te
nt
ia
l D
em
an
d
fo
r
Pr
o
vi
de
rs
W
it
hi
n
2
M
ile
s
o
f
a
Pa
ro
le
e,
b
y
C
ha
ra
ct
er
is
ti
cs
o
f
Pa
ro
le
e
N
um
be
r
o
f
Se
rv
ic
e
Pr
o
vi
de
rs
W
it
hi
n
2
M
ile
s
(3
.2
k
m
)
In
de
pe
nd
en
t
N
um
be
r
o
f
Pa
ro
le
es
N
ea
r
V
ar
ia
bl
es
H
o
us
in
g
Fa
m
ily
C
o
m
m
un
it
y
N
et
w
o
rk
in
g
Le
ga
l
Id
en
ti
fic
at
io
n
Se
rv
ic
es
(l
o
gg
ed
)
A
ge
(
×
1,
00
0)
9.
24
1*
*
7.
59
5*
*
6.
31
7*
*
7.
44
4*
*
14
.5
82
**
5.
43
6*
*
8.
56
2*
*
(1
9.
32
)
(1
7.
10
)
(1
3.
30
)
(1
7.
52
)
(1
9.
59
)
(1
0.
06
)
(1
1.
02
)
A
ge
s
qu
ar
ed
(
×
1,
00
0)
0.
14
0*
*
0.
12
5*
*
0.
09
9*
*
0.
11
8*
*
0.
08
5
0.
15
0*
*
0.
18
0*
*
(4
.7
3)
(4
.4
1)
(3
.3
0)
(4
.3
7)
(1
.7
3)
(4
.2
7)
(3
.6
9)
A
ge
c
ub
ed
(
×
1,
00
0)
− 0
.0
09
**
− 0
.0
07
**
− 0
.0
06
**
− 0
.0
08
**
− 0
.0
08
**
− 0
.0
06
**
− 0
.0
10
**
−(
6.
42
)
−(
5.
42
)
−(
3.
99
)
−(
5.
80
)
−(
3.
73
)
−(
3.
37
)
−(
4.
12
)
A
fr
ic
an
A
m
er
ic
an
0.
27
5*
*
0.
23
9*
*
0.
20
3*
*
0.
24
5*
*
0.
26
5*
*
0.
10
2*
*
0.
65
2*
*
(3
6.
69
)
(3
3.
78
)
(2
7.
21
)
(3
6.
23
)
(2
2.
07
)
(1
1.
82
)
(5
3.
60
)
La
ti
no
0.
05
2*
*
0.
06
5*
*
0.
06
1*
*
0.
06
3*
*
0.
07
1*
*
0.
04
5*
*
0.
27
7*
*
(6
.7
5)
(8
.8
3)
(7
.8
0)
(8
.9
9)
(5
.8
2)
(4
.8
7)
(2
1.
44
)
A
si
an
0.
05
7
0.
01
9
0.
04
3
0.
05
3
−0
.0
05
−0
.1
03
0.
15
7
(1
.5
4)
(0
.5
2)
(1
.1
7)
(1
.5
9)
−(
0.
08
)
−(
2.
21
)
(2
.4
3)
O
th
er
r
ac
e/
et
hn
ic
it
y
0.
07
4*
*
0.
07
0*
*
0.
08
5*
*
0.
08
2*
*
0.
07
9*
*
0.
01
4
0.
19
7*
*
(4
.4
5)
(4
.4
3)
(5
.1
6)
(5
.5
6)
(2
.9
5)
(0
.6
9)
(6
.9
5)
Fe
m
al
e
0.
00
6
0.
02
5*
*
−0
.0
01
0.
01
1
−0
.0
28
−0
.0
37
**
0.
09
2*
*
(0
.6
7)
(3
.1
3)
−(
0.
10
)
(1
.4
9)
−(
1.
98
)
−(
3.
63
)
(6
.6
5)
(c
on
tin
ue
d)
124
T
a
b
le
A
2
.
