GLMM post- hoc comparisons

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

GLMM post- hoc comparisons

Silvina Velez
Hi All,
I have data about seed predation (SP) in fruits of three differents colors (yellow, motted, dark) and in two fruiting seasons (2007, 2008). I performed a GLMM (lmer function, lme4 package) and the outcome showed that the interaction term (color:season) was significant, and some combinations of this interaction have significant Pr(>|z|), but I don't think they are the right significant combinations, because when I look the bwplot, this combinations seems to be very different from the other ones. So, I would like to know if there is any test "a posteriori" to know the p-values ​​for each combination of color:season, and thereby be able to know what conbination/s is/are really significant.

m1=lmer(SP ~ color + season:color +(1|Site:tree), data=datosfl, family="poisson")
AIC   BIC logLik deviance
178.3 196.6 -81.14    162.3
Random effects:
Groups      Name        Variance Std.Dev.
obsBR       (Intercept) 0.064324 0.25362
Site:tree   (Intercept) 0.266490 0.51623
Number of obs: 73, groups: obsBR, 73; Site:tree, 37

                    Estimate Std. Error z value Pr(>|z|)    
(Intercept)            2.5089     0.2750   9.125   <2e-16 ***
colorM                -0.1140     0.3242  -0.352   0.7250    
colorD                -0.6450     0.4178  -1.544   0.1227    
Season2008            -0.7343     0.3104  -2.365   0.0180 *  
colorM:Season2008      0.2505     0.4352   0.576   0.5648    
colorD:Season2008      1.1445     0.5747   1.992   0.0464 *


______________________________________________
[hidden email] 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.
Reply | Threaded
Open this post in threaded view
|

Re: GLMM post- hoc comparisons

Helios de Rosario
>>> El día 08/01/2013 a las 12:40, Silvina Velez
<[hidden email]>
escribió:
> Hi All,
> I have data about seed predation (SP) in fruits of three differents
colors
> (yellow, motted, dark) and in two fruiting seasons (2007, 2008). I
performed
> a GLMM (lmer function, lme4 package) and the outcome showed that the

> interaction term (color:season) was significant, and some
combinations of
> this interaction have significant Pr(>|z|), but I don't think they
are the
> right significant combinations, because when I look the bwplot, this

> combinations seems to be very different from the other ones. So, I
would like
> to know if there is any test "a posteriori" to know the p-values for
each
> combination of color:season, and thereby be able to know what
conbination/s

> is/are really significant.
>
> m1=lmer(SP ~ color + season:color +(1|Site:tree), data=datosfl,
> family="poisson")
> AIC   BIC logLik deviance
> 178.3 196.6 -81.14    162.3
> Random effects:
> Groups      Name        Variance Std.Dev.
> obsBR       (Intercept) 0.064324 0.25362
> Site:tree   (Intercept) 0.266490 0.51623
> Number of obs: 73, groups: obsBR, 73; Site:tree, 37
>
>                     Estimate Std. Error z value Pr(>|z|)    
> (Intercept)            2.5089     0.2750   9.125   <2e-16 ***
> colorM                -0.1140     0.3242  -0.352   0.7250    
> colorD                -0.6450     0.4178  -1.544   0.1227    
> Season2008            -0.7343     0.3104  -2.365   0.0180 *  
> colorM:Season2008      0.2505     0.4352   0.576   0.5648    
> colorD:Season2008      1.1445     0.5747   1.992   0.0464 *

Hi Silvina,

What do you exactly mean with "what combination(s) is/are significant"?
If you mean "what combinations have significantly greater SP than the
baseline combination (yellow:2007)", the table that you have copied may
be what you actually want. If you want to test other contrasts between
color:season combinations, perhaps you can use the function
testInteractions() from package "phia". For instance:

testInteractions(m1)

will give you a test of all the pairwise contrasts between color and
season. You can also test simple main effects, or other specific
contrasts by adding further arguments (see the documentation and the
package vignette). Anyway, the calculation of p-values in mixed models
must always be taken with care.

Helios De Rosario-Martinez
Instituto de Biomecánica de Valencia



INSTITUTO DE BIOMECÁNICA DE VALENCIA
Universidad Politécnica de Valencia • Edificio 9C
Camino de Vera s/n • 46022 VALENCIA (ESPAÑA)
Tel. +34 96 387 91 60 • Fax +34 96 387 91 69
www.ibv.org

  Antes de imprimir este e-mail piense bien si es necesario hacerlo.
En cumplimiento de la Ley Orgánica 15/1999 reguladora de la Protección
de Datos de Carácter Personal, le informamos de que el presente mensaje
contiene información confidencial, siendo para uso exclusivo del
destinatario arriba indicado. En caso de no ser usted el destinatario
del mismo le informamos que su recepción no le autoriza a su divulgación
o reproducción por cualquier medio, debiendo destruirlo de inmediato,
rogándole lo notifique al remitente.

______________________________________________
[hidden email] 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.