for MADAM PROFESSOR

For the first time in my life, I didn’t have the feel- ing that I had to go to Coxsackie, to Woodburn, and then to Sing Sing. I had the feeling now that any- thing could happen, anything that I decided to do. It seemed a little bit crazy, but I even had the feel-

ing that if I wanted to become a doctor or some- thing like that, I could go on and do it. This was the first time in my life that I’d had that kind of feel- ing, and getting out of Harlem was the first step toward that freedom.

— Claude Brown (1965:178)

Research reveals that ex-prisoners tend tobe geographically concentrated in a rela- tively small number of neighborhoods within metropolitan areas, often returning to the same criminogenic neighborhoods where they resided prior to incarceration (La Vigne et al. 2003). Alarmingly, a majority of ex-prisoners will be back in confinement within three years (Langan and Levin 2002). Yet it may be premature to implicate geographic context as a culprit in this vicious cycle. Estimating the causal impact of place of residence on the likelihood of recidi- vism is complicated by the issue of selection bias—that is, the possibility that some unmea- sured characteristic of individuals influences both where they live and their criminal behav- ior and may account for any relation between place of residence and recidivism. Individuals with a high propensity toward criminal offend- ing, and therefore a high likelihood of recidi-

A Natural Experiment on Residential Change and Recidivism: Lessons from Hurricane Katrina

David S. Kirk University of Texas at Austin

Ex-prisoners tend to be geographically concentrated in a relatively small number of

neighborhoods within the most resource deprived sections of metropolitan areas.

Furthermore, many prisoners return “home” to the same criminogenic environment with

the same criminal opportunities and criminal peers that proved so detrimental prior to

incarceration. Yet estimating the causal impact of place of residence on the likelihood of

recidivism is complicated by the issue of selection bias. In this study, I use a natural

experiment as a means of addressing the selection issue and examine whether the

migration of ex-prisoners away from their former place of residence will lead to lower

levels of recidivism. In August 2005, Hurricane Katrina ravaged the Louisiana Gulf

Coast, damaging many of the neighborhoods where ex-prisoners typically reside. The

residential destruction resulting from Hurricane Katrina is an exogenous source of

variation that influences where a parolee will reside upon release from prison. Findings

reveal that moving away from former geographic areas substantially lowers a parolee’s

likelihood of re-incarceration.

AMERICAN SOCIOLOGICAL REVIEW, 2009, VOL. 74 (June:484–505)

Direct correspondence to David Kirk, Department of Sociology, University of Texas at Austin, 1 University Station A1700, Austin, TX 78712 ([email protected]). This work has been sup- ported by a grant from the National Poverty Center at the University of Michigan. I thank John Laub, Andy Papachristos, David Thacher, David Weisburd, Chris Wildeman, the editors, the anonymous review- ers, and participants of the National Poverty Center’s 2008 grantee workshop for comments on an earlier draft of this manuscript. I also thank Bill Sabol and Rob Sampson for helpful advice on the conceptual and methodological design of the study, and I am grateful to Sean Goodison, Maggie Pendzich-Hardy, and Wendy Stickle for research assistance. Finally, I am indebted to the Louisiana Department of Public Safety & Corrections for their support, especially Secretary James LeBlanc and Susan Lindsey. Any findings or conclusions expressed are those solely of the author.

vism, might self-select into certain geographic contexts, and the characteristics of these con- texts might have little causal bearing on indi- viduals’ behavior. Moreover, parolees who return “home” following incarceration to the same neighborhoods where they resided prior to incarceration might be fundamentally dif- ferent on unobserved characteristics relative to parolees who move to different neighborhoods following incarceration. Without accounting for such processes of residential selection, observed correlations between recidivism and residen- tial migration likely reflect the effect of unmea- sured individual or family characteristics that motivated residential choice. Consequently, empirical estimates of the effects of residential migration on recidivism may be biased and inconsistent. If scholars of social context and “neighborhood effects” are to make causal statements about the impact of place of resi- dence or residential change on a given outcome measure, they must address the issue of selec- tion bias by separating out selection effects from true contextual effects.

In this study, I use a natural experiment as a means of minimizing the potential for selection bias in a study of recidivism. Hurricane Katrina and the tragic events that followed have impli- cations for prisoner reentry,1 and, in particular, the selection of residence for those prisoners released from incarceration following Katrina.2

In Orleans Parish in Louisiana, 71.5 percent of housing units suffered some damage following Hurricane Katrina, with 56 percent of housing units significantly damaged (U.S. Department of Housing and Urban Development 2006). The extent of housing destruction was similar in adjacent parishes. In both St. Bernard Parish and

Plaquemines Parish, 80 percent of housing units were damaged, while 70 percent were damaged in St. Tammany Parish and 53 percent were damaged in Jefferson Parish.

The tragedy of Katrina presents an opportu- nity—a natural experiment—to assess the effects of residential change on recidivism. The residential destruction resulting from Hurricane Katrina is an exogenous source of variation that influences where a parolee will reside upon release from prison. In the absence of complete data on why a given parolee moves to one geo- graphic area versus another, to estimate the causal effect of residential migration (i.e., the treatment) it is advantageous to have an exoge- nous source of variation that substantially influ- ences this treatment.

This study aims to answer the following research questions: First, how has the geo- graphic distribution of parolees released from Louisiana prisons changed following Hurricane Katrina? Second, if there has been a geograph- ic displacement of parolees following the hur- ricane, has this hurricane-induced migration had any negative, or even beneficial, effects on the likelihood of recidivism?

If Claude Brown’s (1965) tale of residential change and desistance from crime generalizes to the parolees in Louisiana induced to move because of Hurricane Katrina, we may find that Katrina actually led to positive outcomes for this particular slice of the population. The lesson from Katrina may be that residential change leads to turning points in the lives of parolees.

THE GEOGRAPHIC CONTEXT OF PRISONER REENTRY

The Bureau of Justice Statistics reports that presently more than 700,000 prisoners are released from state and federal facilities each year (Sabol and Couture 2008); estimates suggest that up to half of these releasees will have been in prison before (Langan and Levin 2002). By some estimates, two-thirds of former prisoners in the United States are rearrested within three years of prison release, and half are re-incarcerated (Langan and Levin 2002). These figures should not be separated from the geographic context to which prisoners return. Ex-prisoners tend to be geographically concentrated within the most resource deprived sections of metropolitan areas,

RESIDENTIAL CHANGE AND RECIDIVISM—–485

1 “Prisoner reentry” refers to the process of leav- ing prison and returning to the community (National Research Council 2007).

2 Hurricane Katrina initially developed on August 23rd, 2005. It made its first landfall on the 25th near Ft. Lauderdale, Florida as a Category 1 hurricane. The storm then moved into the Gulf of Mexico, becom- ing a Category 5 hurricane by the 28th. On August 29th, Katrina made its second landfall, in southern Louisiana, and then moved inland toward New Orleans. Levee walls on the Industrial Canal in New Orleans breached from the storm a couple of hours later, and additional levee breaches followed soon after.

often returning to the same neighborhoods where they resided prior to incarceration. For instance, research by the Urban Institute on reentry in Illinois reveals that over half of prisoners released from Illinois prisons return to the city of Chicago, and one-third of those returning to Chicago are concentrated in just six community areas (La Vigne et al. 2003). These six communities are among the most economically disadvantaged in the city. Given that many prisoners return home to their old neighborhoods, it is important to develop a theoretical understanding for why this fact is consequential.

PLACE OF RESIDENCE, RESIDENTIAL MIGRATION, AND RECIDIVISM

The alarming rates of recidivism in the United States demonstrate that there is much continu- ity in criminal behavior over the life course. Sampson and Laub (1993) explain such conti- nuity by the absence of social controls (e.g., sta- ble employment and a healthy marriage). Yet Sampson and Laub also argue that continuity in behavior is not inevitable; change is possible, despite the deleterious consequences that the mark of a criminal record has for future life events. Crucial to understanding the mecha- nisms underlying changes in criminal behavior is the notion of turning points, defined as “con- sequential shifts that redirect a process” (Abbott 1997:101).

Hurricane Katrina may have induced turning points in the lives of former prisoners for a number of different reasons. First, Katrina was a macro-level shock to the Gulf Coast region. As life-course researchers have shown, histor- ical shocks such as the Great Depression and wars have profound consequences for the life course of affected individuals (Elder 1974). A second, but no less important, implication of Hur ricane Katrina is residential change. Widespread property destruction in the Gulf Coast region had an immense impact on resi- dential relocation.

In a study of the life course of crime through age 70, Laub and Sampson (2003:149) contend that, “offenders desist [from crime] in response to structurally induced turning points that serve as the catalyst for sustaining long-term behav- ioral change.” Based on interviews with 70- year-old desisters, they find that residential change is often a fundamental turning point

leading to desistance from crime.3 Similarly, in an analysis of data from the Cambridge Study of Delinquent Development, Osborn (1980) finds that delinquents who subsequently moved away from London were significantly less like- ly to be reconvicted of a crime than were delin- quents who stayed in London. Laub and Sampson, as well as Osborn, find that a change of residence allows individuals to separate from past situations and criminal peers, thus elimi- nating some of the factors contributing to the persistence of criminal behavior.

Claude Brown’s (1965) memoir, Manchild in the Promised Land, illustrates the benefits of separating from criminogenic environments. During his troubled youth, Brown committed countless crimes, used and sold a variety of drugs, was expelled from school, asserted his dominance over both the Wiltwyck School for Boys and the Warwick State Training School, and was even shot during a burglary. Yet after this antisocial childhood, Brown was able to desist from crime while many of his peers ulti- mately went to prison, became heroin addicts, or died (or some combination thereof ). One sure reason why Brown avoided such a fate is because he did not take to heroin as did many of his peers (Brown did not enjoy his brief experimentation with the drug). Yet another compelling reason why he avoided a life of crime is change of residence. As he entered adulthood, Brown made a conscious decision to move away from Harlem and the peers, rou- tines, and temptations of a familiar environ- ment:

One thing began to scare me more than anything else about jail. This was the fact that if I went to jail and got that sheet on me, any time I decided that I didn’t want to go the crime way, that I want- ed to do something that was straight, I’d have a lot of trouble doing it behind being in jail. I didn’t want that sheet on me, and I knew if I kept hanging around Harlem I was going to get busted. (P. 177)

By moving to Greenwich Village, Brown both physically and mentally separated himself from many of the criminogenic influences in his life.

