A Method for Constructing Informative Priors for Bayesian Modeling of Occupational Hygiene Data

Harrison Quick, Tran Huynh, Gurumurthy Ramachandran

Research output: Contribution to journalArticle

Abstract

In many occupational hygiene settings, the demand for more accurate, more precise results is at odds with limited resources. To combat this, practitioners have begun using Bayesian methods to incorporate prior information into their statistical models in order to obtain more refined inference from their data. This is not without risk, however, as incorporating prior information that disagrees with the information contained in data can lead to spurious conclusions, particularly if the prior is too informative. In this article, we propose a method for constructing informative prior distributions for normal and lognormal data that are intuitive to specify and robust to bias. To demonstrate the use of these priors, we walk practitioners through a step-by-step implementation of our priors using an illustrative example. We then conclude with recommendations for general use.

Original languageEnglish (US)
Pages (from-to)67-75
Number of pages9
JournalAnnals of Work Exposures and Health
Volume61
Issue number1
DOIs
StatePublished - Jan 1 2017

Fingerprint

Bayes Theorem
Normal Distribution
Statistical Models
Hygiene

Keywords

  • decision making
  • hierarchical modeling
  • prior sample size
  • sparse data
  • truncated priors

ASJC Scopus subject areas

  • Public Health, Environmental and Occupational Health

Cite this

A Method for Constructing Informative Priors for Bayesian Modeling of Occupational Hygiene Data. / Quick, Harrison; Huynh, Tran; Ramachandran, Gurumurthy.

In: Annals of Work Exposures and Health, Vol. 61, No. 1, 01.01.2017, p. 67-75.

Research output: Contribution to journalArticle

@article{75ab0c56aaee4e019294d564aaf028c6,
title = "A Method for Constructing Informative Priors for Bayesian Modeling of Occupational Hygiene Data",
abstract = "In many occupational hygiene settings, the demand for more accurate, more precise results is at odds with limited resources. To combat this, practitioners have begun using Bayesian methods to incorporate prior information into their statistical models in order to obtain more refined inference from their data. This is not without risk, however, as incorporating prior information that disagrees with the information contained in data can lead to spurious conclusions, particularly if the prior is too informative. In this article, we propose a method for constructing informative prior distributions for normal and lognormal data that are intuitive to specify and robust to bias. To demonstrate the use of these priors, we walk practitioners through a step-by-step implementation of our priors using an illustrative example. We then conclude with recommendations for general use.",
keywords = "decision making, hierarchical modeling, prior sample size, sparse data, truncated priors",
author = "Harrison Quick and Tran Huynh and Gurumurthy Ramachandran",
year = "2017",
month = "1",
day = "1",
doi = "10.1093/annweh/wxw001",
language = "English (US)",
volume = "61",
pages = "67--75",
journal = "Annals of Work Exposures and Health",
issn = "2398-7308",
publisher = "Oxford University Press",
number = "1",

}

TY - JOUR

T1 - A Method for Constructing Informative Priors for Bayesian Modeling of Occupational Hygiene Data

AU - Quick, Harrison

AU - Huynh, Tran

AU - Ramachandran, Gurumurthy

PY - 2017/1/1

Y1 - 2017/1/1

N2 - In many occupational hygiene settings, the demand for more accurate, more precise results is at odds with limited resources. To combat this, practitioners have begun using Bayesian methods to incorporate prior information into their statistical models in order to obtain more refined inference from their data. This is not without risk, however, as incorporating prior information that disagrees with the information contained in data can lead to spurious conclusions, particularly if the prior is too informative. In this article, we propose a method for constructing informative prior distributions for normal and lognormal data that are intuitive to specify and robust to bias. To demonstrate the use of these priors, we walk practitioners through a step-by-step implementation of our priors using an illustrative example. We then conclude with recommendations for general use.

AB - In many occupational hygiene settings, the demand for more accurate, more precise results is at odds with limited resources. To combat this, practitioners have begun using Bayesian methods to incorporate prior information into their statistical models in order to obtain more refined inference from their data. This is not without risk, however, as incorporating prior information that disagrees with the information contained in data can lead to spurious conclusions, particularly if the prior is too informative. In this article, we propose a method for constructing informative prior distributions for normal and lognormal data that are intuitive to specify and robust to bias. To demonstrate the use of these priors, we walk practitioners through a step-by-step implementation of our priors using an illustrative example. We then conclude with recommendations for general use.

KW - decision making

KW - hierarchical modeling

KW - prior sample size

KW - sparse data

KW - truncated priors

UR - http://www.scopus.com/inward/record.url?scp=85037599037&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85037599037&partnerID=8YFLogxK

U2 - 10.1093/annweh/wxw001

DO - 10.1093/annweh/wxw001

M3 - Article

C2 - 28395307

AN - SCOPUS:85037599037

VL - 61

SP - 67

EP - 75

JO - Annals of Work Exposures and Health

JF - Annals of Work Exposures and Health

SN - 2398-7308

IS - 1

ER -