Model selection

An integral part of inference

S. T. Buckland, K. P. Burnham, N. H. Augustin

Research output: Contribution to journalArticle

906 Citations (Scopus)

Abstract

We argue that model selection uncertainty should be fully incorporated into statistical inference whenever estimation is sensitive to model choice and that choice is made with reference to the data. We consider different philosophies for achieving this goal and suggest strategies for data analysis. We illustrate our methods through three examples. The first is a Poisson regression of bird counts in which a choice is to be made between inclusion of one or both of two covariates. The second is a line transect data set for which different models yield substantially different estimates of abundance. The third is a simulated example in which truth is known.

Original languageEnglish
Pages (from-to)603-618
Number of pages16
JournalBiometrics
Volume53
Issue number2
DOIs
Publication statusPublished - 1 Jun 1997

Keywords

  • AIC
  • BIC
  • Information criteria
  • Model selection uncertainty
  • Simulated inference

ASJC Scopus subject areas

  • Statistics and Probability
  • Biochemistry, Genetics and Molecular Biology(all)
  • Immunology and Microbiology(all)
  • Agricultural and Biological Sciences(all)
  • Applied Mathematics

Cite this

Model selection : An integral part of inference. / Buckland, S. T.; Burnham, K. P.; Augustin, N. H.

In: Biometrics, Vol. 53, No. 2, 01.06.1997, p. 603-618.

Research output: Contribution to journalArticle

Buckland, S. T. ; Burnham, K. P. ; Augustin, N. H. / Model selection : An integral part of inference. In: Biometrics. 1997 ; Vol. 53, No. 2. pp. 603-618.
@article{be5ebd71f5b04834aba93b043ce02975,
title = "Model selection: An integral part of inference",
abstract = "We argue that model selection uncertainty should be fully incorporated into statistical inference whenever estimation is sensitive to model choice and that choice is made with reference to the data. We consider different philosophies for achieving this goal and suggest strategies for data analysis. We illustrate our methods through three examples. The first is a Poisson regression of bird counts in which a choice is to be made between inclusion of one or both of two covariates. The second is a line transect data set for which different models yield substantially different estimates of abundance. The third is a simulated example in which truth is known.",
keywords = "AIC, BIC, Information criteria, Model selection uncertainty, Simulated inference",
author = "Buckland, {S. T.} and Burnham, {K. P.} and Augustin, {N. H.}",
year = "1997",
month = "6",
day = "1",
doi = "10.2307/2533961",
language = "English",
volume = "53",
pages = "603--618",
journal = "Biometrics",
issn = "0006-341X",
publisher = "Wiley-Blackwell",
number = "2",

}

TY - JOUR

T1 - Model selection

T2 - An integral part of inference

AU - Buckland, S. T.

AU - Burnham, K. P.

AU - Augustin, N. H.

PY - 1997/6/1

Y1 - 1997/6/1

N2 - We argue that model selection uncertainty should be fully incorporated into statistical inference whenever estimation is sensitive to model choice and that choice is made with reference to the data. We consider different philosophies for achieving this goal and suggest strategies for data analysis. We illustrate our methods through three examples. The first is a Poisson regression of bird counts in which a choice is to be made between inclusion of one or both of two covariates. The second is a line transect data set for which different models yield substantially different estimates of abundance. The third is a simulated example in which truth is known.

AB - We argue that model selection uncertainty should be fully incorporated into statistical inference whenever estimation is sensitive to model choice and that choice is made with reference to the data. We consider different philosophies for achieving this goal and suggest strategies for data analysis. We illustrate our methods through three examples. The first is a Poisson regression of bird counts in which a choice is to be made between inclusion of one or both of two covariates. The second is a line transect data set for which different models yield substantially different estimates of abundance. The third is a simulated example in which truth is known.

KW - AIC

KW - BIC

KW - Information criteria

KW - Model selection uncertainty

KW - Simulated inference

UR - http://www.scopus.com/inward/record.url?scp=0030613470&partnerID=8YFLogxK

U2 - 10.2307/2533961

DO - 10.2307/2533961

M3 - Article

VL - 53

SP - 603

EP - 618

JO - Biometrics

JF - Biometrics

SN - 0006-341X

IS - 2

ER -