Using Bayes Factors to Get the Most out of Linear Regression: A Practical Guide Using R
Year: 2015
Type: article
Abstract: This guide is for readers who want to make rich inferences from their data. It does not contain anything new with regard to R code or theoretical development, but it does piece together information in an easy to follow guide. An article was recently published in a journal that is probably not well known by most researchers, Multivariate Behavioral Research, where the authors discuss the theoretical background of Bayesian inference and provide a simple web calculator for linear regression ( Rouder & Morey, 2012; http://pcl.missouri.edu/bf-reg). More recently, R code has been developed (Morey, Rouder , & Jamil, 2015) and compiled to conduct many different types of Bayesian analyses (Morey, 2015; http://bayesfactorpcl.r-forge.r-project.org/ ) . However, this leaves information scattered across websites and journal pages, and so it is not easily taken in by a researcher who doesn't know where to start with either Bayesian theory or R software. This guide has been put together in a way that a researcher who has never used Bayesian inference or R software could conduct, report, and interpret a Bayesian linear regression. It skips over the complicated derivations and mathematics behind the theory, and focuses on why Bayes factors are valuable and desirable to researchers. It then shows step-by-step how R software can be used for Bayesian model comparison as well as parameter estimation in linear regression problems.
Source: The Winnower
Author: Alexander Etz
Language: English
Subfield: Statistics and Probability
Field: Mathematics
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W2157378142