BIOSTATISTICS FALL FORUM 2011

November 16, 2011

Coker Life Science Building (RM 510)
715 Sumter Street

by: David Darmofal
Associate Professor
Department of Political Science
University of South Carolina

"Bayesian Spatial Survival Modeling with an Application to Political Science Data"

Abstract: Given that many of the research questions of interest to epidemiologists and political scientists are similar, the two disciplines could benefit from a greater dialogue with each other. This talk seeks to promote this dialogue by examining how methods of interest to epidemiologists can be applied to political science data. Specifically, both epidemiologists and political scientists have strong interests in explaining event processes. Both epidemiological and political science survival data often carry a strong spatial component that is central to explaining the timing of events. This talk examines how Bayesian spatial survival models can be applied to political science data. The talk examines both semiparametric and parametric survival models in which spatial autocorrelation is modeled via random effects at neighboring locations.