Date(s) - Apr 12, 2011
1:00 PM - 2:30 PM
Center for Community Health
Professor, UCLA Department of Biostatistics
The modern statistical toolbox is incomplete without Bayesian methods. This talk presents an introduction to Bayesian methods for general audiences. The talk is at a very general level and does not assume much more than passing familiarity with statistics and requires no previous knowledge of Bayesian statistics. We talk about the benefits and strengths of Bayesian statistics and illustrate with a key example that everyone will follow. Bayesian inference starts with (i) a prior distribution that numerically represents the information we have about a particular problem before we collect data and (ii) a model for the data. We sample data, then update the prior distribution using the data and the model to produce a posterior distribution which represents the inferences from our analysis. Bayesian methods are illustrated in the setting of estimating the speaker’s blood pressure. We will discuss priors for blood pressure, then collect some data and illustrate the conclusions that we draw. A homework and excel spreadsheet are available (and should be attached to this notice) to allow you to illustrate the methods step by step with your own blood pressure data. We then illustrate how much better Bayesian inferences are compared to classical inferences using data from the Nurse’s Blood Pressure Study. We illustrate the generality of Bayesian analyses in the context of a linear regression analysis.