Multilevel Latent Class and Social Network Models for Observational Adolescent Obesity Data

Research Question
All of the research questions are addressed using survey data collected in EAT 2010 and Project F-EAT (PI: Neumark-Sztainer) which includes 2500 adolescents (ages 11-18) and their parents from urban neighborhoods in Minneapolis/St. Paul, Minnesota.

  1. Are there obesogenic clusters of behaviors and environments in adolescents that can be empirically identified by summarizing neighborhood, family, and peer characteristics data collected on adolescents and their parents?
  2. How do adolescents' peers influence physical activity, nutrition, and weight control behaviors?
  3. Does having a supportive family environment buffer the effects of an unsupportive neighborhood on weight and health behaviors?

Modeling Approach
Because this is a methodological grant, the following list represents general methodology that is being developed. The methods are also applied to address the specific research questions above regarding adolescent obesity.

  1. To develop multilevel statistical models that can incorporate latent class variables as well as cross-cluster correlations specifically geographic correlation induced by neighborhood level information.
  2. To embed social-network information into a multilevel statistical modeling framework that uses statistically sound estimation methods, and also to develop guidelines for sampling peer-networks within schools including sample size and power considerations.
  3. To investigate the theoretical justification and empirical utility of incorporating multiple propensity scores for multiple causal predictors with flexibility to handle non-dichotomous and multilevel predictors.

Principal Investigator

Melanie M. Wall, PhD
Professor
Biostatistics Columbia University

Co-Investigators

Rich Maclehose, PhD
Division of Epidemiology and Community Health School of Public Health, Division of Biostatistics School of Public Health
University of Minnesota

David Knoke, PhD
Department of Sociology
University of Minnesota

Peter Hannan, MStat
Division of Epidemiology and Community Health School of Public Health
University of Minnesota

Mary Story, PhD RD
Division of Epidemiology and Community Health School of Public Health
University of Minnesota

Dianne Neumark-Sztainer, PhD MPH RD
Division of Epidemiology and Community Health School of Public Health
University of Minnesota