486—–AMERICAN SOCIOLOGICAL REVIEW

3 Desistance refers to the causal process that sup- ports the termination of criminal offending (Laub and Sampson 2003).

There are a number of potential mechanisms by which residential change may lead to a turn- ing point in the life course of crime, including a change in association with criminal peers and a change in one’s routine activities. Association with criminal peers may influence an individ- ual’s criminality, in part, through a contagion process whereby individuals learn the motiva- tions and techniques for crime. Association with criminal peers may also be consequential to crime because such associations provide access to criminal opportunities. With respect to the former causal mechanism, contagion models of crime generally posit that the likelihood of criminal and antisocial behavior increases with exposure to others who engage in similar behav- ior. As Warr (2002:3) convincingly demon- strates, “Criminal conduct is predominantly social behavior. Most offenders are imbedded in a network of friends who also break the law, and the single strongest predictor of criminal behavior known to criminologists is the number of delinquent friends an individual has.”

If crime is in fact a behavior learned and facilitated through group interaction, then it follows that removing individuals from their criminogenic social networks should reduce their likelihood of engaging in criminal behav- ior. Research evidence supports such a con- tention. For instance, Warr (1993) finds that exposure to delinquent peers declines with age, and this explains the decline in crime with age. In a later work, Warr (1998) dissects the reasons why marriage is such a powerful predictor of desistance from crime, arguing that it is because marriage often disrupts peer relationships. He finds that marriage reduces the likelihood that an individual will associate with delinquent peers, and it alters the amount of time individ- uals spend with delinquent peers.

In addition to a learning process, peers may influence an individual’s criminal conduct through several other mechanisms, including the creation of criminal opportunities (Laub and Sampson 2003; Warr 2002). Association with delinquent peers does not necessarily mean that an individual is learning the motivations and techniques for committing crime; rather, it may just mean that the person is exposed to more opportunities for committing crime than would be available in the absence of criminal peers. For instance, an individual may use a criminal asso- ciate to fence stolen goods or to purchase drugs.

Separating individuals from criminal networks via residential change may thus reduce their criminal behavior by eliminating opportunities for engaging in crime.

Besides altering peer associations and any corresponding criminal opportunities, residen- tial change may lead to desistance by disrupt- ing routine activities and daily temptations that are conducive to criminal conduct. For example, the cue-reactivity paradigm in addiction research suggests that drug addicts are vulnerable to relapse in the presence of familiar environ- mental stimuli (Carter and Tiffany 1999; Wikler 1948). Through a process of classical condi- tioning, drug addicts come to associate certain stimuli with the use of a drug. These stimuli can trigger physiological reactions, including an intense craving for drugs. Addicts are thus more likely to relapse in environments associated with prior drug use. Given that over half of prisoners have some form of drug dependence, and that reduced consumption of drugs is one key factor leading to desistance from crime (Mumola and Karberg 2006; National Research Council 2007), residential change may lower the likelihood of recidivism by separating drug addicts from familiar contexts associated with past drug use.

The life-course literature suggests that to decrease the probability of recidivism, it would be beneficial to separate parolees from their criminal past and their peers. Yet, sorting out the causal consequences of residential migration and peer effects requires a research design that explicitly addresses the issue of selection bias. While I am unaware of any studies that express- ly account for selection bias in investigations of the association between residential migration and recidivism, several studies of residential mobility have found that selection bias can be quite consequential for estimates of neighbor- hood effects. As but one example, with data from the Moving to Opportunity (MTO) hous- ing mobility experiment, Ludwig and Kling (2007) examine the effects of neighborhood conditions on the likelihood of arrest for a vio- lent crime. In their analysis, Ludwig and Kling compare nonexperimental results estimated through ordinary least squares (OLS) with experimental results based on instrumental vari- able (IV) estimates. For male youths, they find substantially lower treatment effects of exposure to neighborhood violence on arrest through IV

RESIDENTIAL CHANGE AND RECIDIVISM—–487

estimates than for OLS estimates. Specifically, the size of the treatment effect declined by near- ly 40 percent after accounting for the differen- tial sorting of youth into neighborhoods. Accounting for selection is thus vital to uncov- ering unbiased estimates of the effects of mov- ing on recidivism.

In this study, I address the first part of a two- part causal story. I seek to uncover whether sep- arating individuals from their for mer criminogenic environments reduces their like- lihood of recidivism. I do this through a research design that expressly accounts for selection bias. If residential migration does reduce the likelihood of recidivism, the second part of the causal story is to investigate why. Two potential answers worthy of future exploration are changes to routine activities and peer associa- tions.

DATA AND RESEARCH DESIGN

This study exploits the benefits of a natural experiment in order to characterize the geo- graphic distribution of prisoner reentry in Louisiana and assess the repercussions of geo- graphic displacement from Hurricane Katrina on the probability of re-incarceration. The ana- lytic sample is drawn from prisoners released from Louisiana correctional facilities. Because my interest is in residential displacement due to Hurricane Katrina, I restrict analyses to those prisoners who resided in affected metropolitan areas prior to incarceration. Accordingly, the analytic sample includes only ex-prisoners who were committed to prison from Orleans Parish and the four parishes adjacent to Orleans (Jefferson, Plaquemines, St. Bernard, and St. Tammany). These areas sustained substantial housing damage in the days following Hurricane Katrina. Therefore, the residential options of prisoners released post-Katrina are significant- ly different than their options if they had been released prior to the hurricane, resulting in some measure of geographic displacement. Finally, I also restrict the sample to exclude sex offend- ers. Given the nature of their offense, sex offend- ers face a number of constraints on their residency choices upon release from prison.

For the purposes of descriptive analyses, I construct three cohorts of prison releasees, two of which were released from prison prior to Hurricane Katrina and one released afterward.

The first cohort is made up of releases from a Louisiana prison to parole supervision anytime between September 2001 and February 2002 (hereafter called the 2001 to 2002 cohort).4 The second cohort consists of releases between September 2003 and February 2004 (the 2003 to 2004 cohort), while the third cohort consists of releases onto parole supervision between September 2005 and February 2006 (the post- Katrina cohort). This study uses two pre-Katrina cohor ts to more fully establish whether Hurricane Katrina altered prior trends in parolee migration. Sample sizes equal 1,538, 1,731, and 1,370 for the 2001 to 2002, 2003 to 2004, and post-Katrina cohorts, respectively.

I use three varieties of data in this study: (1) individual-level data on parolees from the Louisiana Department of Public Safety & Corrections (DPS&C) and the Division of Probation and Parole (DPP), (2) zip code and parish-level characteristics from the U.S. Depar tment of Housing and Urban Development, the Louisiana Department of Labor, and ESRI,5 and (3) Louisiana criminal

488—–AMERICAN SOCIOLOGICAL REVIEW

4 I use the term “parole supervision” to refer to all release mechanisms from incarceration that result in a term of supervision for the returning prisoner. Releases to supervision in Louisiana are generally made through parole board action or through the diminution of the sentence via good time credit. “Parole supervision” encompasses both types of releases. Roughly 90 percent of prisoners released each year from Louisiana prisons are released onto parole supervision (in contrast to unconditional releases, which do not require post-incarceration supervision). To distinguish my analytic sample of paroled releases from unconditional releases, I use the term “parolee.”

5 ESRI is a developer of geographic information systems and associated software applications, includ- ing ArcGIS. ESRI also compiles and distributes an assortment of geographically referenced data, includ- ing yearly demographic estimates of the U.S. popu- lation at county and zip code levels. I use these estimates to produce measures of racial segregation and household income. To derive post-Katrina esti- mates, ESRI (2006a, 2006b) augmented their stan- dard methodology used to compute intercensal demographic estimates by incorporating data from the Federal Emergency Management Agency (FEMA) on property damage and applications for assistance, data from the United States Postal Service National Change of Address file, and data from the American

justice system data from the Supreme Court of Louisiana, DPS&C, DPP, and the Uniform Crime Reports.

Given that macro-level social and economic conditions in Louisiana changed drastically immediately following Hurricane Katrina, one must control for such temporal changes to iso- late the effect of residential change on recidi- vism. The statistical models include controls for segregation, average household income, the unemployment rate, average weekly wages, and fair market rents. Massey and Denton (1993) argue that racial segregation has been instru- mental in creating and perpetuating “under- class” geographic areas characterized by poverty and social isolation. To the extent that underclass areas are associated with high rates of crime, whether from a lack of informal social control or a lack of police protection, temporal changes in segregation may influence temporal changes in the likelihood of recidivism (Massey and Denton 1993; Sampson and Wilson 1995). Low income levels, as well as poverty, are related to crime by a number of mechanisms. Specifically, average income is negatively related to social disorganization (Shaw and McKay 1942) and economic strain (Merton 1938), which are pos- itively predictive of crime. Unemployment may have contrasting effects on recidivism. High unemployment rates may lead to an increase in recidivism as individuals turn to crime in the absence of legitimate employment opportunities (Cantor and Land 1985; Grogger 1998). However, while criminal motivation may increase with the unemployment rate, criminal opportunity may decline. Homes are more like- ly to be occupied during the day, and potential victims are less likely to be away from home or their neighborhoods, when unemployment is high (Cantor and Land 1985). Declining wages and rising rents, by increasing deprivation, may also be related to recidivism (Blau and Blau 1982; Merton 1938).

Hurricane Katrina also had many implica- tions for temporal changes in Louisiana’s crim-

inal justice system (for detailed discussions, see Garrett and Tetlow 2006; Roman, Irazola, and Osborne 2007). Given Katrina’s impact on the criminal justice system, particularly in Orleans and adjacent parishes, it is vital to account for temporal variation in the operation of the justice system to draw causal inferences about the effect of residential change on recidi- vism. The analyses thus include control variables related to parole practices, court operations, and the probability of arrest given the com- mission of a crime.

