The Mobility of Youth in the Justice System: Implications for Recidivism

While each are quasi-experimental, selection effects still exist, as families who chose to participate in both the Moving to Opportunities and the Yonkers Project identified safety (such as escaping gangs and drugs) as the primary reason for wanted to relocate from their high-poverty neighborhoods (Fauth et al. 2005). The Moving to Opportunities and Yonker’s Project studies differ in the following way: the Yonkers Project is stated to have received greater public attention, and (mostly white) residents in the new neighborhoods were aware of their new neighbors, and the townhomes erected for the desegregation purpose; whereas Moving to Opportunities participants were provided vouchers to move to lower-poverty areas of their choice (Fauth et al. 2005).

See Keels et al. (2005) for description of the Gautreaux program. Notable implications for adults involved persistent improvements in neighborhood quality for up to 15 years after families were placed in the program. Furthermore, placement anywhere but the most minority-segregated neighborhoods was found to be associated with a drop in current-neighborhood violent crime rates.

Kling et al. (2005) conclude the increased property offending of male experimental group youth was a result of their taking advantage of a comparative advantage in property offending in their new, more affluent neighborhoods, and not due to gender differences in mobility patterns out of disadvantaged urban areas, or gender differences in discrimination. Succinctly, there was simply more nice stuff to steal in then new neighborhoods, and males were more likely to take advantage of that fact. Kling et al. (2005) also hypothesized the newness wears off in time, which is why differences in property offending between experimental and control group males attenuated over time.

Prior Moving to Opportunities evaluations have controlled for pre-program arrests of participants (e.g., Ludwig et al. 2001). However, prior studies have not examined previously adjudicated youth as a distinct sub-sample. To our knowledge the current study is the first to examine residential mobility in an exclusively adjudicated delinquent population.

Excluding youth without a risk/need assessment reduced the sample by approximately 6000 youth. However, risk/need assessment information is critical to ensuring control of pre-existing differences between movers and non-movers. Importantly, youth not assessed are those individuals who were never formally processed at a juvenile assessment center. These youth were given a notice to appear in court and the judge and/or prosecuting attorney decided against formal processing. As our focus is juveniles involved in the juvenile justice system, we believe that including these youth would distort the pattern of relationships we address. We believe that the results presented here accurately represent the impact of residential relocation on juvenile recidivism for a sample of juvenile justice system involved youth.

In the event a juvenile relocates after community-based services are completed, a new address will be entered in the event the juvenile is re-arrested; meaning the current study includes every address for each arrest of all youth, but may not contain an address if a youth completed community-based services, moved, and was not arrested again within the 2-year period.

To assess the effect of this decision on sample composition and the results of our analysis we compared youth who moved once to those youth who moved two-or-more times using a series of Chi square and difference-in-means tests. Results of these analyses indicate that youth who moved more than once face a number of challenges above those of youth who moved only one time. Specifically multiple movers were more likely to have experienced parental drug or employment problems, be using drugs themselves, have antisocial friends, and have experienced physical or sexual abuse. Youth who moved multiple times were also disproportionately Black and male. In ancillary analyses (not presented) we ran an additional regression model which included an ordinal measure representing the number of times a youth moved (0-8) during the follow up period. Results of this analysis indicate that the number of times a youth moved was positively and significantly related to reoffending among the full sample, indicating that frequent relocation may have additional implications for juvenile delinquency.

Initial efforts were made to use two standard deviations for indications of upward or downward mobility. However, this resulted in almost all youth being lateral movers. Therefore, the decision was made to use a one standard deviation change in disadvantage to introduce variability into changes in neighborhood conditions (and also illustrates the general absence of changes in socioeconomic conditions among a statewide sample of juvenile offenders with respect to residential census tracts).

Mental health problems include formal diagnoses of schizophrenia, bi-polar, mood, thought, personality or adjustment disorders. Conduct disorder and oppositional defiant are not included (based on the risk/needs instrument from which the data were gleaned), which is addressed in the limitations section of the current study. Additionally, substance abuse diagnoses were not included (based on the risk/needs tool), though we have included separate measures of current alcohol and drug use.

Given the large number of covariates included in the models presented, we first assessed the degree of collinearity present. Results of these analyses (not presented) indicate that collinearity is not an issue, as no correlations between the measures used were above .500.