The Association of Neighborhood, Family, and Friends with Pre-Adolescent Aggression

Background: Risk and protective factors for the development of aggressive behavior in youth exist in the neighborhood, family, and peer systems. Negative behaviors of teenagers in the neighborhood, and more proximal negative friend behaviors have consistently been identified as neighborhood and friend risk factors. Caring parents and neighborhood adults, in contrast are promotive of prosocial behavior. Although a subset of youth begin engaging in aggressive behaviors before age 10, less is known about the relative effects of social environmental factors on aggression in middle childhood (~age 6 to 11). The current study builds on existing research by examining the effects on aggression of neighborhood, family, and friend factors among 3^rd through 5^th graders using perspectives of parents and children and sophisticated latent variable modeling procedures appropriate for ordinal, clustered data. It was hypothesized that due to their age, children’s aggression would be directly related to their perceptions of parents and friends, with neighborhood characteristics less influential and only indirectly related.

Methods

Sample. Data from ESSP projects in 2008-2009 in four diverse school districts and 13 schools was combined to create a dataset with 1251 child cases. About half of the sample (48%) was male, 39% was African American, 7% was Latino, and 46% was European American. Forty-three percent of the sample took part in the school lunch program.

Measures. Three child-report and two-parent report scales were used in the analysis: Seven items measured child perceptions of caring adults in the neighborhood. Six measured child perceptions of caring parents, and five items measured child perceptions of negative behaviors (e.g., lying, hitting) among friends. The behavior of teens in the neighborhood was measured with four parent-report items, as was the dependent aggression variable. Child gender, race/ethnicity, and free/reduced lunch program participation were included as control variables.

Analysis. Mplus 6.11 was used to test latent variable models using Weighted Least Square (WLSMV) with a polychoric correlation matrix of the ordinal observed variables and a correction for the clustering of children in schools. Full information Maximum Likelihood allowed the inclusion of cases with missing values.

Results. The final model had excellent fit (RMSEA = .013, CI .008 to .017; CFI = .979; TLI = .977). Children’s perceptions of their friends’ negative behaviors and parents’ caring had direct effects on aggression. Parent caring also had an indirect effect on aggression through friends’ behavior. The statistical effects of caring neighborhood adults were fully mediated by their effects on parent caring. However, although some of the effects of neighborhood teen behavior were mediated through children’s friends’ behavior, this negative neighborhood characteristic also had direct effects on child aggression.

Discussion. Factors in all three domains are related to pre-adolescent aggression, suggesting that social work interventions need to target multiple domains. Caring neighbors may promote caring parenting, which in turns reduces childhood aggression directly and indirectly by reducing the influence of negative friends. Negative teens may increase the likelihood of children having older role models with negative behaviors, as well as directly promoting aggressive behaviors among children.