Few studies of contextual effects attempt to model the effects of schools and neighborhoods at the same time, or to explore the differential impact of these contexts over the course of adolescent development. This study seeks to fill this void in research on development by simultaneously modeling the effects of school and neighborhood environments on youth outcomes, focusing specifically on juvenile arrest and school dropout. The goal of the study is to examine whether neighborhoods and schools matter in explaining dropout and arrest, net of neighborhood and school differences in social composition, and to explain exactly why school and neighborhood conditions influence youth outcomes by determining what specific mechanisms or processes within these contexts influence these delinquent outcomes. This assessment of processes tests and augments prior work related to the concepts of social capital, relational trust, and collective efficacy. This study is also an advancement over previous research by integrating the neighborhood effects and school effects approaches, paying particular attention to the cumulative consequences of each on developmental trajectories. This study also has important implications for policy and practice, such as quantifying the benefits of building social capital or collective efficacy in neighborhoods, and identifying the extent to which providing different educational alternatives may benefit youth.
To achieve these objectives, a multi-wave, multilevel design will be employed to combine individual-level data from the Project on Human Development in Chicago Neighborhoods (PHDCN) Cohort surveys, the Chicago Police Department, and the Illinois State Police with contextual data from the U.S. census, the Chicago Public Schools, and the PHDCN Community Surveys. Participants include three cohorts of Chicago youth, aged 9, 12, and 15 at the time of the first PHDCN interview in 1994. The third and final wave of data collection was completed in 2001. One key methodological benefit of this research is the combined use of self-report and official crime data. This study will utilize recent advances in multilevel regression modeling, and employ a growth curve design to model juvenile arrests and a discrete-time event history model to estimate the probability of school dropout.