INDIVIDUAL-LEVEL VARIABLES

RE-INCARCERATION. Re-incarceration refers to whether a parolee returned to a Louisiana prison for a new criminal conviction or a parole vio- lation within one year of prison release. This is the primary measure of recidivism used in the study. In addition to estimating the effect of migration on the combined measure of re-incar- ceration, I also perform separate analyses for new criminal convictions and parole revoca- tions.

RE-INCARCERATED OR DETAINED. I rely on Louisiana data in this study, but a proportion of the release cohorts from Louisiana may be committed to prison in another state or in the federal system. However, since the three cohorts include only those prisoners released to parole supervision, and parolees are required to remain in Louisiana and have periodic visits with their parole officer, the commitment of parolees to other state correctional systems may be limit- ed. Information on non-Louisiana incarcera- tions, as well as incarcerations in a local facility (as opposed to state facilities), is captured in the DPP case management system in a field for supervision level. A parolee’s supervision level is marked as “detained” if that parolee is con- fined in another jurisdiction.6 In the interest of

RESIDENTIAL CHANGE AND RECIDIVISM—–489

Red Cross on housing unit damage. Estimates for 2006 reflect data current as of February 2006. Data used in this study are published in annual editions of the Community Sourcebook of County Demographics and the Community Sourcebook of ZIP Code Demographics.

6 Information on other jurisdictional detention is obtained if the DPP directly transfers a given parolee to the other jurisdiction, or through a process of tracking absconders. In the event that a parolee absconds and misses a periodic visit with a parole officer, the officer may decide to issue a warrant for the parolee through the FBI’s National Crime Information Center (NCIC) and through local law enforcement. To track absconders, DPP parole offi-

determining whether my inferences are sensi- tive to the operationalization of recidivism, I re- estimate my statistical model with a second measure of recidivism. This second, more inclu- sive, measure combines the re-incarceration measure described previously with information on other detentions and indicates whether a given parolee was re-incarcerated in a Louisiana prison within one year or was detained in anoth- er jurisdiction.

DIFFERENT PARISH FROM CONVICTION. This is a binary treatment variable indicating whether parolees moved to a different parish following incarceration relative to where they were orig- inally convicted. This variable equals zero if parolees returned to the same parish where they were convicted, and one if they migrated to a dif- ferent parish.

DISTANCE OF MIGRATION. This is a continu- ous treatment variable indicating how far released parolees reside relative to where they were originally convicted. This variable equals zero if parolees return to the same parish where they were convicted. Otherwise, it is computed as the Euclidean distance (in miles) between the centroid of the parish where a parolee was orig- inally convicted and the centroid of the zip code where the parolee resided upon release from prison.

POST-KATRINA RELEASE. This is a binary vari- able indicating whether the parolee was released from prison following Hurricane Katrina. This variable is used as an instrument in analyses.

FIRST RELEASE. This is a binary variable indi- cating whether parolees were released from their first term of incarceration (equals one) or from their second or greater term (zero). Parolees as a whole are a heterogeneous group with widely var ying rates of recidivism (National Research Council 2007). Parolees

released from their first term of imprisonment have, on average, substantially lower rates of recidivism than do parolees with multiple prior incarcerations (Rosenfeld, Wallman, and Fornango 2005). Controlling for prior incar- cerations is thus vital in a study of recidivism.

The study employs five additional individual- level measures as correlates of recidivism: race, gender, age at time of release, marital status, and time served. Black parolees make up 74.2 per- cent of the sample, with Whites making up 25.7 percent. Other races make up .1 percent of the sample. I use a binary indicator (Black) in analy- ses to compare Black versus White–other race categories. Male is a binary indicator of gender. Males make up 87 percent of the sample. Married is a binary variable indicating the mar- ital status of parolees at the time of their release (not married equals zero). Ten percent of the sample was married at the time of release. Time served refers to the amount of time a parolee served in prison (in years or fraction thereof ) until release.7 Controlling for time served is crucial to account for any differences between cohorts in the average severity of prior offend- ing. If the likelihood of prisoners being released from custody once they were eligible declined in the post-Katrina period, relative to the pre- Katrina period, accounting for time served will capture differences across release cohorts in prior offending severity.

CONTEXTUAL-LEVEL VARIABLES8

DISSIMILARITY. Dissimilarity (D) is a measure of the evenness of population distribution

490—–AMERICAN SOCIOLOGICAL REVIEW

cers also search the National Law Enforcement Telecommunications System (NLETS) and local warrant systems. If absconders are detained in a local or non-Louisiana facility, information in the DPP data system is updated accordingly.

7 Time served is highly associated with the offense of conviction (e.g., prisoners in Louisiana convicted of violent offenses serve more time relative to other offenses). In the interest of minimizing collinearity, I use time served as a control in the analyses, but I do not use indicators of offense of conviction.

8 Research shows that the level of aggregation (e.g., census tract, zip code, county) used in studies of contextual effects influences inferences derived from the association between a given contextual char- acteristic and a dependent variable (Hipp 2007). In this study, I use zip code and parish levels of aggre- gation. I do so partially because measures such as unemployment and wages are reflective of metro- politan-wide economic conditions, and also because

(Duncan and Duncan 1955). In the case here, it is a measure of segregation between Blacks and Whites, reflecting their relative distributions across zip codes within each parish. D can range in value from 0, indicating complete integration, to 100, indicating complete segregation. For the 2001 to 2002 cohort, I use an estimate of D for a parolee’s parish of residence calculated from the 2002 ESRI data, while for the 2003 to 2004 and post-Katrina (i.e., 2005 to 2006) cohorts I use estimates of D from the 2004 and 2006 data, respectively.

AVERAGE HOUSEHOLD INCOME. This is a meas- ure of the average household income (in 2000 adjusted dollars) in the zip code in which a parolee resides. For the 2001 to 2002 cohort, I use the 2002 ESRI household income estimates, and I use 2004 and 2006 estimates for the 2003 to 2004 and post-Katrina cohorts, respectively.

UNEMPLOYMENT RATE. This is a measure of the parish unemployment rate in parolees’ parish of residence in the quarter during which they were released from prison. Data are drawn from the Louisiana Department of Labor.

AVERAGE WEEKLY WAGE. This is a measure of the average weekly wage (in 2000 adjusted dol- lars) in the parish where parolees reside for the quarter during which they were released from prison. Data are drawn from the Louisiana Department of Labor.

FAIR MARKET RENT. This is a measure of the average fair market rent in the parish where a parolee resides, taken from data compiled by the U.S. Depar tment of Housing and Urban Development. For the 2001 to 2002 cohort, I use the fair market rent average in 2002, while for the 2003 to 2004 and post-Katrina cohorts, I use rent averages from 2004 and 2006, respective- ly. All figures are adjusted to 2000 dollars.

CRIMINAL JUSTICE SYSTEM VARIABLES

AVERAGE PAROLE CONTACTS. This measure is drawn from the DPS&C Quarterly Statistical

Performance Report. Greater scrutiny influ- ences the likelihood that a parolee will get caught for violating conditions of parole (Turner and Petersilia 1992), yet increased scrutiny may also have a deterrent effect on parolees that lowers the likelihood of recidivism. It is thus important to control for temporal variations in average parole contacts (i.e., the total contacts parole officers have across their entire case- loads), as well as between parole district vari- ation, to account for these contrasting influences of scrutiny on recidivism. For analyses, I use the average number of contacts made across parole officers in a parole district during the quarter in which parolees were released from prison. Contacts include those that occurred within the parole off ice, as well as f ield contacts to a parolee’s residence or workplace.

JUDGE CASELOADS. This measure derives from the Supreme Court of Louisiana’s annual report. There is a vast body of research exam- ining the relation between judge caseloads and criminal case dispositions (e.g., Heumann 1975). One key issue in this regard is whether large caseloads, and the need to process large numbers of defendants, pressure courts toward the use of plea bargaining and result in lenient sentencing practices (Eisenstein, Flemming, and Nardulli 1988). Recent research (see, e.g., Johnson 2005; Ulmer and Johnson 2004) reveals that judge caseloads are inversely related to the likelihood of incarceration and the severity of criminal sentences. I thus include a control for judge caseloads given that such caseloads like- ly influence whether a convicted offender is sentenced to a term of imprisonment or some other sanction, such as probation. For the 2001 to 2002 cohort, I use the number of cases per judge in 2002 in a parolee’s parish of residence; for the 2003 to 2004 and post-Katrina cohorts, I use caseload figures from 2004 and 2006, respectively.

UCR ARRESTS PER CRIME. I use this measure, computed from the FBI’s Uniform Crime Reports, to control for the temporal and geo- graphic variation in the likelihood of getting arrested for a Part I offense (i.e., murder, rape, robbery, aggravated assault, burglary, larceny, motor vehicle theft, or arson) given the com- mission of such a crime. For the 2001 to 2002

RESIDENTIAL CHANGE AND RECIDIVISM—–491

monthly, quarterly, and yearly time series data are not available at a lower level of aggregation.

cohort, I construct this measure using the ratio of Part I arrests per reported Part I crimes in 2001 for a parolee’s parish of residence; for the 2003 to 2004 and post-Katrina cohorts, I use the 2003 and 2005 ratios, respectively.

ANALYTIC STRATEGY

Analyses follow two paths to coincide with the objectives of the study. First, I expect that pris- oners released soon after Katrina faced severe- ly constrained opportunities to reside in New Orleans, resulting in far fewer returns to the New Orleans metropolitan area than in the pre- Katrina period and therefore fewer returns to former parishes. This expectation leads to Hypothesis 1:

Hypothesis 1: Because of hurricane-related destruction, significantly fewer parolees moved to New Orleans upon release from prison during the time period immediate- ly following Hurricane Katrina, relative to the pre-Katrina period, and significantly fewer returned to the parish they inhabited prior to incarceration.

To examine this hypothesis, I geocode and map the post-release addresses for the three separate cohorts of parolees, and I provide illus- trations of how the geographic distribution of prisoner reentry changed post-Katrina.

With regard to my second research question, concerning the impact of residential change on recidivism, I expect that separating parolees from their former residential environments will be beneficial with respect to desisting from crime. Induced migration due to Hurricane Katrina allows for a separation between parolees and their criminal past, thus reducing the like- lihood of re-incarceration. These propositions yield Hypothesis 2:

Hypothesis 2: The likelihood of re-incarceration is lower when parolees reside in a geo- graphic area different from where they resided prior to incarceration.