(c
o
n
ti
n
u
e
d
)
N
um
be
r
o
f
Se
rv
ic
e
Pr
o
vi
de
rs
W
it
hi
n
2
M
ile
s
(3
.2
k
m
)
In
de
pe
nd
en
t
N
um
be
r
o
f
Pa
ro
le
es
N
ea
r
V
ar
ia
bl
es
H
o
us
in
g
Fa
m
ily
C
o
m
m
un
it
y
N
et
w
o
rk
in
g
Le
ga
l
Id
en
ti
fic
at
io
n
Se
rv
ic
es
(l
o
gg
ed
)
Y
ea
rs
in
p
ri
so
n
0.
00
5*
*
0.
00
5*
*
0.
00
4*
*
0.
00
4*
*
0.
00
5
0.
00
1
0.
01
4*
*
(4
.2
2)
(4
.4
0)
(3
.3
7)
(4
.3
3)
(2
.5
1)
(0
.9
8)
(7
.6
2)
V
io
le
nt
o
ffe
ns
es
−0
.0
29
**
−0
.0
27
**
−0
.0
19
**
−0
.0
21
**
−0
.0
49
**
−0
.0
10
−0
.0
16
−(
6.
44
)
−(
6.
36
)
−(
4.
29
)
−(
5.
08
)
−(
6.
59
)
−(
1.
99
)
−(
2.
25
)
Pr
o
pe
rt
y
o
ffe
ns
es
−0
.0
24
**
−0
.0
23
**
−0
.0
20
**
−0
.0
24
**
−0
.0
30
**
−0
.0
21
**
−0
.0
27
**
− (
5.
48
)
− (
5.
58
)
− (
4.
42
)
− (
5.
98
)
− (
4.
09
)
− (
4.
03
)
− (
3.
69
)
Se
x
o
ffe
nd
er
0.
05
3*
*
0.
03
0*
*
0.
04
1*
*
0.
03
2*
*
0.
06
6*
*
0.
00
6
0.
14
9*
*
(5
.2
2)
(3
.0
4)
(3
.9
7)
(3
.4
8)
(4
.1
8)
0.
48
)
(8
.6
9)
Ju
st
r
el
ea
se
d
0.
06
2*
*
0.
05
7*
*
0.
05
2*
*
0.
05
7*
*
0.
02
7
0.
06
3*
*
− 0
.0
28
**
f
ro
m
p
ri
so
n
(8
.3
9)
(8
.1
1)
(7
.0
5)
(8
.6
6)
(2
.1
2)
(6
.8
1)
−(
2.
70
)
In
te
rc
ep
t
0.
77
3*
*
0.
53
2*
*
0.
73
8*
*
0.
84
9*
*
−1
.9
56
**
0.
16
7*
*
5.
26
6*
*
(6
9.
75
)
(4
4.
13
)
(6
5.
88
)
(7
8.
98
)
−(
75
.2
4)
(1
1.
43
)
(2
67
.6
3)
R
2
0.
05
0
0.
04
0
0.
04
4
0.
03
1
0.
07
5
0.
05
7
0.
20
9
N
o
te
: F
ix
ed
e
ffe
ct
s
(b
y
co
un
ty
)
ne
ga
ti
ve
b
in
o
m
ia
l r
eg
re
ss
io
n
m
o
de
ls
. S
ta
nd
ar
d
er
ro
rs
c
o
rr
ec
te
d
fo
r
cl
us
te
ri
ng
b
y
pa
ro
le
e.
N
=
2
23
,1
29
; t
v
al
ue
s
ar
e
in
p
ar
en
th
es
es
.
**
p
<
.0
1
(t
w
o
-t
ai
le
d
te
st
).
Hipp et al. 125
Authors’ Note
We thank the Center for Evidence-Based Corrections at the University of California–
Irvine for providing access to the data used in the analyses.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interests with respect to the author-
ship and/or publication of this article.
Funding
The author(s) received no financial support for the research and/or authorship of
this article.
