Questions about mix models

Alena,

1) Can you join a summary of your data. Is it a balanced design?

2) Not sure to understand how your model assigns the residual error... Have you tried to exclude variable 'top' from the random component: only (1|mead/trans) ? does it improve convergence? and do you get the appropriate number of degree of freedom for your fixed effects (based on your experimental design)? 

Julien Beguin
________________________________________
De : r-sig-mixed-models-bounces at r-project.org [r-sig-mixed-models-bounces at r-project.org] de la part de Alena Dra?narov? [drasnarova.alena at gmail.com]
Date d'envoi : 16 ao?t 2010 05:58
? : r-sig-mixed-models at r-project.org
Objet : [R-sig-ME] Questions about mix models

Dear all,
 I have so complicated data and I am trying to gain correct results from them.

I am interested in factors influencing density and diversity of the
soil seed bank on alluvial meadows. I have nested design of my
experiment: 35 meadows (mead=M1-M35), three transects on each meadow
(trans=T1-T3) and  2 plots on each transect (top=A,B).
I found out  a lot of information (about soil properties, moisure,
litter, biomass, vegetation diversity and management).
At first, I tried to use glmer, but sometimes there was error message:

a2<-glmer(number~top+depth+HPV+K+VVS+(1|mead/trans/top),data=dat,family=poisson)

Warning messages:
1: In mer_finalize(ans) :
 Cholmod warning 'not positive definite' at
file:../Cholesky/t_cholmod_rowfac.c, line 432
2: In mer_finalize(ans) :
 Cholmod warning 'not positive definite' at
file:../Cholesky/t_cholmod_rowfac.c, line 432
3: In mer_finalize(ans) : false convergence (8)

So, I decided to use MCMCglmm, but I am not sure with fitting the
model. I tried to fitt it by this way (example below is for one
factor):

prior=list(R=list(V=1, n=0, fix=1), G=list(G1 = list(V =1,n=1),G2=list(V=1,n=1),G3=list(V=1,n=1)))
m1 <- MCMCglmm(number ~ as.factor(top), random=~mead+mead:trans+mead:trans:top, family = "poisson", data=dat,prior=prior)

I am not sure with define prior and random effect.

I will be very happy, if anybody write me own experiences with these
models and similar data and help me which model is the best to use.

With kind regards
Alena Dra?narov?

_______________________________________________
R-sig-mixed-models at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models

An embedded and charset-unspecified text was scrubbed...
Name: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-mixed-models/attachments/20100816/175f3b59/attachment.pl>

Hello,

given that you are interested in investigating the effects of a series of 
predictors (e.g. moisture) on the number of seeds, whilst using random 
effects to account for your sampling design, I would actually suggest to fit 
your model without "top" also as fixed effect. Something like:

glmer(number ~ depth + HPV + K + VVS + 
(1|mead/trans/top),data=dat,family=poisson)

HTH, just my 2 cents.


Cheers,

Luca


----- Original Message ----- 
From: "Alena Drasnarov?" <drasnarova.alena at gmail.com>
To: "Julien Beguin" <julien.beguin.1 at ulaval.ca>
Cc: <r-sig-mixed-models at r-project.org>
Sent: Monday, August 16, 2010 1:38 PM
Subject: Re: [R-sig-ME] RE : Questions about mix models


Julien, thank you for your reaction.
1) Below you can see structura of my data (for 1 meadow)

        mead trans top depth number man litt water pH Ca K Mg P N C  VVS  1
1 A S 605 L 8.6 0 5.28 40.667 8.000 14.292 1.903 0.165 14.068 0.199  1 1 A V
582 L 8.6 0 5.28 40.667 8.000 14.292 1.903 0.165 14.068 0.199  1 1 B S 135 L
10.5 208 4.49 3.629 4.484 2.387 1.889 0.185 10.173 0.096  1 1 B V 153 L 10.5
208 4.49 3.629 4.484 2.387 1.889 0.185 10.173 0.096  1 2 A S 3 L 2.6 182
5.90 114.113 33.967 27.520 1.848 0.167 8.782 0.457  1 2 A V 2 L 2.6 182 5.90
114.113 33.967 27.520 1.848 0.167 8.782 0.457  1 2 B S 18 L 7.7 332 5.48
133.495 9.194 41.580 1.769 0.252 11.612 0.252  1 2 B V 57 L 7.7 332 5.48
133.495 9.194 41.580 1.769 0.252 11.612 0.252  1 3 A S 387 L 5.4 0 5.84
266.500 8.588 51.103 1.777 0.211 18.139 0.232  1 3 A V 462 L 5.4 0 5.84
266.500 8.588 51.103 1.777 0.211 18.139 0.232  1 3 B S 62 L 4.5 5 5.32
227.184 15.444 47.302 1.895 0.337 14.172 0.313  1 3 B V 22 L 4.5 5 5.32
227.184 15.444 47.302 1.895 0.337 14.172 0.313
Only on 2 meadows there are some missing data. But I prefer to use these
plots too.

2)
I did not try my model without top in random part. I can try it, but I think
that the model will lost important information about my design. About deegre
of freedom, I am not sure how to calculate them.

