Bayesian Analyses Using SAS

The course focuses on Bayesian analyses using the PHREG, GENMOD, and MCMC procedures. The examples include logistic regression, Cox proportional hazards model, general linear mixed model, zero-inflated Poisson model, and data containing missing values. A Bayesian analysis of a crossover design and a meta-analysis are also shown.

Learn how to

• explain the concepts of Bayesian analysis
• illustrate Bayesian analyses in PROC GENMOD, PROC PHREG and PROC MCMC
• incorporate prior distributions in a Bayesian analysis
• illustrate a Bayesian analysis approach to a meta-analysis.

Who should attend

Biostatisticians, epidemiologists, and social scientists who are interested in the Bayesian analysis approach

Formats available	Duration
Classroom:	2.0 days

Before attending this course, you should

• be able to create SAS data sets and manipulate data. You can gain this experience from the SAS Programming 2: Data Manipulation Techniques course.
• have completed a statistics course such as the Statistics 1: Introduction to ANOVA, Regression, and Logistic Regression or the Statistics 2: ANOVA and Regression courses.

This course addresses SAS/STAT software.

This course is offered on the Windows platform only using SAS/STAT 14.2.

Introduction to Bayesian Analysis

• introduce the basic concepts of Bayesian analysis
• compute the diagnostic plots and diagnostic statistics for model assessment
• discuss the advantages and disadvantages of Bayesian analysis
• illustrate a Bayesian analysis in PROC GENMOD and PROC PHREG

Fitting Models with the MCMC Procedure

• show the essential statements in PROC MCMC
• show the supported distributions in PROC MCMC
• fit a logistic regression model in PROC MCMC
• fit a general linear mixed model in PROC MCMC
• fit a zero-inflated Poisson model in PROC MCMC
• incorporate missing values in PROC MCMC

Bayesian Approaches to Clinical Trials

• use prior distributions in a Bayesian analysis
• illustrate a Bayesian approach to clinical trials using PROC MCMC
• illustrate the Bayesian approach to meta-analysis