LMM without Random effect: I want to run an LMM both with and without the random factor (ID). And then extract the log-lik values from the two models in order to generate a p-value. with random factor as: lmer(y~x+(1|ID),data) Question: can I simply substitute a dummy var (e.g. populated by zeros) for "ID" to run the model without the random factor? when I try this R returns values that seem reasonable, but I want to be sure this is appropriate.>From other inquires in the forum it seems that substituting ("ID-1") or somevariant like that, does not work. Thanks for your thoughts. Baugh -- View this message in context: http://r.789695.n4.nabble.com/lmm-WITHOUT-random-factor-lme4-tp3384054p3384054.html Sent from the R help mailing list archive at Nabble.com.
On Mar 17, 2011; 11:43am Baugh wrote:>> Question: can I simply substitute a dummy var (e.g. populated by zeros) >> for "ID" to run the model >> without the random factor? when I try this R returns values that seem >> reasonable, but I want to be sure >> this is appropriate.If you can fit the model using lme (and it looks like you easily can) then another check would be: ## Compare models with and without random effects fm <- lm(Reaction ~ Days, sleepstudy) fm1 <- lme(Reaction ~ Days, random= ~1|Subject, sleepstudy) anova(fm1, fm) ## lme-fitted model must come first Regards, Mark. -- View this message in context: http://r.789695.n4.nabble.com/lmm-WITHOUT-random-factor-lme4-tp3384054p3384072.html Sent from the R help mailing list archive at Nabble.com.
Dear Mark,
You cannot compare lm() with lme() because the likelihoods are not the same. Use
gls() instead of lm()
library(nlme)
data("sleepstudy", package = "lme4")
fm <- lm(Reaction ~ Days, sleepstudy)
fm0 <- gls(Reaction ~ Days, sleepstudy)
logLik(fm)
logLik(fm0)
fm1 <- lme(Reaction ~ Days, random= ~1|Subject, sleepstudy)
anova(fm0, fm1)
Best regards,
Thierry
----------------------------------------------------------------------------
ir. Thierry Onkelinx
Instituut voor natuur- en bosonderzoek
team Biometrie & Kwaliteitszorg
Gaverstraat 4
9500 Geraardsbergen
Belgium
Research Institute for Nature and Forest
team Biometrics & Quality Assurance
Gaverstraat 4
9500 Geraardsbergen
Belgium
tel. + 32 54/436 185
Thierry.Onkelinx at inbo.be
www.inbo.be
To call in the statistician after the experiment is done may be no more than
asking him to perform a post-mortem examination: he may be able to say what the
experiment died of.
~ Sir Ronald Aylmer Fisher
The plural of anecdote is not data.
~ Roger Brinner
The combination of some data and an aching desire for an answer does not ensure
that a reasonable answer can be extracted from a given body of data.
~ John Tukey
> -----Oorspronkelijk bericht-----
> Van: r-help-bounces at r-project.org
> [mailto:r-help-bounces at r-project.org] Namens Mark Difford
> Verzonden: donderdag 17 maart 2011 10:55
> Aan: r-help at r-project.org
> Onderwerp: Re: [R] lmm WITHOUT random factor (lme4)
>
> On Mar 17, 2011; 11:43am Baugh wrote:
>
> >> Question: can I simply substitute a dummy var (e.g. populated by
> >> zeros) for "ID" to run the model without the random
factor? when I
> >> try this R returns values that seem reasonable, but I want
> to be sure
> >> this is appropriate.
>
> If you can fit the model using lme (and it looks like you
> easily can) then another check would be:
>
> ## Compare models with and without random effects fm <-
> lm(Reaction ~ Days, sleepstudy)
> fm1 <- lme(Reaction ~ Days, random= ~1|Subject, sleepstudy)
> anova(fm1, fm) ## lme-fitted model must come first
>
> Regards, Mark.
>
> --
> View this message in context:
> http://r.789695.n4.nabble.com/lmm-WITHOUT-random-factor-lme4-t
> p3384054p3384072.html
> Sent from the R help mailing list archive at Nabble.com.
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
On Mar 17, 2011; 04:29pm Thierry Onkelinx wrote:>> You cannot compare lm() with lme() because the likelihoods are not the >> same. Use gls() instead of lm()Hi Thierry, Of course, I stand subject to correction, but unless something dramatic has changed, you can. gls() can be used if you need to accommodate a correlation structure. The method I have outlined, i.e. anova(lme$obj, lm$obj), is detailed in Pinheiro & Bates (2000) beginning page 154. Please refer to this if you doubt me. Regards, Mark. -- View this message in context: http://r.789695.n4.nabble.com/lmm-WITHOUT-random-factor-lme4-tp3384054p3384802.html Sent from the R help mailing list archive at Nabble.com.
On Mar 17, 2011; 04:29pm Thierry Onkelinx wrote:>> You cannot compare lm() with lme() because the likelihoods are not the >> same. Use gls() instead of lm()And perhaps I should have added the following: First para on page 155 of Pinheiro & Bates (2000) states, "The anova method can be used to compare lme and lm objects." Regards, Mark. -- View this message in context: http://r.789695.n4.nabble.com/lmm-WITHOUT-random-factor-lme4-tp3384054p3384823.html Sent from the R help mailing list archive at Nabble.com.