To test Hypothesis 2, I compare the likelihood of re-incarceration within one year of release from prison for parolees who resided in the same parish upon release as where they were originally convicted versus parolees who moved to a different parish. Before describing the spe-

cific statistical analyses that will be performed, it is first relevant to highlight the benefits of using a natural experiment to examine the influ- ence of residential migration on recidivism. There are numerous potential threats to internal validity, but in studies of residential migration, the threat of selection bias is of particular con- cern, where selection refers to the process of assigning individuals to conditions (i.e., treat- ment versus control groups). One approach to remedy the threat of selection bias is simply to measure more characteristics of individuals and more fully specify statistical models with those measures; in essence, the strategy is to subset the data such that two or more individuals are identical except with respect to the treatment condition. The assumption with such an approach is that selection into treatment and control groups is ignorable (i.e., assignment to control and treatment groups is random) after conditioning on the observed characteristics of parolees. The threat of omitted variables is still probable, however, even with extensive use of statistical controls.

An alternative approach to alleviating the threat of selection bias is the use of instrumen- tal variables. With an IV approach, a variable (or variables) that is unrelated to the outcome vari- able is used as a predictor (i.e., instrument) of the key explanatory variable (i.e., the treat- ment), and the outcome variable is then regressed on the predicted treatment measure. Conceptually, this approach removes the spuri- ous correlation between the explanatory variable and unobserved characteristics, in this case unobservable characteristics of parolees. An IV remedies the issue of omitted variables by using only that portion of the variability in the treat- ment variable that is uncorrelated with omitted variables to estimate the causal relation between the treatment and outcome. The key criticism of this approach is that the assumption about the lack of relation between the instrument and the outcome variable may be problematic. However, the use of an instrument derived from a natural experiment obviates this issue. We can have more confidence that the instrument and out- come variable are unrelated if the instrument derives from a random force of nature like a hur- ricane (for a discussion, see Angrist and Krueger 2001). This assumption is known as the exclu- sion restriction—that is, cov(Zi,ui) = 0, where Zi

492—–AMERICAN SOCIOLOGICAL REVIEW

is the instrument and ui is the model error that includes the omitted variables—but it is not directly verifiable (Angrist, Imbens, and Rubin 1996). The validity of inferences from an IV analysis depends on the appropriateness of this assumption. I address potential violations of the exclusion restriction in the Discussion and Appendix.

Using Hurricane Katrina as an exogenous source of variation that influences where a parolee resides, I combine a natural experi- ment with an instrumental variable approach to provide a consistent estimate of the effect of moving to a different parish (i.e., the treat- ment) on re-incarceration (Angrist et al. 1996).9

Angrist and Krueger (2001:77) note that, “instrumental variables provide an estimate for a specific group—namely, people whose behav- ior can be manipulated by the instrument.” In the present context, Hurricane Katrina affect- ed the residential behavior of those parolees who could not or would not have moved away from the New Orleans metropolitan area fol- lowing incarceration, but it did not affect parolees who would have moved even in the absence of the hurricane.10 In short, using an IV approach allows me to compute an estimate of the effect of migrating to a different parish for parolees who otherwise would have moved back to the same parish, had Hurricane Katrina not occur red (this is the Local Average Treatment Effect, or LATE). I do not provide an estimate of the effect of moving to a differ- ent parish on re-incarceration for parolees who would have moved regardless of the hurricane.

I specify in Equations 1 and 2 the two-stage estimation process with the IV technique. The first equation models the key explanatory vari- able Si (different parish from conviction) as a function of an instrumental variable Zi (post- Katrina release) and a vector of control vari-

ables Xi used to account for any observed dif- ferences between the treatment and control groups. I assume that where parolees reside depends, in part, on whether they were released from prison before or after Hurricane Katrina:11

Si = Zi�1 + Xi� + �i (1)

The second-stage of the two-stage estima- tion process models the dependent variable Yi (re-incarceration) as a function of the predict- ed Si from Equation 1 and a vector of control variables Xi :

12

Yi* = � S ^ i + Xi� + ui (2)

where

Yi = 0 if Yi* < 0, Yi = 1 if Yi* ≥ 0, and ui ~ N(0,1).

The coefficient � is the key parameter of interest and will be used to determine whether migrating to a new parish upon release influ- ences a parolee’s likelihood of recidivism.

RESIDENTIAL CHANGE AND RECIDIVISM—–493

09 Consistency means that the IV estimate con- verges to the population parameter as the sample size increases.

10 As an example, Angrist (1990) uses Vietnam-era draft lotteries as an instrument to assess the effect of military service on future earnings. Some individu- als joined the military voluntarily, and some joined because of the draft. Angrist’s IV estimates of the effects of military service on future earnings are applicable for draftees but not volunteers.

11 I estimate Equations 1 and 2 in Stata using the ivprobit function. Via the cluster estimation com- mand with the ivprobit function, I adjust model stan- dard errors to account for the clustering (i.e., interdependence) of parolees within parishes. Note that the ivprobit function treats the endogenous treat- ment variable Si in Equation 1 as continuous, and a linear regression model is used to predict treatment assignment. As Angrist and Krueger (2001:80) reveal, using a linear regression model to produce the first- stage estimates generates consistent second-stage estimates even when the endogenous treatment vari- able is binary (as is the case here). Stata do-files for all models estimated in analyses are available from the author on request.

12 With a true randomized experiment, it is assumed that treatment and control groups are bal- anced (i.e., equivalent) with respect to observable and unobservable characteristics, aside from random vari- ation. Even though I use an exogenous source of variation—a hurricane—to predict treatment assign- ment, in the absence of random assignment it is not necessarily the case that treatment and control groups are identical. It is thus beneficial to include statisti- cal controls (Xi ) for key parolee characteristics, as well as for contextual and criminal justice covariates, to account for any observed differences across groups. The benefit of statistical controls is a gain in statis- tical efficiency (i.e., a reduction in sampling vari- ance).

RESULTS

HYPOTHESIS 1

Figure 1 provides a snapshot of the geograph- ic redistribution of parolees post-Katrina. This figure shows which parish parolees resided in immediately after their exit from prison. I use data from two pre-Katrina cohorts to establish the extent to which Hurricane Katrina prompt- ed a shift in the geographic distribution of parolees. The release figures from 2001 to 2002 and 2003 to 2004 reveal that roughly 50 percent of prisoners convicted in the five-parish area (i.e., Orleans, Jefferson, Plaquemines, St. Bernard, and St. Tammany) subsequently returned to New Orleans. Post-Katrina, this number drops to 20 percent. In the post-Katrina period, proportionally more parolees opted to reside in Baton Rouge (2 percent before Katrina versus 11 percent after), and many other parolees dispersed throughout the state. This figure demonstrates that the proportion of newly released prisoners (who were originally com- mitted from the New Orleans metropolitan area) who returned to the five-parish area upon release declined drastically immediately following Katrina. This f inding provides support for Hypothesis 1. Moreover, recall that one bene- fit of using an instrument derived from a natu-

ral experiment is to assure that the instrument is correlated with the treatment condition (i.e., migration) but otherwise unrelated to the out- come variable (i.e., re-incarceration). Figure 1 suggests that the time period of release (pre- ver- sus post-Katrina) is highly correlated with the treatment condition (see the Appendix for a detailed assessment of the assumptions of the IV framework).

Table 1 contrasts the place of residence across the three cohorts through a cross-tabulation of the proportion of members of each cohort who moved to the same parish following incarcera- tion as where they were originally convicted, rel- ative to the proportion who migrated to a different parish. Results reveal that prior to Hurricane Katrina, roughly three-quarters of parolees returned to their parish of conviction upon release from prison. Post-Katrina, this dis- tribution changed significantly (Chi-Square = 286.65, p < .001), with 49.9 percent of parolees returning to the same parish and 50.1 percent migrating to a different parish. This finding provides additional support for Hypothesis 1.

HYPOTHESIS 2

Results thus far reveal that there has been a geographic redistribution of prisoner reentry

494—–AMERICAN SOCIOLOGICAL REVIEW

Figure 1. Parish of Release for Prisoners Committed from the Five-Parish Area

post-Katrina. Whether the residential migration underlying this redistribution is causally relat- ed to the likelihood of recidivism is an empir- ical question, which I will address through an IV analysis. Before moving onto the IV results, it is necessary to address the topic of treatment noncompliance, so as to clarify precisely which type of treatment effect is estimated in this study. Recall that in the present case the treat- ment Si is represented by a given parolee resid- ing in a different parish upon release relative to where he was convicted prior to incarcera- tion. Perfect treatment compliance would rep- resent the situation where all parolees released post-Katrina (Z = 1) moved to a different parish (S = 1), and all parolees released pre-Katrina (Z = 0) moved to the same parish as where they were originally convicted (S = 0). As revealed in Table 1, this situation certainly does not hold; 49.9 percent of parolees released post- Katrina (Z = 1) moved to the same parish (S = 0), and 24.7 percent of parolees from the two pre-Katrina cohorts (Z = 0) moved to a differ- ent parish (S = 1). While we can assume that the assignment to treatment is ignorable (i.e., conditional on the instrument Z and the other covariates from Equation 1, assignment to con- trol and treatment groups is random), a con- sequence of noncompliance is that the receipt of treatment is nonignorable (Angrist et al. 1996). If this is the case, simply computing the difference between the pre- and post-Katrina cohorts on recidivism will not provide an unbi- ased or consistent estimate of the average causal effect of migrating to a different parish on recidivism. Yet, by using IV methods, I can compute a consistent estimate of the effect of migrating to a different parish on re-incarcer- ation for parolees who would not have moved had it not been for Hurricane Katrina (i.e., an

estimate of LATE).13 Results in the ensuing analyses represent the LATE estimate.

