On Using Bayesian Methods to Address Small Sample Problems

Research output: Contribution to journalArticle

56 Citations (Scopus)

Abstract

As Bayesian methods continue to grow in accessibility and popularity, more empirical studies are turning to Bayesian methods to model small sample data. Bayesian methods do not rely on asympotics, a property that can be a hindrance when employing frequentist methods in small sample contexts. Although Bayesian methods are better equipped to model data with small sample sizes, estimates are highly sensitive to the specification of the prior distribution. If this aspect is not heeded, Bayesian estimates can actually be worse than frequentist methods, especially if frequentist small sample corrections are utilized. We show with illustrative simulations and applied examples that relying on software defaults or diffuse priors with small samples can yield more biased estimates than frequentist methods. We discuss conditions that need to be met if researchers want to responsibly harness the advantages that Bayesian methods offer for small sample problems as well as leading small sample frequentist methods.

Original languageEnglish (US)
Pages (from-to)750-773
Number of pages24
JournalStructural Equation Modeling
Volume23
Issue number5
DOIs
StatePublished - Sep 2 2016
Externally publishedYes

Fingerprint

Bayesian Methods
Small Sample
Specifications
Estimate
Small Sample Size
Prior distribution
Accessibility
Data Model
Empirical Study
Biased
Bayesian methods
Small sample
Continue
Specification
Software
popularity
Simulation
simulation

Keywords

  • Bayes
  • prior distribution
  • small sample

ASJC Scopus subject areas

  • Decision Sciences(all)
  • Modeling and Simulation
  • Sociology and Political Science
  • Economics, Econometrics and Finance(all)

Cite this

On Using Bayesian Methods to Address Small Sample Problems. / McNeish, Daniel.

In: Structural Equation Modeling, Vol. 23, No. 5, 02.09.2016, p. 750-773.

Research output: Contribution to journalArticle

@article{13cf12aa3f1d434284fc3f0d631f598d,
title = "On Using Bayesian Methods to Address Small Sample Problems",
abstract = "As Bayesian methods continue to grow in accessibility and popularity, more empirical studies are turning to Bayesian methods to model small sample data. Bayesian methods do not rely on asympotics, a property that can be a hindrance when employing frequentist methods in small sample contexts. Although Bayesian methods are better equipped to model data with small sample sizes, estimates are highly sensitive to the specification of the prior distribution. If this aspect is not heeded, Bayesian estimates can actually be worse than frequentist methods, especially if frequentist small sample corrections are utilized. We show with illustrative simulations and applied examples that relying on software defaults or diffuse priors with small samples can yield more biased estimates than frequentist methods. We discuss conditions that need to be met if researchers want to responsibly harness the advantages that Bayesian methods offer for small sample problems as well as leading small sample frequentist methods.",
keywords = "Bayes, prior distribution, small sample",
author = "Daniel McNeish",
year = "2016",
month = "9",
day = "2",
doi = "10.1080/10705511.2016.1186549",
language = "English (US)",
volume = "23",
pages = "750--773",
journal = "Structural Equation Modeling",
issn = "1070-5511",
publisher = "Psychology Press Ltd",
number = "5",

}

TY - JOUR

T1 - On Using Bayesian Methods to Address Small Sample Problems

AU - McNeish, Daniel

PY - 2016/9/2

Y1 - 2016/9/2

N2 - As Bayesian methods continue to grow in accessibility and popularity, more empirical studies are turning to Bayesian methods to model small sample data. Bayesian methods do not rely on asympotics, a property that can be a hindrance when employing frequentist methods in small sample contexts. Although Bayesian methods are better equipped to model data with small sample sizes, estimates are highly sensitive to the specification of the prior distribution. If this aspect is not heeded, Bayesian estimates can actually be worse than frequentist methods, especially if frequentist small sample corrections are utilized. We show with illustrative simulations and applied examples that relying on software defaults or diffuse priors with small samples can yield more biased estimates than frequentist methods. We discuss conditions that need to be met if researchers want to responsibly harness the advantages that Bayesian methods offer for small sample problems as well as leading small sample frequentist methods.

AB - As Bayesian methods continue to grow in accessibility and popularity, more empirical studies are turning to Bayesian methods to model small sample data. Bayesian methods do not rely on asympotics, a property that can be a hindrance when employing frequentist methods in small sample contexts. Although Bayesian methods are better equipped to model data with small sample sizes, estimates are highly sensitive to the specification of the prior distribution. If this aspect is not heeded, Bayesian estimates can actually be worse than frequentist methods, especially if frequentist small sample corrections are utilized. We show with illustrative simulations and applied examples that relying on software defaults or diffuse priors with small samples can yield more biased estimates than frequentist methods. We discuss conditions that need to be met if researchers want to responsibly harness the advantages that Bayesian methods offer for small sample problems as well as leading small sample frequentist methods.

KW - Bayes

KW - prior distribution

KW - small sample

UR - http://www.scopus.com/inward/record.url?scp=84974808077&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84974808077&partnerID=8YFLogxK

U2 - 10.1080/10705511.2016.1186549

DO - 10.1080/10705511.2016.1186549

M3 - Article

AN - SCOPUS:84974808077

VL - 23

SP - 750

EP - 773

JO - Structural Equation Modeling

JF - Structural Equation Modeling

SN - 1070-5511

IS - 5

ER -