The effects of neighborhood context on the development and experiences of children and youth increasingly have been emphasized. Yet although ethnographic studies have revealed that youth spend their time in multiple neighborhoods (Burton, 2001), the literature has neglected to consider neighborhood contexts other than those of youths' home neighborhoods. Ours is among the first study to examine school neighborhoods for their effects on youth development. In particular, this study examines whether the impacts of violence exposure and disadvantage on youth problem behavior and mental health vary by whether they occur within home neighborhoods or within schools and their neighborhoods.
Respondents in this study were 742 youth from a planned follow-up survey of participants in a large school-based violence prevention program in New York City (NYC). In addition to demographic variables, youth-reported received parenting and exposure to violence (at or near home or school) were used as predictors. The outcome variables, conduct disorder and major depressive disorder, were measured using symptom counts from the Diagnostic Interview Schedule for Children-IV. Individual respondent data were merged with data on NYC neighborhoods (for home and school) as well as with characteristics of schools. In order to correct standard errors for nested data and to nest children within their home neighborhoods and schools/school neighborhoods simultaneously, the analyses involved cross-classified multilevel models (Hox, 2002).
Although home neighborhood, school, and school neighborhood factors were not highly predictive of adolescent mental health, variation in outcomes by contextual level was relatively high. With no independent variables included in the model, 8.0% of the variance in conduct disorder was at the school level and 4.4% of the variation was at the home neighborhood level. Only 2.2% of the variance in depressive symptom count was at the school level and 2.4% was at the home neighborhood level. These percentages were not reduced when individual-level predictors were included, suggesting that experiences of schools, school neighborhoods, and home neighborhoods vary a great deal for individuals.
Of the individual, home, and school factors, individual violence victimization was most predictive of conduct disorder symptoms. For each additional type of victimization, the symptom count increases by 2.2 (p<.001). Additionally, the total number of experiences of violence in the home neighborhood increased conduct disorder symptoms. Punitive parenting practices were highly predictive of conduct disorder. Individual violence victimization was also highly predictive of depressive symptoms (coef = 1.8, p<.001). Violence experiences in the home, home neighborhood, and school neighborhood all significantly contributed to depressive symptoms.
Conclusions and Implications
School and home neighborhood characteristics were poor predictors of adolescent conduct disorder and depression, but experiences of violence were highly predictive of both conduct disorder and depression. These results confirm that proximal indicators of violence exposure are more important predictors of youth conduct disorder and depression than more distal indicators of neighborhood and school contexts. Efforts to improve neighborhoods would thus be expected to improve youth mental health and behavior only if they reduce youth exposure to violence.