Table 2 presents the IV probit results of re- incarceration (a detailed assessment of the assumptions of the IV framework is presented in the Appendix). Focusing on the first column of results (Re-incarceration), males are more likely than females to be re-incarcerated, and married parolees are less likely to be re-incar- cerated. Age and time served in prison are neg- atively related to re-incarceration. As expected, first releases from prison are significantly and substantially less likely to recidivate than are repeat offenders (this relation will be explored in greater detail in Table 3). With respect to the contextual-level and criminal justice system covariates, only fair market rent is significant- ly associated with re-incarceration. In this case, higher rents correspond to a lower likelihood of re-incarceration, which may indicate that areas with higher rent have relatively more informal social control and surveillance, and therefore less crime.

Turning to the main finding of the study, results show that individuals who migrated to

RESIDENTIAL CHANGE AND RECIDIVISM—–495

Table 1. Post-incarceration Place of Residence

Post-incarceration Residence

Cohort Same Parish Different Parish

2001 to 2002 (N = 1,538) 75.5% 24.5% 2003 to 2004 (N = 1,731) 75.2% 24.8% Post-Katrina (N = 1,370) 49.9% 50.1%

Note: “Same” means the proportion of parolees in a given cohort who moved to the same parish following incar- ceration as where they were originally convicted. “Different” refers to the converse. Chi-Square = 286.65, p < .001.

13 Through the IV method, I adjust the recidivism difference between the pre- and post-Katrina cohorts for the fact that not all post-Katrina parolees moved to a different parish and not all pre-Katrina parolees returned to the same parish. This adjustment yields the Local Average Treatment Effect, which is com- puted as follows (Angrist et al. 1996):

{E[Yi | Zi =1] – {E[Yi | Zi = 0]} / {E[Si | Zi =1] – {E[Si | Zi = 0]} = � ,

where Yi is the outcome measure of re-incarcera- tion, Zi is the instrument (post-Katrina release), and Si is the key explanatory variable (different parish from conviction).

a different parish were significantly and sub- stantially less likely to be re-incarcerated with- in one year of release from prison. With respect to the marginal effect, the probability of re- incarceration is .15 lower for parolees who did not move back to the parish where they were originally convicted, relative to parolees who did return (net of individual, contextual, and crim- inal justice system correlates). The 95 percent confidence interval of the marginal effect ranges from –.082 to –.221. The predicted probability of re-incarceration for male parolees who returned to the same parish as where they were convicted is .265. By contrast, the predicted probability for males released to a different parish is .110. As noted, these inferences pertain to the local average treatment effect for com- pliers—that is, the effect of migrating to a dif- ferent parish for parolees who otherwise would have moved back to the same parish had Hurricane Katrina not occurred.

Re-incarceration typically results from one of two paths: a new criminal conviction or a parole revocation due to violations of the conditions of

parole (e.g., a failed drug test). To determine if the significant treatment effect of migration identified in the first column of results is robust to these two different forms of recidivism, I re- estimated the model using parole revocation and new conviction as dependent variables. The second and third columns in Table 2 reveal that individuals who migrated to a different parish were significantly less likely to be re-incarcer- ated, whether through parole revocation or a new criminal conviction.

Given that offenders with multiple prior incar- cerations have drastically higher rates of recidi- vism than do first releases (Rosenfeld et al. 2005), it may be the case that repeat offenders are relatively immune to the apparent benefit of residential migration. The risk factors that pro- pel repeat offenders toward crime, and the dis- advantages associated with the mark of a criminal record, may render the influence of a change of residence essentially meaningless. Table 3 explores this possibility by estimating Equations 1 and 2 separately by group.

496—–AMERICAN SOCIOLOGICAL REVIEW

Table 2. Instrumental Variable Probit Estimates of Re-incarceration

Re-incarceration Parole Revocation New Conviction

Coef. Robust SE Coef. Robust SE Coef. Robust SE

Different Parish from Conviction –.597 (.156)*** –.586 (.151)*** –.481 (.226)* Individual-Level —Black –.004 (.084) –.009 (.090) .024 (.066) —Male .163 (.054)** .132 (.063)* .316 (.109)** —Married –.205 (.056)*** –.236 (.051)*** –.019 (.098) —Age at release –.008 (.002)*** –.008 (.002)*** –.009 (.003)** —Time served –.034 (.013)** –.031 (.015)* –.045 (.007)*** —First release –.304 (.076)*** –.320 (.085)*** –.127 (.046)** Context and Criminal Justice System —Unemployment rate .067 (.072) .037 (.085) .198 (.129) —Average weekly wage –.003 (.005) –.004 (.005) .004 (.006) —Average household income .003 (.002) .003 (.002) .007 (.005) —Dissimilarity .021 (.040) .021 (.044) –.025 (.073) —Fair market rent –.008 (.004)* –.007 (.004)* –.011 (.006)* —Average parole contacts –.005 (.012) –.009 (.011) .017 (.022) —Judge caseloads .000 (.001) .000 (.001) –.004 (.002) —UCR arrests per crime (parish) –.244 (.245) –.407 (.508) .017 (.108) Intercept .382 (.402) .436 (.469) –1.260 (.663)* N 4,639 4,514 3,817

Note: Re-incarceration includes both parole revocations and new convictions. The instrument Zi is a binary indi- cator of the release period (pre- versus post-hurricane). The coefficient and standard error for average household income are multiplied by 1,000. Coefficients and standard errors for all other context and criminal justice system measures, except UCR arrests per crime, are multiplied by 10. Significance tests are calculated from robust standard errors. The analytic sample for the Parole Revocation model excludes new conviction recidivists. The sample for the New Conviction model excludes parole revocation recidivists. * p ≤ .05; ** p ≤.01; *** p ≤ .001 (one-tailed test).

Results reveal that all parolees, whether first releases or those with multiple prior incarcera- tions, are significantly less likely to be re-incar- cerated if they migrate to a different parish. While the likelihood of re-incarceration with- in one year differs for these two groups, with a probability of .174 for first releases and a prob- ability of .270 for repeat offenders, the marginal effect of migration is equivalent for the groups (.15).

SENSITIVITY ANALYSES

To assess whether findings are robust to dif- ferent operationalizations of the recidivism measure, the treatment variable, the sample, and the instrumental variables, I perform a series of sensitivity analyses, with results presented in Table 4.14 First, I re-estimate Equations 1 and 2 with an alternative measure of recidivism,

which combines indicators of re-incarceration and detainers. Recall that detainers refer to incarcerations in another jurisdiction (i.e., fed- eral, another state, or local). Results in the first column of Table 4 reveal that there is some vari- ation in the correlates of this second measure of recidivism relative to the re-incarceration model in Table 2, but I still find that parolees who migrate to a different parish relative to where they were originally convicted are significant- ly less likely to recidivate (re-incarcerated or detained) within one year of prison release. The probability of re-incarceration or detention is .08 lower for parolees who did not move back to the parish where they were originally convicted, relative to parolees who did return. The decline in the treatment effect of migration relative to the .15 effect from the re-incarceration model in Table 2 is likely due to the detained measure containing relatively more minor forms of deten- tion (e.g., short-term stays in jail following an arrest). Residential migration thus appears to be more consequential for desistance from seri- ous forms of criminal offending. Claude Brown’s (1965) memoir is instructive in this regard. After moving to Greenwich Village,

RESIDENTIAL CHANGE AND RECIDIVISM—–497

Table 3. IV Probit Estimates of Re-incarceration, First Release versus Repeat Offenders

First Release Repeat Offenders

Coef. Robust SE Coef. Robust SE

Different Parish from Conviction –.684 (.195)*** –.499 (.202)** Individual-level —Black .044 (.084) –.141 (.110) —Male .208 (.069)** .075 (.072) —Married –.191 (.063)** –.230 (.093)** —Age at release –.008 (.004)* –.009 (.003)** —Time served –.035 (.014)** –.042 (.047) Context and Criminal Justice System —Unemployment rate –.013 (.088) .204 (.185) —Average weekly wage .001 (.005) –.007 (.004)* —Average household income .002 (.002) .006 (.003)* —Dissimilarity –.003 (.045) .056 (.053) —Fair market rent –.009 (.005)* –.007 (.008) —Average parole contacts –.008 (.007) .000 (.025) —Judge caseloads .000 (.001) .000 (.001) —UCR arrests per crime (parish) –.482 (.636) –.136 (.108) Intercept .134 (.376) .410 (.693) N 3,275 1,364

Note: The instrument Zi is a binary indicator of the release period (pre- versus post-hurricane). The coefficient and standard error for average household income are multiplied by 1,000. Coefficients and standard errors for all other context and criminal justice system measures, except UCR arrests per crime, are multiplied by 10. Significance tests are calculated from robust standard errors. * p ≤ .05; ** p ≤ .01; *** p ≤ .001 (one-tailed test).

14 For the four different sensitivity analyses, only one model specification is varied at a time (e.g., I do not test alternate measures of recidivism and treat- ment at the same time).

498—–AMERICAN SOCIOLOGICAL REVIEW

T a b le

4 .