Notes
1. We also empirically tested the degree of clustering among the initial 13 types of ser-
vices. A principal components analysis of the number of such types of services near
the parolees in our sample found that they loaded on a single factor. This suggests
that there is considerable clustering of all types of services in particular geographic
areas. We therefore focused on models with the four theoretically derived outcome
measures described here.
2. We also estimated our models using a 5-mile circle around parolees and found very
similar results.
3. With a log-transformation, the coefficients of this model can be interpreted as
percentage changes in the outcome measure. We also estimated models with the
unlogged outcome, and the results were substantively the same.
4. We tested higher level polynomials and found no significant effects. We also cre-
ated a series of categorical measures of the age of parolees and found a similar
nonlinear effect. The age categories were (a) less than or equal to 18 years of age;
(b) 19-21 years of age; (c) 22-25 years; (d) 26-29 years; (e) 30-34 years; (f)
35-39 years; (g) 40-44 years; (h) 45-49 years; (i) 50-54 years; (j) 55-59 years;
and (k) 60 years and older.
5. As Allison and Waterman (2002) discussed, the conditional fixed effects negative
binomial regression of Hausman, Hall, and Griliches (1984) does not appropriately
account for differences across units as it accounts only for the difference in the
distribution of the overdispersion across units rather than accounting for the dif-
ferences in the parameters.
6. Our data come from the time before the passage and implementation of “Jessica’s
Law” in California. The additional mobility restraints of this law will likely affect
sex offenders’ access to services even further than what we observed in this study.
126 Crime & Delinquency 57(1)
References
Allard, S. W. (2004). Access to social services: The changing urban geography of
poverty and service provision. Washington, DC: Brookings Institution.
Allard, S. W., & Danziger, S. (2003). Proximity and opportunity: How residence and race
affect the employment of welfare recipients. Housing Policy Debate, 13, 675-700.
Allard, S. W., Tolman, R. M., & Rosen, D. (2003). Proximity to service providers and
service utilization among welfare recipients: The interaction of place and race.
Journal of Policy Analysis and Management, 22, 599-613.
Allison, P. D., & Waterman, R. P. (2002). Fixed-effects negative binomial regression
models. Sociological Methodology, 32, 247-265.
Anderson, R. M. (1995). Revisiting the behavioral model and access to medical care:
Does it matter? Journal of Health and Social Behavior, 36, 1-10.
Anglin, M., Prendergast, M., Farabee, D., & Cartier, J. (2002). Final report on the
Substance Abuse Program at the California Substance Abuse Treatment Facility
(SATF-SAP) and state prison at Corcoran. Sacramento, CA: California Department
of Corrections, Office Substance Abuse Programs.
Bloom, B., Owen, B., & Covington, S. (2002). Gender-responsive strategies:
Research, practice, and guiding principles for women offenders. Washington, DC:
National Institute of Corrections.
Blumenberg, E., & Manville, M. (2004). Beyond the spatial mismatch: Welfare recipi-
ents and transportation policy. Journal of Planning Literature, 19, 182-205.
Blumenberg, E., & Ong, P. (1998). Job accessibility and welfare usage: Evidence from
Los Angeles. Journal of Policy Analysis and Management, 17, 639-657.
Bouffard, J. A., Mackenzie, D. L., & Hickman, L. J. (2000). Effectiveness of voca-
tional education and employment programs for adult offenders: A methodology-
based analysis of the literature. Journal of Offender Rehabilitation, 31, 1-41.
Brameld, K. J., & Holman, C. D. J. (2006). The effect of locational disadvantage on
hospital utilisation and outcomes in Western Australia. Health & Place, 12, 490-502.
Byrne, J. M., Taxman, F. S., & Young, D. (2002). Emerging roles and responsibilities
in the reentry partnership initiative: New ways of doing business. Washington, DC:
National Institute of Justice, U.S. Department of Justice, Office of Justice Programs.
