From: Katharine Thayer (kthayer@pacifier.com)
Date: 11/07/00-05:50:03 AM Z
I think you missed my point. I have no problem with theories and models
that make sense, and I never said such things shouldn't be discussed, I
was simply pointing out, as a statistician, that I don't think the
particular multivariate models being proposed are a good choice at all
for our situation, and would more likely lead us farther astray than
give us better answers. There are places that those models make sense,
but they are used inappropriately and carelessly these days more often
than not I'm afraid, resulting in bad science everywhere one looks.
On a sub topic, to answer my question "how would we measure the outcome
quantitatively?" Wayde points to chemistry, step tablets and
sensitrometry. If these are your outcome variables, we're not on the
same page at all. To me they are a means to an end but not an end in
themselves surely; these are all "predictor variables" not outcome
variables. The outcome is the print and whether I think the print
communicates effectively. And I still don't see a *meaningful* way to
measure that quantitatively. Lest my overall point has been lost in this
side issue, let me repeat that even if there were a way to measure the
outcome that we all agreed on, the models being recommended would give
no less confusing information than we have now, trust me. As for being
able to get more information with less testing, well, yes, but you still
should have at least 20 times as many observations as you have
variables, even if the model is appropriate for the data. Of course if
it's not, it doesn't matter, because the results will be uninterpretable
no matter what you do.
Katharine Thayer
J. Wayde Allen wrote:
>
> If theories and models are bogus concepts, then why do we talk about how
> any of these processes work? We've all listened to Judy's theories about
> gum printing. Katherine Thayer has also offered theories of her own. If
> theories aren't worth discussing then do we throw these out too? Sure
> theories fail, but they do give one something to work from. What is
> particularly amazing is that I've not given any offer of a printing theory
> here at all. I've also not specified any particular printing model
> either. What I've tried to do is offer a different framework to hang the
> information we've already been gathering on. The idea is to try and give
> us some hopefully better insight into these processes so that we can
> improve on, and help develop new and/or better models.
>
> Anyway, if you feel that what I've said doesn't apply to alt process work,
> that is certainly your prerogative. Those of you who might be interested
> have some starting references.
>
> (Oh, and by the way Judy, there is a theoretical explanation for why
> bumblebee's fly. The problem wasn't so much that it couldn't be
> explained. It just doesn't follow the classical Bernoulli principle like
> airplanes do. You have to resort to turbulence theory.)
>
> - Wayde
> (wallen@lug.boulder.co.us)
This archive was generated by hypermail 2b30 : 12/01/00-11:46:56 AM Z CST