Inference in mixed models in R - beyond the usual asymptotic likelihood ratio test

Ulrich Halekoh; Søren Højsgaard

Abstract

Mixed models in R (www.r-project.org) are currently usually handled with the lme4 package.
Until recently, inference (hypothesis test) in linear mixed models with lme4 was commonly based on the limiting χ² distribution of the likelihood ratio statistic. The pbkrtest package provides two alternatives: 1) A Kenward-Roger approximation for calculating (or estimating) the numerator degrees of freedom for an ”F-like” test statistic. 2) p-values based on simulating the reference
distribution of the likelihood ratio statistic via parametric bootstrap. A recent addition to the package is a Satterthwaite approximation of the degrees of freedom. We will illustrate the package through various examples, and discuss some directions for further developments.

References
[1] Halekoh, U., Højsgaard, S. (2014). A Kenward-Roger Approximation and Parametric Bootstrap
Methods for Tests in Linear Mixed Models The R Package pbkrtest. Journal of Statistical
Software 59, 1–32.
36

Original language	English
Publication date	2018
Number of pages	1
Publication status	Published - 2018
Event	27th Nordic Conference in Mathematical Statistics - Dorpat Convention Centre, Tartu, Estonia Duration: 26 Jun 2018 → 29 Aug 2018 Conference number: 27 http://nordstat2018.ut.ee/

Conference

Conference	27th Nordic Conference in Mathematical Statistics
Number	27
Location	Dorpat Convention Centre
Country	Estonia
City	Tartu
Period	26/06/2018 → 29/08/2018
Internet address	http://nordstat2018.ut.ee/

Keywords

Kenward-Roger approximation
parametric bootstrap
Satterthwaite approximation

Access to Document

http://nordstat2018.ut.ee/sites/default/files/nordstat2018/files/nordstat2018-abstract.pdf

Fingerprint Dive into the research topics of 'Inference in mixed models in R - beyond the usual asymptotic likelihood ratio test'. Together they form a unique fingerprint.

Cite this

@conference{10db0cdb1d8e492fb7d822a160f12964,

title = "Inference in mixed models in R - beyond the usual asymptotic likelihood ratio test",

abstract = "Mixed models in R (www.r-project.org) are currently usually handled with the lme4 package.Until recently, inference (hypothesis test) in linear mixed models with lme4 was commonly based on the limiting χ2 distribution of the likelihood ratio statistic. The pbkrtest package provides two alternatives: 1) A Kenward-Roger approximation for calculating (or estimating) the numerator degrees of freedom for an ”F-like” test statistic. 2) p-values based on simulating the referencedistribution of the likelihood ratio statistic via parametric bootstrap. A recent addition to the package is a Satterthwaite approximation of the degrees of freedom. We will illustrate the package through various examples, and discuss some directions for further developments.References[1] Halekoh, U., H{\o}jsgaard, S. (2014). A Kenward-Roger Approximation and Parametric BootstrapMethods for Tests in Linear Mixed Models The R Package pbkrtest. Journal of StatisticalSoftware 59, 1–32.36",

keywords = "Kenward-Roger approximation, parametric bootstrap, Satterthwaite approximation, Kenward-Roger approximation, parametric bootstrap, Satterthwaite approximation",

author = "Ulrich Halekoh and S{\o}ren H{\o}jsgaard",

year = "2018",

language = "English",

pages = "31",

note = "27th Nordic Conference in Mathematical Statistics , NORDSTAT 2018 ; Conference date: 26-06-2018 Through 29-08-2018",

url = "http://nordstat2018.ut.ee/",

}

TY - ABST

T1 - Inference in mixed models in R - beyond the usual asymptotic likelihood ratio test

AU - Halekoh, Ulrich

AU - Højsgaard, Søren

N1 - Conference code: 27

PY - 2018

Y1 - 2018

N2 - Mixed models in R (www.r-project.org) are currently usually handled with the lme4 package.Until recently, inference (hypothesis test) in linear mixed models with lme4 was commonly based on the limiting χ2 distribution of the likelihood ratio statistic. The pbkrtest package provides two alternatives: 1) A Kenward-Roger approximation for calculating (or estimating) the numerator degrees of freedom for an ”F-like” test statistic. 2) p-values based on simulating the referencedistribution of the likelihood ratio statistic via parametric bootstrap. A recent addition to the package is a Satterthwaite approximation of the degrees of freedom. We will illustrate the package through various examples, and discuss some directions for further developments.References[1] Halekoh, U., Højsgaard, S. (2014). A Kenward-Roger Approximation and Parametric BootstrapMethods for Tests in Linear Mixed Models The R Package pbkrtest. Journal of StatisticalSoftware 59, 1–32.36

AB - Mixed models in R (www.r-project.org) are currently usually handled with the lme4 package.Until recently, inference (hypothesis test) in linear mixed models with lme4 was commonly based on the limiting χ2 distribution of the likelihood ratio statistic. The pbkrtest package provides two alternatives: 1) A Kenward-Roger approximation for calculating (or estimating) the numerator degrees of freedom for an ”F-like” test statistic. 2) p-values based on simulating the referencedistribution of the likelihood ratio statistic via parametric bootstrap. A recent addition to the package is a Satterthwaite approximation of the degrees of freedom. We will illustrate the package through various examples, and discuss some directions for further developments.References[1] Halekoh, U., Højsgaard, S. (2014). A Kenward-Roger Approximation and Parametric BootstrapMethods for Tests in Linear Mixed Models The R Package pbkrtest. Journal of StatisticalSoftware 59, 1–32.36

KW - Kenward-Roger approximation

KW - parametric bootstrap

KW - Satterthwaite approximation

KW - Kenward-Roger approximation

KW - parametric bootstrap

KW - Satterthwaite approximation

M3 - Conference abstract for conference

SP - 31

T2 - 27th Nordic Conference in Mathematical Statistics

Y2 - 26 June 2018 through 29 August 2018

ER -