[R] unordered multinomial logistic regression (or logit model) with repeated measures (I think)

[drlucyasher]

I am attempted to examine the temporal independence of my data set and think
I need an unordered multinomial logistic regression (or logit model) with
repeated measures to do so. The data in question is location of chickens.
Chickens could be in any one of 5 locations when a snapshot sample was
taken. The locations of chickens (bird) in 8 pens (pen) were scored twice a
day (AMPM) for 20 days (date) one a minute for 15 minutes (time). I want to
build a model which specifies that bird is repeated as is time and AMPM and
time is autoregressive. Then I want to use this to see the independence of
location from time to see how much the the data I can legitimally consider
separate independent recordings.

My data looks like this:

Location    Date    Bird    AMPM    Pen     Time
f            15/08/09   1       AM       1        1
p           15/08/09   1       AM       1        2
l           15/08/09   1       AM       1        3
l           15/08/09   1       AM       1        4
l           15/08/09   1       AM       1        5
.
.
.

What I have thought of so far is 

glm(Location~ Pen| Bird*AMPM*Time, family=mulinomial)

I know this isn't correct and also would like to build in that time is
autoregressive and eaxmine this specificy but can't think how to do this.

Maybe I'm totally on the wrong track and I'd be better off looking at the
order of markov chain that best fits that data and take it from there.

Any help or sdvice would be much appreciated.

Lucy
-- 
View this message in context: http://www.nabble.com/unordered-multinomial-logistic-regression-%28or-logit-model%29-with-repeated-measures-%28I-think%29-tp25803223p25803223.html
Sent from the R help mailing list archive at Nabble.com.