In article <[EMAIL PROTECTED]>, [EMAIL PROTECTED] wrote:
> Yes, you will want to do studies that involve orthogonal arrays, for
> those question that can be formulated to use them. I wonder about your
> issue of order/sequence, nonetheless. If order makes a difference, or
> is adjusted, then the order is a factor, and the number of factors you
> must consider just blossomed, greatly.
>
Thanks much for your reply. Why do I need orthogonal arrays? Not sure what
you mean about order. We have subjects do a task in which they first do x,
then y, such as they see the problem, then read it, then see an answer,
then judge it's correctness, then make a response, etc. Don't understand
how the order of events is a factor. Most doing this kind of
research/analysis simply create a regressor for each of the periods, such
that it is 0 at all times except during the reading, or solving, or
calculating, etc. By determining how much each of these regressors
contribute to the overall movement of a brain spot's intensity variance
throughout the entire experiment, we can see what spot in the brain
responds particularly to that cogntive activity (and not others).
For example, it is expected the regressor for making the response,
elevated around the time our reaction time indicates subjects to respond,
should select regions in the left motor cortex that controls right hand
movement.
> traditional DoE's address a (small) portion of the system at hand, in
> such a way that they usually avoid feedback _within_ the experimental
> conditions. Only you can understand if this is an issue, although an
> experienced DoE person (statistician or otherwise) can often help dig
> out the phenomena.
I don't understand this at all, you see why I asked for a book reference.
>
> If you are willing to handle the real thing, I'd suggest Box & Draper,
> 1987, Empirical Model Building & Response Surfaces. A bit heavy on the
> math, but watch the pictures. And orthogonal designs are all there.
>
> Jay
Thanks for that, I'll look for it.
Jim
--
Remove SPAMBLOCK to reply
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
