Presented By: The Center for the Study of Complex Systems
Bridging the Gap: Statistical Methods and Agent-Based Modeling in Social Epidemiology
Sarah Cherng, School of Public Health
CSAAW Event
ABSTRACT
Calls for greater integration of complex systems methods in the study of social factors contributing to health have gone largely unanswered due to the difficulty of integrating data with computational modeling methods in these contexts. This study utilizes a generalized linear mixed-effects model of risk factors at the school and individual levels for smoking experimentation in order to present parameter estimates for an agent-based model to identify the mechanisms of social connectedness that contribute to smoking initiation. While many studies identify adolescent popularity as a risk factor for smoking initiation, none have identified the specific characteristics of network connectivity that are responsible for the perpetuation of smoking behavior. This attempt to integrate modeling methods presents a potential solution for the integration of computational modeling with traditional statistical methods prevalent among social epidemiologists. The results from this study emphasize the critical importance of accounting for the behavior of immediate friends, particularly when considering global centrality network measures as a proxy for measuring popularity.
ABSTRACT
Calls for greater integration of complex systems methods in the study of social factors contributing to health have gone largely unanswered due to the difficulty of integrating data with computational modeling methods in these contexts. This study utilizes a generalized linear mixed-effects model of risk factors at the school and individual levels for smoking experimentation in order to present parameter estimates for an agent-based model to identify the mechanisms of social connectedness that contribute to smoking initiation. While many studies identify adolescent popularity as a risk factor for smoking initiation, none have identified the specific characteristics of network connectivity that are responsible for the perpetuation of smoking behavior. This attempt to integrate modeling methods presents a potential solution for the integration of computational modeling with traditional statistical methods prevalent among social epidemiologists. The results from this study emphasize the critical importance of accounting for the behavior of immediate friends, particularly when considering global centrality network measures as a proxy for measuring popularity.
Related Links
Explore Similar Events
-
Loading Similar Events...