California Coalition Against Sexual Assault. (2006). Proposition 83 CALCASA position
paper. Sacramento, CA: California Coalition Against Sexual Assault.
California Department of Corrections and Rehabilitation, Offender Information Services
Branch. (2007). Movement of prison population: Calendar year 2006. Sacramento,
CA: California Department of Corrections and Rehabilitation.
Fleming, K., Hirsch, W., Lal, R., Piper, J., Sharma, A., Shimada, T., et al. (2005).
Understanding the challenges of offender reentry in Allegheny County, phase II:
Human services supply gaps and policy simulation model. Pittsburgh, PA: Heinz
School, Carnegie Mellon University.
Hipp et al. 127
Gottfredson, S. D., & Gottfredson, D. C. (1986). Accuracy of prediction models. In
Criminal Careers and “Career Criminals.” Vol. II. Washington, DC: National
Academy Press.
Greenwood, P. W., Rydell, C. P., Abrahamse, A. F., Caulkins, J. P., Chiesa, J.,
Model, K. E., et al. (1994). Three strikes and you’re out: Estimated benefits and
costs of California’s new mandatory-sentencing law. Santa Monica, CA: RAND.
Gregory, P. M., Malka, E. S., Kostis, J. B., Wilson, A. C., Arora, J. K., & Rhoads, G. G.
(2000). Impact of geographic proximity to cardiac revascularization services on
service utilization. Medical Care, 38, 45-57.
Harrison, P. M., & Beck, A. J. (2006). Prison and jail inmates at midyear 2005.
Washington, DC: Bureau of Justice Statistics (NCJ 213133).
Hausman, J. A., Hall, B. H., & Griliches, Z. (1984). Econometric models for count data
with an application to the patents-R&D relationship. Econometrica, 52(4), 909-938.
Hess, D. B. (2005). Access to employment for adults in poverty in the Buffalo-Niagara
region. Urban Studies, 42, 1177-1200.
La Vigne, N. G., Kachnowski, V., Travis, J., Naser, R., & Visher, C. (2003). A portrait
of prisoner reentry in Maryland. Washington, DC: The Urban Institute.
La Vigne, N., Wolf, S. J., & Jannetta, J. (2004). Voices of experience: Focus group
findings on prisoner reentry in the state of Rhode Island. Washington, DC: Urban
Institute: Justice Policy Center.
Lin, J., & Turner, S. (2007). Considering secure reentry centers in California. Irvine,
CA: Center for Evidence-Based Corrections.
Lynch, J. P., & Sabol, W. J. (2001). Prisoner reentry in perspective. Washington, DC:
Urban Institute.
Moss, P., & Tilly, C. (1996). ‘Soft’ skills and race: An investigation of Black men’s
employment problems. Work and Occupation, 23, 252-276.
National Research Council. (2007). Parole, desistance from crime, and community
integration. Washington, DC: National Academies Press.
Pager, D., & Quillian, L. (2005). Walking the talk? What employers say versus what
they do. American Sociological Review, 70, 355-380.
Pager, D. (2003). The mark of a criminal record. American Journal of Sociology, 108,
937-975.
Painter, G., Gabriel, S., & Myers, D. (2001). Race, immigrant status, and housing
tenure choice. Journal of Urban Economics, 49, 150-167.
Pearce, J., Witten, K., Hiscock, R., & Blakely, T. (2007). Are socially disadvantaged
neighbourhoods deprived of health-related community resources? International
Journal of Epidemiology, 36, 348-355.
Petersilia, J. (2003). When prisoners come home: Parole and prisoner reentry. New York:
Oxford.
Piette, J. D., & Moos, R. H. (1996). The influence of distance on ambulatory care
use, death, and readmission following a myocardial infraction. Health Services
Research, 31, 573-591.
128 Crime & Delinquency 57(1)
Raphael, S., & Stoll, M. A. (2004). The effect of prison releases on regional crime rates.
Brookings-Wharton Papers on Urban Affairs, 207-255.
