[From Rick Marken (970413.1040 PDT)]
CHUCK TUCKER (970407) --
This is a post from another list (called CRTNET) regarding a book
on statistics which proposes a solution to the problem of relating
aggregate data to individual data.
I've bookmarked it. I'm not sure I understand what this has to do
with the "problem of relating aggregate data to individual data"
that we have discussed on CSGNet. The problem (the mistake, really) that
we have discussed is that of coming to conclusions about individual
behavior based on aggregate data (see Powers (1990)
Control theory and statistical generalizations, American Behavioral
Scientist, (34) p. 24- 31). There is no "solution" to this problem other
than just not making this mistake; it is always a mistake to come to
conclusions about individual behavior based on aggregate data. Perhaps
this article is solving a different problem. Is that it?
Mary Powers (99704.11) --
When we got back from Arizona Friday there was a letter from
Fort Lewis College essentially cancelling us out for the agreed
dates (July 30-Aug 3).
How sweet. Did they give a reason?
We have three alternatives: change the dates, change the location,
or cancel the conference.
I'm fine with all three. But there are people who were planning to come
to the conference on the the original dates, some from far away (like
Wolfgang Zocher and his wife). Maybe those of us who can make
it should just get together in Durango on the agreed on dates (July
30-Aug 3) and meet at your place (we'll stay in accomodations in
Durango). A little more expensive, perhaps, but a small price to pay to
get together and discuss the greatest intellectual revelation of the
century (millenium?)
Martin Taylor (970411 15:15) --
Even in your e-coli demonstration, there aren't any changes at
all when the moves are in an improving direction, no matter how
high the error
It's depends on the gain parameter of the model. If an e-coli has
very high gain (which is basically the probability of a tumble/size
of error) it is very likely to tumble when moving toward the target.
Real E. coli apparently vary (individually) in terms of this gain
parameter; a very high rate of tumbling is observed in some E. coli
even when they are in a positive (toward the target) chemical gradient.
I think you might want to _reorgnize_ your thinking a bit regarding how
the PCT "E. coli" reorganization system works.
Best
Rick