S en

si ti

v it

y A

n al

y se

s: I

n st

ru m

en ta

l V

ar ia

b le

P ro

b it

E st

im at

es o

f R

e- in

ca rc

er at

io n

(1 )

A lt

er n

at e

O u

tc o m

e (2

) A

lt er

n at

e T

re at

m en

t (3

) A

lt er

n at

e S

am p

le (4

) A

lt er

n at

e IV

R e-

in ca

rc er

at io

n /D

et ai

n ed

R e-

in ca

rc er

at io

n (

D is

ta n

ce )

R e-

in ca

rc er

at io

n (

O rl

ea n

s) R

e- in

ca rc

er at

io n

( 5 I

V s)

C o ef

. R

o b u

st S

E C

o ef

. R

o b u

st S

E C

o ef

. R

o b u

st S

E C

o ef

. R

o b u

st S

E

T re

at m

en t

— D

if fe

re n t

p ar

is h f

ro m

c o nv

ic ti

o n

– .2

1 7

(. 0 7 3 )*

* – .3

8 8

(. 2 0 7 )*

– .5

3 4

(. 1 8 8 )*

* —

D is

ta n ce

o f

m ig

ra ti

o n (

m il

es )

– .0

0 4

(. 0 0 1 )*

* *

In d iv

id u al

-L ev

el —

B la

ck .0

4 7

(. 0 8 4 )

.0 2 1

(. 0 5 8 )

– .3

5 8

(. 0 9 2 )*

* *

– .0

0 1

(. 0 8 8 )

— M

al e

.2 4 0

(. 0 4 0 )*

* *

.1 7 2

(. 0 5 0 )*

* *

.2 0 3

(. 0 7 4 )*

* .1

6 5

(. 0 5 8 )*

* —

M ar

ri ed

– .1

1 2

(. 0 4 9 )*

– .2

4 2

(. 0 6 5 )*

* *

– .3

4 4

(. 1 2 8 )*

* – .2

0 9

(. 0 5 2 )*

* *

— A

g e

at r

el ea

se – .0

1 3

(. 0 0 2 )*

* *

– .0

0 8

(. 0 0 3 )*

* *

– .0

0 6

(. 0 0 3 )*

– .0

0 8

(. 0 0 2 )*

* *

— T

im e

se rv

ed – .0

0 7

(. 0 0 6 )

– .0

3 4

(. 0 1 3 )*

* – .0

4 3

(. 0 1 2 )*

* *

– .0

3 5

(. 0 1 2 )*

* —

F ir

st r

el ea

se – .2

2 5

(. 0 4 7 )*

* *

– .3

2 0

(. 0 7 1 )*

* *

– .2

1 0

(. 0 5 3 )*

* *

– .3

0 3

(. 0 7 3 )*

* *

C o n te

x t

an d C

ri m

in al

J u st

ic e

S y st

em —

U n em

p lo

y m

en t

ra te

– .1

6 7

(. 0 5 9 )*

* .0

0 1

(. 0 5 0 )

.2 2 0

(. 1 7 9 )

.0 4 1

(. 0 8 3 )

— A

ve ra

g e

w ee

k ly

w ag

e .0

0 5

(. 0 0 5 )

– .0

0 7

(. 0 0 4 )

– .0

1 7

(. 0 0 8 )*

– .0

0 3

(. 0 0 5 )

— A

ve ra

g e

h o u se

h o ld

i n co

m e

– .0

0 1

(. 0 0 2 )

.0 0 2

(. 0 0 2 )

.0 1 1

(. 0 0 6 )*

.0 0 3

(. 0 0 2 )*

— D

is si

m il

ar it

y .0

0 3

(. 0 2 9 )

.0 2 8

(. 0 3 0 )

.0 6 2

(. 0 8 3 )

.0 1 8

(. 0 3 9 )

— F ai

r m

ar ke

t re

n t

– .0

0 1

(. 0 0 2 )

– .0

0 6

(. 0 0 3 )*

– .0

0 1

(. 0 0 6 )

– .0

0 7

(. 0 0 4 )

— A

ve ra

g e

p ar

o le

c o n ta

ct s

.0 1 1

(. 0 1 3 )

– .0

0 1

(. 0 1 1 )

– .0

0 2

(. 0 2 3 )

– .0

0 4

(. 0 1 0 )

— Ju

d g e

ca se

lo ad

s .0

0 0

(. 0 0 1 )

.0 0 0

(. 0 0 1 )

– .0

0 4

(. 0 0 2 )*

* .0

0 0

(. 0 0 1 )

— U

C R

a rr

es ts

p er

c ri

m e

(p ar

is h )

– .1

6 9

(. 1 0 4 )

– .2

6 0

(. 3 1 0 )

.0 1 4

(. 1 1 7 )

– .2

5 4

(. 2 6 3 )

In te

rc ep

t – .1

2 2

(. 4 6 5 )

.3 5 7

(. 3 9 2 )

.6 5 4

(. 6 5 8 )

.3 1 5

(. 3 1 4 )

N 4 ,6

3 9

4 ,6

3 9

2 ,2

0 7

4 ,6

3 9

N o te

: In

M o d el

4 , “5

I V

s” d

en o te

s th

at t

h e

m o d el

w as

e st

im at

ed w

it h f

iv e

in st

ru m

en ts

m ea

su ri

n g t

h e

in te

ra ct

io n b

et w

ee n t

im e

p er

io d o

f re

le as

e (p

re -

ve rs

u s

p o st

-K at

ri n a)

a n d

p ar

is h o

f co

nv ic

ti o n (

O rl

ea n s,

J ef

fe rs

o n , P

la q u em

in es

, S

t. B

er n ar

d, o

r S

t. T

am m

an y ).

T h e

co ef

fi ci

en t

an d s

ta n d ar

d e

rr o r

fo r

av er

ag e

h o u se

h o ld

i n co

m e

ar e

m u lt

ip li

ed b

y 1

,0 0 0 .

C o ef

fi ci

en ts

a n d s

ta n d ar

d e

rr o rs

f o r

al l

o th

er c

o n te

x t

an d c

ri m

in al

j u st

ic e

sy st

em m

ea su

re s,

e x ce

p t

U C

R a

rr es

ts p

er c

ri m

e, a

re m

u lt

ip li

ed b

y 1

0 . S

ig n if

ic an

ce t

es ts

a re

c al

cu la

te d

fr o m

r o b u st

s ta

n d ar

d e

rr o rs

. F

o r

M o d el

3 , th

e sa

m p le

i s

re st

ri ct

ed t

o p

ar o le

es o

ri g in

al ly

c o nv

ic te

d i

n O

rl ea

n s

P ar

is h .

* p

≤ .0

5 ;

* * p

≤ .0

1 ;

* * * p

≤ .0

0 1 (

o n e-

ta il

ed t

es t)

.

Brown avoided the serious forms of criminali- ty that had led to his detention in juvenile reform institutions. However, he did not abstain from criminal behavior altogether (e.g., he frequent- ly used marijuana and occasionally got into fights). Residential migration may thus lead to a de-escalation to less serious forms of offend- ing, rather than a total cessation from criminal behavior.

To assess the sensitivity of results to the type of treatment variable used, I re-estimated Equations 1 and 2 with an alternative treatment measure. For analyses performed up to this point, I used a binary treatment measure indi- cating whether parolees moved to a different parish relative to where they were originally convicted. As an alternate treatment, I use a measure of distance indicating how far parolees moved relative to the parish where they were originally convicted (i.e., distance of migra- tion). In this sense, the measure of distance is analogous to a treatment dosage, and the model reveals whether the level of dosage affects the likelihood of re-incarceration. Results in the second set of columns in Table 4 once again demonstrate the significant effect of migration on re-incarceration. In this case, the marginal treatment effect of migration is equivalent to a .01 reduction in the likelihood of re-incarcera- tion for every 10 miles parolees migrate away from their prior parish.

Recall that the sample used in this study is made up of prisoners originally convicted in metropolitan areas affected by Hur ricane Katrina (i.e., Orleans, Jefferson, Plaquemines, St. Bernard, and St. Tammany parishes). As a third sensitivity analysis, I restrict the sample to individuals who were convicted in Orleans Parish and re-estimate Equations 1 and 2 with this restricted sample. Given that Orleans Parish typically has a higher crime rate than the other parishes (Federal Bureau of Investigation 2007), residential migration away from criminogenic areas may be particularly advantageous for indi- viduals convicted in Orleans Parish, while less advantageous for individuals convicted in lower crime areas. The third model in Table 4 reveals that there is a significant negative effect of migration on re-incarceration for parolees who were originally convicted in Orleans Parish, although the treatment effect is smaller rela- tive to the full sample (the marginal effect of treatment is .15 for the full sample and .11 for

the Orleans Parish sample). For comparison purposes, I also estimated the model excluding the Orleans convictions but including individ- uals convicted in the four other parishes, and I found a marginal treatment effect of .19 (results available from the author on request). In sum, I find a negative effect of migration regardless of the pre-incarceration location of the parolee, although the size of the treatment effect varies by parish. One potential reason for this diver- gence may be related to distance. If the likeli- hood of recidivism declines with distance from the original parish, then the lower marginal effect of treatment for parolees convicted in Orleans Parish may be due to the fact that movers did not migrate very far when they left the parish (at least relative to movers from other parishes). Such a scenario is a worthy topic for future investigation.

As a final sensitivity analysis, I re-estimated Equation 1 using an alternative specification of instrumental variables. As described earlier, property damage from Hurricane Katrina var- ied by parish; St. Bernard and Plaquemines parishes suffered the most damage, while Jefferson Parish had the least amount of dam- age. As depicted in Figure 1, the likelihood of a parolee moving to a given parish post-incar- ceration was influenced by the extent of hous- ing destruction in the parish. Given that both the time period of release from prison (pre- versus post-Katrina) and the geographic variation in property damage affect whether parolees will return to the same parish where they were orig- inally convicted, I use an interaction between time period of release and parish of conviction as an alternative specification of instrumental variables.15 With this specification, I assume that

RESIDENTIAL CHANGE AND RECIDIVISM—–499

15 In essence, I have five instruments. The first instrument (post-Katrina release and Orleans con- viction) equals 1 for all parolees who were released following Katrina and who were originally convict- ed in Orleans Parish; it equals 0 otherwise. I construct similar instruments for parolees originally convict- ed in Jefferson, Plaquemines, St. Bernard, and St. Tammany parishes. I entered all five instruments into Equation 1 simultaneously. One key benefit of estimating a model with more instruments than endogenous regressors (i.e., treatments) is that I can test the IV assumption that the dependent variable and instruments are unrelated (see the Appendix for details).

whether parolees migrate to a different parish or not depends on whether they were released from prison before or after Hurricane Katrina and where they were originally convicted.

Findings presented in the fourth model in Table 4 are consistent with inferences derived in Table 2. Parolees who migrated to a differ- ent parish were significantly less likely to be re- incarcerated, such that the marginal effect of re-incarceration is .14 lower for parolees who did not move back to the parish where they were originally convicted, relative to parolees who did return.

In summary, findings suggest that moving away from former places of residence serves some benefit for parolees. A change of resi- dence allows individuals to separate from their peers and the temptations that contributed to their criminality in the first place. This finding holds for both first releases and repeat offend- ers, and it holds under alternative specifica- tions of the dependent variable, treatment variable, sample, and instruments.

DISCUSSION

Selection bias is very much an issue with research on both residential migration and neighborhood effects. The innovation of this study has been to use a natural experiment to investigate a theoretical question whose answer has largely eluded researchers because of selec- tion—just how consequential is a change of residence to behavioral outcomes such as crime? In the absence of perfect information on why a given individual moves to one place of resi- dence versus another, it is difficult to answer this question because omitted variables may intro- duce bias into estimates of the effects of migra- tion. While tragic, the residential destruction and migration resulting from Hurricane Katrina present a unique opportunity for understanding how place and migration affect outcomes such as crime and recidivism.