Sabol, W. J., & Harrison, P. M. (2007). Prison and jail inmates at midyear 2006.
Washington, DC: Bureau of Justice Statistics (NCJ 217675).
Solomon, A. L., Thomson, G. L., & Keegan, S. (2004). Prisoner reentry in Michigan.
Washington, DC: Urban Institute.
Uggen, C. (2000). Work as a turning point in the life course of criminals: A duration
model of age, employment, and recidivism. American Sociological Review, 65,
529-546.
Visher, C., & Courtney, S. M. E. (2007). One year out: Experiences of prisoners
returning to Cleveland. Washington, DC: Urban Institute.
Visher, C., & Farrell, J. (2005). Chicago communities and prisoner reentry. Washington,
DC: Urban Institute.
Visher, C., Palmer, T., & Gouvis Roman, C. (2007). Cleveland stakeholders’ perceptions
of prisoner reentry. Washington, DC: Urban Institute.
Watson, J., Solomon, A. L., La Vigne, N. G., Travis, J., Funches, M., & Parthasarathy, B.
(2004). A portrait of prisoner reentry in Texas. Washington, DC: Urban Institute.
Weiss, J. E., & Greenlick, M. R. (2007). Determinants of medical care utilization:
The effect of social class and distance on contacts with the medical care system.
Medical Care, 8, 456-462.
Wexler, H., DeLeon, G., Thomas, G., Kressel, D., & Peters, J. (1999). The Amity
Prison TC Evaluation: Reincarceration outcomes. Criminal Justice and Behavior,
26, 147-167.
Williams, J. H., Pierce, R., Young, N. S., & Van Dorn, R. A. (2001). Service utilization
in high-crime communities: Consumer views on supports and barriers. Families in
Society, 82, 409-417.
Zhang, S. X., Roberts, R. E. L., & Callanan, V. J. (2006). Preventing parolees from
returning to prison through community-based reintegration. Crime & Delinquency,
52, 551-571.
Bios
John R. Hipp is an assistant professor in the department of Criminology, Law and
Society, and Sociology at the University of California–Irvine. His research interests
focus on how neighborhoods change over time, how that change both affects and is
affected by neighborhood crime, and the role networks and institutions play in that
change. He has published substantive work in such journals as American Sociological
Review, Criminology, Social Forces, Social Problems, Mobilization, City & Community,
Urban Studies, and Journal of Urban Affairs. He has published methodological work
in such journals as Sociological Methodology, Psychological Methods, and Structural
Equation Modeling.
Hipp et al. 129
Jesse Jannetta is a research specialist in the Center for Evidence Based Corrections
at the University of California–Irvine. He received his master’s degree in public policy
from the John F. Kennedy School of Government at Harvard University in 2005. His
work for the Center has included projects on GPS monitoring of sex offender parolees,
adapting the COMPSTAT management system to a correctional agency, the role of the
Division of Juvenile Justice in the California juvenile justice system, the scope of
correctional control in California, and assessment of CDCR programs in terms of
evidence-based program design principles. His primary interest is in applied research
for policy application. He has also served on the California Comprehensive Approaches
to Sex Offender Management Task Force and as a member of the support team to the
CDCR Expert Panel on Adult Offender and Recidivism Reduction Programming.
Rita Shah is a criminology, law, and society doctoral student at the University of
California–Irvine. Her research interests include the effect of neighborhood character-
istics on parolee outcomes, the lived experience of parole, and women in the criminal
justice system.
Susan Turner, PhD, is a professor of criminology, law, and society and codirector of
the Center for Evidence-Based Corrections at the University of California–Irvine. Before
joining UCI in 2005, she was a senior behavioral scientist at the RAND Corporation in
Santa Monica, California, for more than 20 years. Her areas of expertise include the
design and implementation of randomized field experiments and research collaborations
with state and local justice agencies. She is a member of the American Society of
Criminology and the Association for Criminal Justice Research (California) and is a
fellow of the Academy of Experimental Criminology.