[R] contrasts for lm

[Dieter Menne]

Ann Hess <hess <at> stat.colostate.edu> writes:

> 
> Here is my model statement:
> 
> Model<-lm(log2PM~P+T+P*T)
> 
> where P has 16 levels, T(treatment) has 12 levels and I am interested in 
> looking at different treatment comparisons.

With so many levels, you should do something against fishing for error 
accumulation and use simint(simtest) in package multcomp, which can handle lm-
models.

Dieter