This study addresses whether separating indi- viduals from their former residential environ- ments reduces their likelihood of recidivism. My findings support this assertion. In particular, the marginal effect of migrating to a different parish upon release from prison is a .15 decline in the probability of re-incarceration. This find- ing holds for both first releases from incarcer-

ation and parolees with multiple prior incar- cerations.

We must consider whether such findings are internally valid. Recall my discussion of the exclusion restriction, which states that the instru- ment and the outcome variable are unrelated except indirectly through the effect of the instru- ment on the causal treatment. The validity of the IV analysis rests on the assumption that any effect of the time period of release (pre- versus post-Katrina) on the likelihood of re-incarcer- ation must be captured through the effect of time period on residential migration. Tests pre- sented in the Appendix provide support for the assumption that the instruments and the re- incarceration measure are not significantly cor- related. While I cannot completely rule out the possibility of violating the exclusion restric- tion, my findings suggest that even with such a violation, the potential bias would not wholly eliminate the sizable treatment effect observed in this study.

The next step in this causal story is to further investigate why there is such a powerful treat- ment effect from changing place of residence. For instance, the causal mechanism underlying the effect of migration may be related to social ties. Parolees who migrate to a different parish may be more likely to sever ties with criminal peers than would parolees who return to the same parish. They would therefore have less opportunity and provocation for engaging in crime. While I reason that changes to peer asso- ciations and routine activities explain why migration is consequential to recidivism, I attempted to account for several alternative explanations for the reduced likelihood of recidi- vism among parolees who migrated. For instance, judges may be less likely to convict or sentence an offender to prison and parole offi- cers may be less likely to revoke the parole of parole violators in destination parishes, rela- tive to the original parishes. By controlling for court, parole, and police practices in my statis- tical models, I attempted to determine if migra- tion has a causal effect on recidivism after consideration of the variation in criminal justice practices across space and time.

Similarly, to account for the possibility that parolees who migrated were more likely to abscond and leave Louisiana, I examined the effect of migration not only on re-incarcera- tions in Louisiana (i.e., Table 2), but also on

500—–AMERICAN SOCIOLOGICAL REVIEW

detentions in local, federal, or other state facil- ities (i.e., the first column in Table 4). Still, the potential for unmeasured variation in the oper- ation of the criminal justice system across space and time presents a possible alternative expla- nation for the effect of migration on recidivism (in contrast to the hypothesized mechanisms of changes in peer association and routine activi- ties). Exploration of mechanisms underlying the causal effect of migration is an important avenue for future research.

Yet, even before the full causal story bears out, the findings established thus far with respect to residential migration and recidivism have significant policy implications. If separating ex-prisoners from their former residential envi- ronments actually benefits those prisoners with- out sacrificing public safety, then a logical next step is to consider how to disperse the popula- tion of ex-prisoners on a large scale. Research by the Urban Institute suggests that substantial proportions of returning prisoners would wel- come the opportunity to move away from their former criminogenic environments (Visher and Farrell 2005). For instance, in a sample of recently released males from the Illinois Department of Corrections, Visher and Farrell (2005) find that 45 percent of returning pris- oners explicitly expressed a desire to move to a different neighborhood than where they lived prior to prison. Over half of this group wished to live in a different neighborhood to avoid drugs and the other temptations and troubles of familiar settings.

Yet many ex-prisoners still end up moving back to their former counties and neighbor- hoods despite an expressed interest to avoid such places. One prime reason is because of legal barriers. In most states, prisoners released to some form of parole supervision are legally required to return to their county of last resi- dence (National Research Council 2007). Criminal justice policies in most states are designed to return prisoners to the same famil- iar surroundings where they got into trouble with the law in the first place. In Louisiana, there are no such geographic restrictions. The find- ings from Louisiana presented in this study sug- gest that allowing ex-prisoners to move to different parishes or counties will reduce recidi- vism. It may thus be fruitful for states to recon- sider the residency restrictions imposed on returning prisoners.

As noted, however, the sizable treatment effect of moving found in this study is only applicable to parolees who moved to a different parish post-Katrina who otherwise would have moved back to where they lived before. To fully explore the policy implications of this finding, we must understand why prisoners return to their former parishes and neighborhoods even in a state like Louisiana where there are no legal barriers preventing parolees from moving to a different parish. We know from the Urban Institute research that another key reason why released prisoners move back to their former neighborhoods is to be close to family (Visher and Farrell 2005). To capitalize on the apparent benefits of residential migration, it may thus be necessary to both remove legal barriers pre- venting ex-prisoners from moving and provide ex-prisoners and their families with opportuni- ties and incentives to move away from old neigh- borhoods. While forcing ex-prisoners to move away from their old neighborhoods is neither realistic nor ethical, providing opportunities for ex-prisoners to move away from old neighbor- hoods is a policy prescription that may be worth pursuing.

Before enacting policy initiatives to promote residential change among the parole popula- tion, more research must be done to both vali- date the findings demonstrated in this study and uncover the mechanism underlying the treat- ment effect. With respect to the former, repli- cating the current study may prove challenging given that the research design is based on a nat- ural experiment. One possibility for future research is to contrast recidivism rates of states that require prisoners to return to their county of last residence with states that do not. A sec- ond approach would be to identify states that changed the residency restrictions imposed on released prisoners and then compare recidivism rates before and after the changes were made.

Two particular extensions to the current study would add another layer of nuance to under- standing the relation between migration and recidivism. First, results presented in Table 4 reveal a negative relationship between distance of migration and recidivism, yet it remains unclear just how far parolees must move to sep- arate themselves from past situations and crim- inal associates. For Claude Brown (1965), the roughly seven miles between Harlem and Greenwich Village was sufficient to separate

RESIDENTIAL CHANGE AND RECIDIVISM—–501

him from the daily criminal temptations of his former social environment. The key question is whether there is a distance threshold after which parolees have a substantial decline in the like- lihood of recidivism. Moreover, does a parolee’s likelihood of desisting from crime increase with each increment after this threshold, or does it remain approximately the same after passing the threshold? Further research is needed to answer such questions. Second, as Abbott (1997:89) notes, “what makes a turning point a turning point rather than a minor ripple is the passage of sufficient time ‘on a new course’ such that it becomes clear that direction has indeed been changed.” To more definitively conclude that Hurricane Katrina and the resulting migration produced turning points in the life course of crime for many former prisoners, it is thus nec- essary to follow up with the same cohorts of individuals in future analyses to determine if the observed differences in recidivism across cohorts remain after a sufficient passage of time (e.g., three to five years).

In addition to these various opportunities to further investigate the relation between res- idential migration and recidivism, other wor- thy research topics include the repercussions of macro-structural changes due to Katrina and the psychological strain associated with such a tragedy. With respect to macro-struc- tural change, Katrina not only influenced res- idential decisions at the individual-level but also the macro-level distribution of parolees and criminals more generally. De-concentrat- ing the spatial clustering of criminals, just like de-concentrating poverty, may attenuate the cultural disorganization that characterizes so many urban neighborhoods, as well as reverse the deleterious consequences of social isolation (Sampson and Wilson 1995). With respect to psychological strain, one plausible hypothesis is that strain associated with the hurricane would increase the likelihood of recidivism. The fact that I found a sizable treatment effect despite the likely psychological strain from Katrina means that the effect of migration may in fact be underestimated. Still, the psycho- logical impact of Hurricane Katrina may bear upon recidivism, and it is a worthy topic of investigation.

While there are plenty of opportunities for future research on the effects of Katrina, as well as the relation between migration, place of res-

idence, and recidivism, findings thus far suggest that successful prisoner reentry and reintegra- tion depend on providing opportunities for pris- oners to separate from their criminogenic past. Such a finding provides initial support for the life-course propositions outlined earlier. If crim- inal peers and familiar criminogenic contexts causally influence criminal behavior, separating individuals from their peers and familiar con- texts should lead to a reduced likelihood of criminal behavior. That is precisely what I find. In a practical sense, to lessen the likelihood of recidivism and to foster the path to desistance, it is beneficial to separate ex-prisoners from their criminal past.

David S. Kirk is an Assistant Professor in the Department of Sociology and a Faculty Research Associate of the Population Research Center at the University of Texas at Austin. His current research explores the influence of social context and neigh- borhood change on criminal behavior. One ongoing project examines the structural and cultural predic- tors of neighborhood violence. Kirk’s recent research has appeared in Demography and Criminology.

APPENDIX: ASSUMPTIONS OF THE IV FRAMEWORK

We must consider potential violations of the IV framework before concluding that the treat- ment effect of residential migration described in Tables 2, 3, and 4 is in fact valid. As men- tioned previously, one key assumption is that the instruments and the outcome variable are unre- lated, except through the treatment condition (Angrist and Krueger 2001). This is known as the exclusion restriction. While the exclusion restriction is not directly verifiable, it is possi- ble to indirectly assess this assumption in overi- dentif ied models where there are more instruments than endogenous regressors. Specifically, for the results presented in Table 2, models are just identified—that is, the model has the same number of instruments as endoge- nous variables (one). Yet with five instruments for the fourth model estimated in Table 4, the model is overidentified. It is thus possible to indirectly test the exclusion restriction—that is, the instruments are uncorrelated with the model error—by correlating the residuals from the estimation of Equation 2 with the excluded instruments. I use Hansen’s (1982) J statistic (see also Baum, Schaffer, and Stillman 2003) to

502—–AMERICAN SOCIOLOGICAL REVIEW

test the joint null hypothesis that the instru- ments are in fact valid instruments (i.e., they are uncorrelated with the error term).16 The J sta- tistic for the fourth model estimated in Table 4 equals 5.996 ( p = .200). I thus fail to reject the null hypothesis, thereby providing necessary support for the exclusion restriction.

