R help - Jun 2004 - Repeated measures

Dear R-gurus,

I am pretty much new on R.
I am trying to to do a repeated analysis of a linear mixed model with 
R, and I consistently fail...

The problem is: Cow is the random factor, treatment is the fixed 
factor. The dependent variable is milk yield, which is measured several 
times (repeatedly over time), thus there is another variable which is 
time (i.e. week).

The model would be something like this: milk yield = cow + treatment + 
time + treatment*time

With time as a repeated measure.

Would some one be kind enough to guide on how could I set up the 
statement in R?
I imagine I have to use LME but I have not been smart enough to figure 
out how to do it by just reading its manual. Also, with SAS there is a 
nice contrast, called SLICE that allows you to test when in time there 
is difference between treatments. I do not know if there is something 
like this is R.

Thank you very much!

PD. For SAS users, what I am using in SAS to perform this analysis 
(with an Autoregressive covariance structure) the program would read 
like this:

proc MIXED covtest;
CLASS cow treat time;
MODEL yield=  treat time treat*time;
REPEATED time/SUB=cow(treat) TYPE=ARH(1) R RCORR;
RANDOM cow;
LSMEANS treat time treat*time/SLICE = time;
RUN;


Thank you very much,

Sincerely,

Alex

Dear R-gurus,

I am pretty much new on R.
I am trying to to do a repeated analysis of a linear mixed model with 
R, and I consistently fail...

The problem is: Cow is the random factor, treatment is the fixed 
factor. The dependent variable is milk yield, which is measured several 
times (repeatedly over time), thus there is another variable which is 
time (i.e. week).

The model would be something like this: milk yield = cow + treatment + 
time + treatment*time

With time as a repeated measure.

Would some one be kind enough to guide on how could I set up the 
statement in R?
I imagine I have to use LME but I have not been smart enough to figure 
out how to do it by just reading its manual. Also, with SAS there is a 
nice contrast, called SLICE that allows you to test when in time there 
is difference between treatments. I do not know if there is something 
like this is R.

Thank you very much!

PD. For SAS users, what I am using in SAS to perform this analysis 
(with an Autoregressive covariance structure) the program would read 
like this:

proc MIXED covtest;
CLASS cow treat time;
MODEL yield=  treat time treat*time;
REPEATED time/SUB=cow(treat) TYPE=ARH(1) R RCORR;
RANDOM cow;
LSMEANS treat time treat*time/SLICE = time;
RUN;


Thank you very much,

Sincerely,

Alex

If you fit the model as suggested below and store it in an object, say
"model.lme", then you can update the model as follows to include the
continuous AR1 structure
 
model2.lme<-update(model.lme, correlation=corCAR1(form=~time|cow) )
 
To compare the two models, you might use the LRT as:
 
anova(model.lme,model2.lme)
 
Harold

	-----Original Message----- 
	From: r-help-bounces@stat.math.ethz.ch on behalf of Spencer Graves 
	Sent: Sun 6/6/2004 12:21 PM 
	To: Alex Bach 
	Cc: r-help@stat.math.ethz.ch 
	Subject: Re: [R] Repeated measures
	
	

	      1.  You didn't say which "manual" you were reading on
"lme", if
	you have not consulted Pinheiro and Bates (2000) Mixed-Effects Models in
	S and S-Plus (Springer), I suggest you do so.  The issues are discussed
	in greater depth with many examples.  I found this book well worth the
	time and money I invested in it.
	
	      2.  Have you considered the following:
	
	      lme(milk yield = cow + treatment + time + treatment*time, random 
~time|cow ... )
	
	      This will NOT have the ARCH(1) R RCORR error structure;  "lme"
can
	handle certain types of autocorrelated error structures, but I don't
	remember the details at the moment.  Pinheiro and Bates discuss some
	capabilities of this nature.
	
	      hope this helps.
	      spencer graves
	
	Alex Bach wrote:
	
	> Dear R-gurus,
	>
	> I am pretty much new on R.
	> I am trying to to do a repeated analysis of a linear mixed model with
	> R, and I consistently fail...
	>
	> The problem is: Cow is the random factor, treatment is the fixed
	> factor. The dependent variable is milk yield, which is measured
	> several times (repeatedly over time), thus there is another variable
	> which is time (i.e. week).
	>
	> The model would be something like this: milk yield = cow + treatment +
	> time + treatment*time
	>
	> With time as a repeated measure.
	>
	> Would some one be kind enough to guide on how could I set up the
	> statement in R?
	> I imagine I have to use LME but I have not been smart enough to figure
	> out how to do it by just reading its manual. Also, with SAS there is a
	> nice contrast, called SLICE that allows you to test when in time there
	> is difference between treatments. I do not know if there is something
	> like this is R.
	>
	> Thank you very much!
	>
	> PD. For SAS users, what I am using in SAS to perform this analysis
	> (with an Autoregressive covariance structure) the program would read
	> like this:
	>
	> proc MIXED covtest;
	> CLASS cow treat time;
	> MODEL yield=  treat time treat*time;
	> REPEATED time/SUB=cow(treat) TYPE=ARH(1) R RCORR;
	> RANDOM cow;
	> LSMEANS treat time treat*time/SLICE = time;
	> RUN;
	>
	>
	> Thank you very much,
	>
	> Sincerely,
	>
	> Alex
	>
	> ______________________________________________
	> R-help@stat.math.ethz.ch mailing list
	> https://www.stat.math.ethz.ch/mailman/listinfo/r-help
	> PLEASE do read the posting guide!
	> http://www.R-project.org/posting-guide.html
	
	______________________________________________
	R-help@stat.math.ethz.ch mailing list
	https://www.stat.math.ethz.ch/mailman/listinfo/r-help
	PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
	


	[[alternative HTML version deleted]]