R help - Apr 2014 - Mixed Effects MANOVA

Greetings,

I have a question regarding data analysis of habitat use of animals.  
These animals were radio collared and tracked periodically throughout 
the year.  When they were sighted/detected, the habitat type was 
marked.  Our dataset recorded the sex of the animal, and we know the 
data when the surveys were performed.  The goal was to address the 
questions: does habitat use differ between the sexes, and does habitat 
use vary between seasons?  Below is a summary table, ignoring seasons.

dattab <- matrix(c(190,87,206,170,103,23,66,72,53,22),nrow=5,byrow=T)
rownames(dattab) <-
c("Rock","Burrow","Cactus","Brushpile","Other")
colnames(dattab) <- c("Female","Male")
dattab
           Female Male
Rock         190   87
Burrow       206  170
Cactus       103   23
Brushpile     66   72
Other         53   22

We could perform a test of independence, but the problem lies with our 
assumptions.  Because individual animals were tracked through time, each 
animal give a different number of datapoints (min=1, max=126), which 
violates our assumption of independence.  Thus, our sampling unit should 
be at the level of the skunk and analysis should proceed from there.  
I'm familiar (theory and practice) with linear mixed effect models, but 
I believe that these data call for a mixed effects MANOVA.  Is there 
such a test in R?  Or, would it be better to analyze the data using a 
standard MANOVA where our y1, y2, ... are the percentage of data points 
within that various habitats? My problem with this last analysis is that 
each skunk will carry the same weight even though both could have a 
large difference in the number of data points.  Thanks...

This is R-help, not a list that provides statistical help (primarily;
they do intersect at times). Post to the r-sig-mixed-models list
instead. You're likely to do better there anyway for this sort of
thing.

Cheers,
Bert

Bert Gunter
Genentech Nonclinical Biostatistics
(650) 467-7374

"Data is not information. Information is not knowledge. And knowledge
is certainly not wisdom."
H. Gilbert Welch




On Thu, Apr 3, 2014 at 6:29 AM, Nick Negovetich <nj.negovetich at
gmail.com> wrote:
> Greetings,
>
> I have a question regarding data analysis of habitat use of animals.  These
> animals were radio collared and tracked periodically throughout the year.
> When they were sighted/detected, the habitat type was marked.  Our dataset
> recorded the sex of the animal, and we know the data when the surveys were
> performed.  The goal was to address the questions: does habitat use differ
> between the sexes, and does habitat use vary between seasons?  Below is a
> summary table, ignoring seasons.
>
> dattab <- matrix(c(190,87,206,170,103,23,66,72,53,22),nrow=5,byrow=T)
> rownames(dattab) <-
c("Rock","Burrow","Cactus","Brushpile","Other")
> colnames(dattab) <- c("Female","Male")
> dattab
>           Female Male
> Rock         190   87
> Burrow       206  170
> Cactus       103   23
> Brushpile     66   72
> Other         53   22
>
> We could perform a test of independence, but the problem lies with our
> assumptions.  Because individual animals were tracked through time, each
> animal give a different number of datapoints (min=1, max=126), which
> violates our assumption of independence.  Thus, our sampling unit should be
> at the level of the skunk and analysis should proceed from there.  I'm
> familiar (theory and practice) with linear mixed effect models, but I
> believe that these data call for a mixed effects MANOVA.  Is there such a
> test in R?  Or, would it be better to analyze the data using a standard
> MANOVA where our y1, y2, ... are the percentage of data points within that
> various habitats? My problem with this last analysis is that each skunk
will
> carry the same weight even though both could have a large difference in the
> number of data points.  Thanks...
>
> ______________________________________________
> 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.