A second key assumption of the IV frame- work is that the covariance between the treat- ment and instrument differs from zero: cov(Si, Zi ) ≠ 0. If the instrument (i.e., post-Katrina release) does not affect residential migration, then use of Katrina as an instrument is inap- propriate. Results from Figure 1 support the assumption that the treatment and instrument co- vary, in that the spatial distribution of the parole population changed following Hur ricane Katrina. However, even if the treatment and instrument are significantly associated, prob- lems may arise in the second-stage estimation (Equation 2) if they are only weakly related (i.e., a nonzero yet small correlation). This is known as the weak instruments problem (Staiger and Stock 1997). A primary consequence is that weak instruments produce inconsistent IV estimators (Bound, Jaeger, and Baker 1995). Additionally, a weak instrument increases stan- dard errors of the IV estimates and therefore affects hypothesis testing.

To assess the potential for a weak instru- ment, it is useful to examine the explanatory power of the instrument in the first-stage (i.e., Equation 1) of the two-stage approach. For the re-incarceration model from Table 2, to assess the instrument’s explanatory power, I use an F-test of the significance of the instrument. Results reveal that the instrument, post-Katrina release, is signif icantly correlated with the treatment variable (F = 60.25; df = 1, 46; p < .001). Yet, a statistically significant F statistic may still represent bias in the IV estimator, so the F statistic must be compared against a suit- able threshold. Staiger and Stock (1997) sug- gest that an F statistic below 10 is indicative of a weak instrument. In the case here, the instrument post-Katrina release is statistical- ly signif icant and far exceeds Staiger and Stock’s threshold. Such findings alleviate con-

cerns over violating assumptions of the IV approach.17

REFERENCES

Abbott, Andrew. 1997. “On the Concept of Turning Point.” Comparative Social Research 16:85–105.

Angrist, Joshua D. 1990. “Lifetime Earnings and the Vietnam Era Draft Lottery: Evidence from Social Security Administration Records.” American Economic Review 80:450–77.

Angrist, Joshua D., Guido W. Imbens, and Donald B. Rubin. 1996. “Identification of Causal Effects Using Instrumental Variables.” Journal of the American Statistical Association 91:444–55.

Angrist, Joshua D. and Alan B. Krueger. 2001. “Instr umental Variables and the Search of Identif ication: From Supply and Demand to Natural Experiments.” Journal of Economic Perspectives 15:69–85.

Baum, Christopher F., Mark E. Schaffer, and Steven Stillman. 2003. “Instrumental Variables and GMM: Estimation and Testing.” Stata Journal 3:1–31.

Blau, Judith R. and Peter M. Blau. 1982. “The Cost of Inequality: Metropolitan Structure and Violent Crime.” American Sociological Review 47:114–29.

Bound, John, David A. Jaeger, and Regina M. Baker. 1995. “Problems with Instrumental Variables Estimation When the Correlation Between the Instruments and the Endogenous Explanatory Variable Is Weak.” Journal of the American Statistical Association 90:443–50.

Brown, Claude. 1965. Manchild in the Promised Land. New York: MacMillan.

C antor, D avid and Kenneth C . L and. 1985. “Unemployment and Crime Rates in the Post- World War II United States: A Theoretical and Empirical Analysis.” American Sociological Review 50:317–32.

Carter, Brian L. and Stephen T. Tiffany. 1999. “Meta- Analysis of Cue-Reactivity in Addiction Research.” Addiction 94:327–40.

Duncan, Otis D. and Beverly Duncan. 1955. “A Methodological Analysis of Segregation Indexes.” American Sociological Review 20:210–17.

Eisenstein, James, Roy Flemming, and Peter Nardulli.

RESIDENTIAL CHANGE AND RECIDIVISM—–503

16 Tests of IV assumptions are available in Stata through the ivreg2 function.

17 Another common strategy for assessing weak instruments is to examine the partial R2 of the first- stage equation, which isolates the squared partial correlation between the instrument and the endoge- nous regressor (Bound et al. 1995). The magnitude of the bias of IV estimators increases as the partial R2 approaches zero. First-stage results from the re- incarceration model in Table 2 reveal a partial R2 of .173.

1988. The Contours of Justice: Communities and Their Courts. Boston, MA: Little, Brown.

Elder, Glen H., Jr. 1974. Children of the Great Depression: Social Change in Life Experience. Chicago, IL: University of Chicago Press.

ESRI. 2006a. ESRI Demographic Update Methodology: 2006/2011. Redlands, CA: ESRI. Retrieved October 24, 2007 (http://www.esri.com/ library/whitepapers/pdfs/ demographic-update- methodology.pdf).

———. 2006b. ESRI Gulf Coast Updates Methodology: 2006/2011. Redlands, CA: ESRI. Retrieved October 24, 2007 (http://www.esri. c o m / l i b r a r y / wh i t e p a p e r s / p d f s / g u l f - c o a s t - methodology.pdf).

Federal Bureau of Investigation. 2007. Uniform Crime Reporting Program Data [United States]: County- Level Detailed Arrest and Offense Data, 2005 [Computer File]. Ann Arbor, MI: Inter-university Consortium for Political and Social Research.

Gar rett, Brandon L. and Tania Tetlow. 2006. “Criminal Justice Collapse: The Constitution after Hurricane Katrina.” Duke Law Journal 56:127–78.

Grogger, Jeff. 1998. “Market Wages and Youth Crime.” Journal of Labor Economics 16:756–91.

Hansen, Lars P. 1982. “Large Sample Properties of Generalized Method of Moments Estimators.” Econometrica 50:1029–54.

Heumann, Milton. 1975. “A Note on Plea Bargaining and Caseload Pressure.” Law & Society Review 9:515–28.

Hipp, John R. 2007. “Block, Tract, and Levels of Aggregation: Neighborhood Structure and Crime and Disorder as a Case in Point.” American Sociological Review 72:659–80.

Johnson, Brian D. 2005. “Contextual Disparities in Guidelines Depar tures: Cour troom Social Contexts, Guidelines Compliance, and Extralegal Disparities in Criminal Sentencing.” Criminology 43:761–96.

Langan, Patrick A. and David J. Levin. 2002. Recidivism of Prisoners Released in 1994. Washington, DC: Bureau of Justice Statistics.

Laub, John H. and Robert J. Sampson. 2003. Shared Beginnings, Divergent Lives: Delinquent Boys to Age 70. Cambridge, MA: Harvard University Press.

La Vigne, Nancy G., Cynthia A. Mamalian, Jeremy Travis, and Christy Visher. 2003. A Portrait of Prisoner Reentry in Illinois. Washington, DC: Urban Institute.

Ludwig, Jens and Jeffrey R. Kling. 2007. “Is Crime Contagious?” Journal of Law & Economics 50:491–518.

Massey, Douglas S. and Nancy A. Denton 1993. American Apartheid: Segregation and the Making of the Underclass. Cambridge, MA: Harvard University Press.

Merton, Robert K. 1938. “Social Structure and

Anomie.” American Sociological Review 3:672–82.

Mumola, Christopher J. and Jennifer C. Karberg. 2006. Drug Use and Dependence, State and Federal Prisoners, 2004. Bureau of Justice Statistics Special Report. Washington, DC: U.S. Department of Justice.

National Research Council. 2007. Parole, Desistance from Crime, and Community Integration. Committee on Community Supervision and Desistance from Crime. Committee on Law and Justice, Division of Behavioral and Social Sciences and Education. Washington, DC: The National Academies Press.

Osborn, S. G. 1980. “Moving Home, Leaving London and Delinquent Trends.” British Journal of Criminology 20:54–61.

Roman, Caterina Gouvis, Seri Irazola, and Jenny W. L. Osborne. 2007. After Katrina: Washed Away? Justice in New Orleans. Washington, DC: Urban Institute.

Rosenfeld, Richard, Joel Wallman, and Robert For nango. 2005. “The Contribution of Ex- Prisoners to Crime Rates.” Pp. 80–104 in Prisoner Reentry and Crime in America, edited by J. Travis and C. Visher. New York: Cambridge University Press.

Sabol, William J. and Heather Couture. 2008. Prison Inmates at Midyear 2007. Bureau of Justice Statistics Bulletin. Washington, DC: U.S. Department of Justice.

Sampson, Robert J. and John H. Laub. 1993. Crime in the Making: Pathways and Turning Points through Life. Cambridge, MA: Harvard University Press.

Sampson, Robert J. and William J. Wilson. 1995. “Toward a Theory of Race, Crime, and Urban Inequality.” Pp. 37–54 in Crime and Inequality, edited by J. Hagan and R. D. Peterson. Stanford, CA: Stanford University Press.

Shaw, Clifford R. and Henry D. McKay. 1942. Juvenile Delinquency and Urban Areas. Chicago, IL: University of Chicago Press.

Staiger, Douglas and James H. Stock. 1997. “Instrumental Variables Regression with Weak Instruments.” Econometrica 65:557–86.

Turner, Susan and Joan Petersilia. 1992. “Focusing on High-Risk Parolees: An Experiment to Reduce Commitments to the Texas Depar tment of Corrections.” Journal of Research in Crime and Delinquency 29:34–61.

Ulmer, Jeffer y and Brian D. Johnson. 2004. “Sentencing in Context: A Multilevel Analysis.” Criminology 42:137–75.

U.S. Depar tment of Housing and Urban Development. 2006. Current Housing Unit Damage Estimates: Hurricanes Katrina, Rita, and Wilma. Washington, DC: U.S. Department of Housing and Urban Development. Retrieved

504—–AMERICAN SOCIOLOGICAL REVIEW

December 8, 2006 (http://www.huduser.org/ publications/destech/GulfCoast_HsngDmgEst. html).

Visher, Christy and Jill Farrell. 2005. Chicago Communities and Prisoner Reentry. Washington, DC: Urban Institute.

Warr, Mark. 1993. “Age, Peers, and Delinquency.” Criminology 31:17–40.

———. 1998. “Life-Course Transitions and Desistance from Crime.” Criminology 36:183–216.

———. 2002. Companions in Crime: The Social Aspects of Criminal Conduct. Cambridge, UK: Cambridge University Press.

Wikler, Abraham. 1948. “Recent Prog ress in Research on the Neurophysiologic Basis of Mor phine Addiction.” American Journal of Psychiatry 105:329–38.

RESIDENTIAL CHANGE AND RECIDIVISM—–505