[R] Random Effects Model

[Mike Dunbar]

Hi Ryan

Your model code suggests that Type is a variable that you have in your FREE data frame, but you don't mention it, that is probably part of your problem?

The second question is how are your data structured? Sex goes in the fixed part of the model. Swimmer is almost certainly goes in the random component, as you have identified. The question is what to do with Swim. This could either be the key variable of interest or simply a label representing replicates. What is the hypothesis you're trying to test? When swimmers 1..n undertake swim 1 is this all at the same time? If this is the case, do you have reason to believe that there could be some similarlites (not sure what, maybe temperature, wind conditions, fatigue) between swimmers on the same swim? 

Without information on the data structure then its difficult to suggest anything further. As a first guess you could try comparing:

m1 <- lme(Difference~Sex, random=~1|Swimmer, data=FREE)
m2 <- lme(Difference~Sex, random=~1|Swimmer/Swim, data=FREE)

With Swim in the fixed component of the model you're suggesting a linear relationship between Difference and Swim, are you sure this is what you want? If you really think Swim should be in the fixed compoent then you might want to compare:

m3 <- lme(Difference~Sex*Swim, random=~1|Swimmer, data=FREE)
m4 <- lme(Difference~Sex*Swim, random=~Swim|Swimmer, data=FREE)

The latter will fit an overall relationship between Difference and Swim, and then estimate both intercept and slope random effects for Swim.

regards

Mike D


>>> <readams at colby.edu> 19/12/2006 03:18 >>>
Hello,

I am new to R, and I am trying to figure out how to use it for a  
random effects model. I am using version 2.4.0, and I also have the  
book Applied Linear Regression by Sanford Weisberg.

I have four variables: Swimmer, Sex, Swim, and Difference.  Swimmer  
identifies the number assigned to a particular person. Sex is  
male/female. Swim identifies the number swim from 1 to 6. Difference  
is my variable of interest (random).

The book says I should run something like this:

> library(nlme)
> data(FREE) #FREE is my dataset
> xyplot(Difference~Swim|Type, group=Swimmer, data=FREE,
+     panel.groups=function(x,y,...){
+       panel.linejoin(x,y,horizontal=FALSE,...)}
+     )
> m1 <- lme(Difference~Swim+Type, data=FREE, random=~1+Type|Swimmer)
> m2 <- update(m1, random=~1|Swimmer)

I've pretty much decided that xyplot doesn't work. Instead, I think  
plot will work much better.  I keep getting errors about Type. Can  
someone explain this to me? Does the code above look right?

Any kind of help will be appreciated...

Thanks in advance!

Ryan

______________________________________________
R-help at stat.math.ethz.ch 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.


-- 
This message (and any attachments) is for the recipient only...{{dropped}}