Bayesian Statistics Seminar
North Carolina State University
presents
Dr. Hrishikesh Chakraborty
Research Traiangle Institute
"Bivariate Random Effect Models -- A Frequentist and Bayesian Approaches to Evaluate Relationship"
ABSTRACT
Analysis of correlated data in a longitudinal setting is quite popular in epidemiological and clinical research. Bivariate mixed models are commonly used to model correlated data assuming that the random effects and within subject errors are normally distributed to evaluate relationship. However, the normality assumption may not always give reliable results, particularly if the data exhibits skewness. We may use a bayesian approach to bivariate mixed model and relax the normality assumption by using a multivariate skew-normal distribution. HIV RNA study data will be used to demonstrate the two different types of modeling techniques.
Tuesday, November, 8, 2005
4:00 - 5:00 pm
208 Patterson Hall