Alena
























































































































































































































































































































































































































Dne 16. srpna 2010 15:26 Julien Beguin <julien.beguin.1 at ulaval.ca>
napsal(a):

Alena,

1) Can you join a summary of your data. Is it a balanced design?

2) Not sure to understand how your model assigns the residual error...

Have you tried to exclude variable 'top' from the random component: only
(1|mead/trans) ? does it improve convergence? and do you get the appropriate
number of degree of freedom for your fixed effects (based on your
experimental design)?

Julien Beguin

________________________________________
De : r-sig-mixed-models-bounces at r-project.org [

r-sig-mixed-models-bounces at r-project.org] de la part de Alena Dra??narov?? [
drasnarova.alena at gmail.com]

Date d'envoi : 16 ao??t 2010 05:58
?? : r-sig-mixed-models at r-project.org
Objet : [R-sig-ME] Questions about mix models

Dear all,
 I have so complicated data and I am trying to gain correct results from

them.

I am interested in factors influencing density and diversity of the
soil seed bank on alluvial meadows. I have nested design of my
experiment: 35 meadows (mead=M1-M35), three transects on each meadow
(trans=T1-T3) and  2 plots on each transect (top=A,B).
I found out  a lot of information (about soil properties, moisure,
litter, biomass, vegetation diversity and management).
At first, I tried to use glmer, but sometimes there was error message:

a2<-glmer(number~top+depth+HPV+K+VVS+(1|mead/trans/top),data=dat,family=poisson)

Warning messages:
1: In mer_finalize(ans) :
 Cholmod warning 'not positive definite' at
file:../Cholesky/t_cholmod_rowfac.c, line 432
2: In mer_finalize(ans) :
 Cholmod warning 'not positive definite' at
file:../Cholesky/t_cholmod_rowfac.c, line 432
3: In mer_finalize(ans) : false convergence (8)

So, I decided to use MCMCglmm, but I am not sure with fitting the
model. I tried to fitt it by this way (example below is for one
factor):

prior=list(R=list(V=1, n=0, fix=1), G=list(G1 = list(V

=1,n=1),G2=list(V=1,n=1),G3=list(V=1,n=1)))

m1 <- MCMCglmm(number ~ as.factor(top),

random=~mead+mead:trans+mead:trans:top, family = "poisson",
data=dat,prior=prior)

I am not sure with define prior and random effect.

I will be very happy, if anybody write me own experiences with these
models and similar data and help me which model is the best to use.

With kind regards
Alena Dra??narov??

_______________________________________________
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

Dear Alena,

As other's have said, its hard to assess the problem without more  
information. However, with regards to MCMCglmm, the model you  
specified is equivalent to the model you tried to fit in glmer (in  
terms of random effects) but models over-dispersion, which is  
important. However, you have fixed the residual variance to one which  
you should not do - this is only for categorical and ordinal responses.

If glmer is issuing warnings of this sort it often suggests there may  
be some problems with the model and so I would be very careful. If  
there is little replication for the random effects the priors you use  
will be quite informative. There is no perfect prior but I often find  
parameter expanded priors to work well:

prior=list(R=list(V=1, n=0), G=list(G1 = list(V =1,n=1, alpha.mu=0,  
alpha.V=1000),G2=list(V=1,n=1, alpha.mu=0,  
alpha.V=1000),G3=list(V=1,n=1, alpha.mu=0, alpha.V=1000)))

Cheers,

Jarrod

Dear all,
I have so complicated data and I am trying to gain correct results  
from them.

I am interested in factors influencing density and diversity of the
soil seed bank on alluvial meadows. I have nested design of my
experiment: 35 meadows (mead=M1-M35), three transects on each meadow
(trans=T1-T3) and  2 plots on each transect (top=A,B).
I found out  a lot of information (about soil properties, moisure,
litter, biomass, vegetation diversity and management).
At first, I tried to use glmer, but sometimes there was error message:

a2<-glmer(number~top+depth+HPV+K+VVS+(1|mead/trans/ 
top),data=dat,family=poisson)

Warning messages:
1: In mer_finalize(ans) :
Cholmod warning 'not positive definite' at
file:../Cholesky/t_cholmod_rowfac.c, line 432
2: In mer_finalize(ans) :
Cholmod warning 'not positive definite' at
file:../Cholesky/t_cholmod_rowfac.c, line 432
3: In mer_finalize(ans) : false convergence (8)

So, I decided to use MCMCglmm, but I am not sure with fitting the
model. I tried to fitt it by this way (example below is for one
factor):

prior=list(R=list(V=1, n=0, fix=1), G=list(G1 = list(V  
=1,n=1),G2=list(V=1,n=1),G3=list(V=1,n=1)))
m1 <- MCMCglmm(number ~ as.factor(top), random=~mead+mead:trans 
+mead:trans:top, family = "poisson", data=dat,prior=prior)

I am not sure with define prior and random effect.

I will be very happy, if anybody write me own experiences with these
models and similar data and help me which model is the best to use.

With kind regards
Alena Dra?narov?

_______________________________________________
R-sig-mixed-models at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models

The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.