What Can Bayesian Inference Do for Accounting Research?

Abstract

Bayesian statistics is a framework for combining new data with existing forms of information to yield more precise inferences than is possible using the data alone. Its greatest practical advantages are the flexibility it offers in incorporating prior information and beliefs, modeling heterogeneity, modeling latent constructs, and combining multiple data sources. The goal of this paper is twofold: first, to introduce accounting researchers to Bayesian inference and distinguish it from classical frequentist inference and, second, to showcase when Bayesian modeling can improve inferences in many applications that are of interest to accounting researchers.

Keywords: Accounting research, Bayesian statistics, quantifying uncertainty, measuring latent constructs

JEL Classification: C11, C53, G17, M40