Dear list, In lmer, no significances for the t-values of fixed effects are given, while in a logistic model with glmer, significances for z-values are shown. Where does this difference come from? Does it mean, that the problem of computing the "right" nr. of degrees of freedom for the t-values of a linear model does not exits (or is less of a problem) for non-linear models? Are the significances that glmer produces (at least in case of logistic regression) based on the Wald statistic or may be on an LRT test? Thanks for any explanation! Ben.
(no) significance value (gl)mer
6 messages · Mollie Brooks, Marko Bachl, Ben Bolker +1 more
Hi Ben,
The latest documentation on that topic should be available with the code
help("pvalues",package="lme4")
cheers,
Mollie
------------------------
Mollie Brooks, PhD
Postdoctoral Researcher, Population Ecology Research Group
Department of Evolutionary Biology & Environmental Studies, University of Z?rich
http://www.popecol.org/team/mollie-brooks/
On 25Apr 2016, at 15:52, Ben Pelzer <b.pelzer at maw.ru.nl> wrote: Dear list, In lmer, no significances for the t-values of fixed effects are given, while in a logistic model with glmer, significances for z-values are shown. Where does this difference come from? Does it mean, that the problem of computing the "right" nr. of degrees of freedom for the t-values of a linear model does not exits (or is less of a problem) for non-linear models? Are the significances that glmer produces (at least in case of logistic regression) based on the Wald statistic or may be on an LRT test? Thanks for any explanation! Ben.
_______________________________________________ R-sig-mixed-models at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
Dear Ben, Just a short info, others may have something more substantial to add. You will find plenty of information on significance tests and inference in lme4 in the top sections of this website: http://glmm.wikidot.com/faq. A short info inside the lme4 package is given by require(lme4) ?pvalues Best, Marko 2016-04-25 15:52 GMT+02:00 Ben Pelzer <b.pelzer at maw.ru.nl>:
Dear list, In lmer, no significances for the t-values of fixed effects are given, while in a logistic model with glmer, significances for z-values are shown. Where does this difference come from? Does it mean, that the problem of computing the "right" nr. of degrees of freedom for the t-values of a linear model does not exits (or is less of a problem) for non-linear models? Are the significances that glmer produces (at least in case of logistic regression) based on the Wald statistic or may be on an LRT test? Thanks for any explanation! Ben.
_______________________________________________ R-sig-mixed-models at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
Dr. Marko Bachl Universit?t Hohenheim Institut f?r Kommunikationswissenschaft (540C) T 0711 459 228 66 M marko.bachl at uni-hohenheim.de W www.komm.uni-hohenheim.de/bachl
For what it's worth, I've lately been playing with the FAQ; a (*very* slightly) updated version is at https://rawgit.com/bbolker/mixedmodels-misc/master/glmmFAQ.html (source at https://github.com/bbolker/mixedmodels-misc/blob/master/glmmFAQ.rmd ). At some point when I think the new version is sufficiently improved over the old one to warrant it, I'll replace the old (wikidot.com) version with a pointer to the Github repo. Comments (and pull requests) welcome!
On 16-04-25 10:27 AM, Marko Bachl wrote:
Dear Ben, Just a short info, others may have something more substantial to add. You will find plenty of information on significance tests and inference in lme4 in the top sections of this website: http://glmm.wikidot.com/faq. A short info inside the lme4 package is given by require(lme4) ?pvalues Best, Marko 2016-04-25 15:52 GMT+02:00 Ben Pelzer <b.pelzer at maw.ru.nl>:
Dear list, In lmer, no significances for the t-values of fixed effects are given, while in a logistic model with glmer, significances for z-values are shown. Where does this difference come from? Does it mean, that the problem of computing the "right" nr. of degrees of freedom for the t-values of a linear model does not exits (or is less of a problem) for non-linear models? Are the significances that glmer produces (at least in case of logistic regression) based on the Wald statistic or may be on an LRT test? Thanks for any explanation! Ben.
_______________________________________________ R-sig-mixed-models at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
For what it's worth, I've lately been playing with the FAQ; a (*very* slightly) updated version is at https://rawgit.com/bbolker/mixedmodels-misc/master/glmmFAQ.html (source at https://github.com/bbolker/mixedmodels-misc/blob/master/glmmFAQ.rmd ). At some point when I think the new version is sufficiently improved over the old one to warrant it, I'll replace the old (wikidot.com) version with a pointer to the Github repo. Comments (and pull requests) welcome!
On 16-04-25 10:27 AM, Marko Bachl wrote:
Dear Ben, Just a short info, others may have something more substantial to add. You will find plenty of information on significance tests and inference in lme4 in the top sections of this website: http://glmm.wikidot.com/faq. A short info inside the lme4 package is given by require(lme4) ?pvalues Best, Marko 2016-04-25 15:52 GMT+02:00 Ben Pelzer <b.pelzer at maw.ru.nl>:
Dear list, In lmer, no significances for the t-values of fixed effects are given, while in a logistic model with glmer, significances for z-values are shown. Where does this difference come from? Does it mean, that the problem of computing the "right" nr. of degrees of freedom for the t-values of a linear model does not exits (or is less of a problem) for non-linear models? Are the significances that glmer produces (at least in case of logistic regression) based on the Wald statistic or may be on an LRT test? Thanks for any explanation! Ben.
_______________________________________________ R-sig-mixed-models at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
Dear Milly, Marko and Ben, Thanks a lot for your replies and pointing me to these packages, great! Kind regards, Ben.
On 25-4-2016 19:43, Ben Bolker wrote:
For what it's worth, I've lately been playing with the FAQ; a (*very* slightly) updated version is at https://rawgit.com/bbolker/mixedmodels-misc/master/glmmFAQ.html (source at https://github.com/bbolker/mixedmodels-misc/blob/master/glmmFAQ.rmd ). At some point when I think the new version is sufficiently improved over the old one to warrant it, I'll replace the old (wikidot.com) version with a pointer to the Github repo. Comments (and pull requests) welcome! On 16-04-25 10:27 AM, Marko Bachl wrote:
Dear Ben, Just a short info, others may have something more substantial to add. You will find plenty of information on significance tests and inference in lme4 in the top sections of this website: http://glmm.wikidot.com/faq. A short info inside the lme4 package is given by require(lme4) ?pvalues Best, Marko 2016-04-25 15:52 GMT+02:00 Ben Pelzer <b.pelzer at maw.ru.nl>:
Dear list, In lmer, no significances for the t-values of fixed effects are given, while in a logistic model with glmer, significances for z-values are shown. Where does this difference come from? Does it mean, that the problem of computing the "right" nr. of degrees of freedom for the t-values of a linear model does not exits (or is less of a problem) for non-linear models? Are the significances that glmer produces (at least in case of logistic regression) based on the Wald statistic or may be on an LRT test? Thanks for any explanation! Ben.
_______________________________________________ R-sig-mixed-models at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
_______________________________________________ R-sig-mixed-models at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models