|InterJournal Complex Systems, 529
|Manuscript Number: |
Submission Date: 20501
|The Sum of the Parts: Two Studies of Interpersonal Influence on Aggressive Behaviors in Small Group Settings|
The Sum of the Parts: Two Studies of Interpersonal Influence on Aggressive Behaviors in Small Group Settings Keith Warren The Ohio State University College of Social Work Elena Irwin The Ohio State University Department of Agricultural Economics Brian Roe The Ohio State University Department of Agricultural Economics William Sean Newsome The Ohio State University College of Social Work Economists and other social scientists have recently shown considerable interest in interactions-based models that address the question of how often and when an individualís choices depend on those of peers (Brock & Durlauf, 2000; Manski, 1995). Most of these studies have focused on neighborhoods and schools as the units of analysis, in an attempt to understand interactions among large numbers of individuals. However, most social interventions occur in classrooms, small groups or dyads, and there is a need to model interactions in these smaller-scale settings. This paper describes studies that attempt to model the effect of interpersonal interactions on aggressive behavior in two different settings, the first a sample of elementary school classrooms and the second a group home for developmentally disabled adults. In each case we argue that the value of aggressive behavior will depend on whether others are behaving aggressively; aggression will, therefore, tend to breed aggression. In the first study, we employ a logistic regression model and find that the average level of aggression in elementary school classrooms, measured by the teacher, is positively correlated with clinically significant levels of aggression in individual children, measured by their parents, both concurrently and at one and two year follow-up. This correlation remains statistically significant when controlling for family and neighborhood poverty, family conflict, gender, academic achievement and level of aggression in the previous time period. This suggests that early classroom-based interventions might have a lasting effect on aggressive behaviors in children. In the second study, we use multivariate nonlinear time series analysis to study the interactions of two individuals, one male and one female, living together in a group home. We find evidence of correlated volatility in their aggressive behaviors. In this case, the volatility of the female residentís behavior is negatively correlated with that of the male resident on the next time lag. The volatility of the male residentís behavior is positively correlated with that of the female resident. This mix of positive and negative correlation implies that staff interventions can have unexpected iatrogenic effects at times of resident behavioral volatility. This presentation has methodological as well as substantive implications. It emphasizes the critical importance of choosing the correct unit of analysis, whether it is neighborhood, classroom, or peer group, if we wish to find evidence of interpersonal interactions. Further, some types of interactions, such as correlated volatility between individuals, may only be possible to document in a nonlinear time series framework. Brock, W. & Durlauf, S. (2000). Interactions-based models. In Heckman, J. & Leamer, E. (eds) Handbook of Econometrics, 5th Edition. Amsterdam: North Holland. Manski, C. (1995). Identification Problems in the Social Sciences. Cambridge, MA: Harvard University Press.
|Submit referee report/comment|