[R-sig-ME] likelihood ratio tests vs. bootstrapping inquasi-poisson models

[ONKELINX, Thierry]

Dear marty,

Why don't you derive p-values from your bootstrap data? Just find the
largest confidence level at which zero is still excluded from the
confidence interval.

Furthermore I'm interested in how you simulate data from a quasipoisson
distribution. Can you share your code?

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-sig-mixed-models-bounces at r-project.org 
> [mailto:r-sig-mixed-models-bounces at r-project.org] Namens Marty Kardos
> Verzonden: dinsdag 15 juni 2010 22:28
> Aan: r-sig-mixed-models at r-project.org
> Onderwerp: [R-sig-ME] likelihood ratio tests vs. 
> bootstrapping inquasi-poisson models
> 
> Hello mixed-model folks;
> 
> I am assessing the significance of regression coefficients in 
> quasi-poisson mixed models using lmer.  Based on previous 
> discussion in the mixed models archives, it appears that 
> p-values from likelihood ratio tests on this kind of model 
> may often be unreliable.
> 
> I have been debating with myself over the merits of two 
> alternative approaches to assess the significance of coefficients...
> 
> I have done likelihood ratio tests (comparing nested models), 
> and I have written a program in R to build 'bias-corrected 
> and accelerated' bootstrap confidence intervals for 
> coefficient estimates.  In most cases the inferences based on 
> the bootstrap intervals and likelihood ratio tests 
> qualitatively agree (i.e. when the bootstrap confidence
> interval does not contain zero the likelihood ratio test is 
> significant for the particular coefficient estimate). However 
> in a few important cases, the inferences based on bootstrap 
> intervals and likelihood ratio tests are different (i.e.
> a few coefficients would be determined to be siginficant 
> based on the bootstrap intervals and either marginally
> significant or insignificant based on the likelihood ratio 
> test).  I am attracted to the bootstrapping confidence 
> intervals approach because it appears that it is likely to be 
> more reliable.  I am attracted to the likelihood ratio tests 
> because they provide p-values (which editors and readers of 
> scientific journals seem to enjoy having).
> 
> Here is an example of the models I am working with:
> nongene<-
> lmer(data$offspring~data$age+data$weight+data$agesq+data$Het+(
> 1|data$ID)+(1|factor(data$year)),family=quasipoisson,REML=FALS
> E,na.action=na.omit)
> 
> Do any of you have an opinion on which of these approaches 
> may be best, or ideas for better a better approach?
> 
> Thanks for any thoughts you might have!
> --
> Marty Kardos
> Ph.D. Candidate
> Montana Conservation Genetics Lab
> Organismal Biology & Ecology
> MEID IGERT Program
> University of Montana
> 
> --
> Marty Kardos
> Ph.D. Candidate
> Montana Conservation Genetics Lab
> Organismal Biology & Ecology
> MEID IGERT Program
> University of Montana
> 406-599-1358
> http://dbs.umt.edu/people/gradStudents.php
> 
> 	[[alternative HTML version deleted]]
> 
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list 
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
> 

Druk dit bericht a.u.b. niet onnodig af.
Please do not print this message unnecessarily.

Dit bericht en eventuele bijlagen geven enkel de visie van de schrijver weer 
en binden het INBO onder geen enkel beding, zolang dit bericht niet bevestigd is
door een geldig ondertekend document. The views expressed in  this message 
and any annex are purely those of the writer and may not be regarded as stating 
an official position of INBO, as long as the message is not confirmed by a duly 
signed document.