[R] lme and aov

[Gang Chen]

Thanks for the response!

It is indeed a balanced design. The results are different in the  
sense all the F tests for main effects are not the same. Do you mean  
that a random interaction is modeled in the aov command? If so, what  
would be an equivalent command of aov to the one with lme?

Thanks,
Gang

On Aug 3, 2007, at 3:52 PM, Peter Dalgaard wrote:

> Gang Chen wrote:
>> I have a mixed balanced ANOVA design with a between-subject  
>> factor  (Grp) and a within-subject factor (Rsp). When I tried the  
>> following  two commands which I thought are equivalent,
>>
>>  > fit.lme <- lme(Beta ~ Grp*Rsp, random = ~1|Subj, Model);
>>  > fit.aov <- aov(Beta ~ Rsp*Grp+Error(Subj/Rsp)+Grp, Model);
>> 	
>> I got totally different results. What did I do wrong?
>>
>>
> Except for not telling us what your data are and what you mean by  
> "totally different"?
>
> One model has a random interaction between Subj and Rsp, the other  
> does not. This may make a difference, unless the interaction term  
> is aliased with the residual error.
>
> If your data are unbalanced, aov is not guaranteed to give  
> meaningful results.
>
>    -pd