iain Wrote:
-------------------------------------------------------
> To clarify, I was wondering if there was ever a
> reason to have something like this [1 -3 -1 1 3],
> an arbitrary contrast that sets some values
> positive, others negative, and they do not sum to
> zero. This is more the kind of thing that puzzled
> me. Why would someone do that? Or would they? I
> think the implication of your answer is that
> non-zero sum GLTs are basically used for tests
> where all terms are positive and unitary. That
> right?

I can't say for sure that non-zero sum GLTs with nonuniform weights and mixed signs don't make sense under some context, but I can't think of a real example right now.

Gang