Academic Risks, Neighborhood Characteristics and School Suspensions for Youth on Probation: A Spatial Dependence Analysis

Background/purpose:

School connections have been identified as an important prosocial resource for at-risk youth, which may also be a malleable target for interventions. School behavior has been found to be influenced more by neighborhood factors than family processes (Bowen, Bowen, & Ware, 2002). While social disorganization theory holds that neighborhood factors can produce geographically concentrated levels of risk and delinquency, methods assessing spatial dependence for clustering on maps are seldom employed in the literature. While spatial dependence of neighborhood factors affecting juvenile reentry rates have been examined (Abrams & Freisthler, 2010), the literature is scant on justice-involved youth, meaning there are many unanswered questions about how geographical trends and clustering may affect their outcomes. The aim of this study is to examine how school and neighborhood factors affect rates of suspensions for youth on probation, and to determine if the results are spatially dependent.

Methods:

Data were gathered and aggregated at the zip code level from the Youth Assessment Screening Inventory records of youth on probation (2009-2011) from a large, diverse, urban jurisdiction in the Midwest (N=5,279). The sample was 89% male and 11% female; 6% White, 76% African American (AA), 13% Hispanic,  4% Mixed & 2% Other; and included ages 12-17. Geographic and demographic neighborhood characteristics are derived from census data for 2011. Using OLS regressions, we tested school (special education status, school problems, and school affect) and neighborhood factors (housing vacancy, poverty, unemployment, and demographics), by themselves and combined, and their effects on in-school and out-of-school suspensions. Spatial error regression models were also ran with the same variables to examine spatial dependence.

Results:

Both school affect and special education status predicted in-school suspensions, however, there was no spatial dependence and neighborhood factors were not significant. Out-of-school suspensions were related to school problems and the percentage of non-white residents; both results were spatially dependent.

Conclusion/Implications:

Results suggest some individual academic characteristics (special education status, school affect) are risk factors for in school suspension, regardless of location. The lack of spatial dependence could be viewed as a positive indicator: structural neighborhood factors do not affect school disciplinary problems for this population, bolstering the potential of school engagement as one avenue to improve youth outcomes. The finding that the percentage of nonwhite residents and school problems was significant for out-of-school suspensions (but not in-school suspensions) suggests that disparate use of more severe disciplinary approaches may be used with African American and Latino students in our sample, which is also already evident in the broader school going population in the United States (Finn & Servoss, 2013). These results strengthen the case for alternative behavior interventions, such as mindfulness-based behavioral interventions (Mendelson et al., 2010), which are gaining recognition for their utility in high-risk schools. This study also shows the importance of taking spatial dependence into account when examining place-based environmental risk factors to assess what level they may be disproportionately concentrated on the map and population level.