Biostatistics Seminar

Lawrence McCandless, PhD

Associate Professor, Faculty of Health Sciences, Simon Fraser University

Unmeasured confounding, large datasets, and the role of Bayesian statistics

Abstract:

Unmeasured confounding creates terrible problems in observational studies using large administrative databases.  The massive sample size crushes p-values and standard errors to zero that are calculated from standard biostatistical techniques.  While this may delight researchers who discover that everything is significant, it obscures the role of bias, including unmeasured confounding.  The Bayesian approach to statistics provides an appealing way forward because uncertainty about bias can be brought into the analysis using prior distributions.  In this talk I will illustrate Bayesian sensitivity analysis for unmeasured confounding in observational studies using administrative data from British Columbia.  I will show how to use Stan, which is new software developed by Andrew Gelman and others.  Stan allows the careful study of posterior distribution in a vast collection of Bayesian models, including nonidentifiable models for bias in epidemiology, which are poorly suited to conventional Gibbs sampling. 

Bio:

Lawrence McCandless is associate professor in the Faculty of Health Sciences at Simon Fraser University.  His research interests include epidemiology and Bayesian causal inference.