[R] Permutation Test on Interactions {coin}

[Torsten Hothorn]

Axel,

you need a model for such type of analyses and coin is completely 
model-free.

Torsten

On Mon, 23 Sep 2013, Axel Urbiz wrote:

> Dear List,
> I'm interested in performing a permutation test on the interaction between a binary treatment
> indicator and a covariate (either continuous or categorical). I'm interested in the p-value
> of the interaction effect from a permutation test, and I'm using the coin package for that
> purpose. 
> 
> As I haven't seen any examples like this in the package documentation (or anywhere else), I'm
> not sure how to specify the test in this case. For example, should I interpret the p-value in
> the example below as the pvalue of the interaction effect between group and the covariate x. 
> 
> set.seed(1)
> library("coin")
> data("rotarod", package = "coin")
> x <- rnorm(24)
> rotarod <- cbind(rotarod, x)
> pvalue(independence_test(time ~ group * x, data = rotarod))
> 
> 
> Your advice would be much appreciated. 
> 
> Regards,
> Axel.   
> 
>