Re: Success-fail data: very different results with Binomial or Wilcoxon

Thanks, JW, that sounds like a much better way to do it. I'm slightly
intimidated by generalized linear models, but I am fairly comfortable
with general linear models so hopefully I'll manage. Actually another
thing I wanted to do was compare different treatment groups, as well as
just looking for a deviation from 50/50 in one, so using a generalized
LM has got to make sense.

Cheers,

Ben

.