A Web-Based Application to Support Applied Bayesian Data Analysis

• In Event: Statistical Inference and Bayesian Estimation

Abstract

I present a freely accessible web-based Bayesian data analysis application built on the Stan computation engine. The goal of the application is to make Bayesian data analysis accessible while demonstrating two practices of applied Bayesian data analysis. The first practice is that of including subjective prior information in data analysis. The second practice is communicating study effects probabilistically using posterior samples. To demonstrate these practices, the application elicits prior information from the user and returns plots that communicate effects probabilistically. The application implements two-group comparisons for: unbounded continuous outcomes (t-regression and quantile regression); bounded continuous outcomes (beta regression); and binary outcomes (binomial regression). The application is under continuous development; by the date of presentation, the application will include additional techniques.

Author

• James O. Uanhoro, University of North Texas