···
Hans Blom 971128
What actually made me leave was not the contents of the discussions
but their tone. Take this, for instance:
Bill Powers (971125.0706 MST) to Martin Taylor (971121 23:08)
I appreciate the sudden change in tone, now that you understand what
we have been talking about.
I've met Martin,... He
doesn't deserve this kind of derision.
----------------------------
I feel I have to comment on this.
In Bill's messages I have never felt "derision" or insult of any kind. I
have experienced frustration at my inability to get him to understand
things that seem to me to be self-evident, as I am sure he has at my
inability to understand things he sees as self-evident. And I'm sure
I have said things that he has perceived as being every bit as insulting
as you perceive this quote to be.
On the particular quote in question, I was amused rather than insulted,
because I had perceived in my messages on the topic (use of group data
for studies of individuals) no change in either tone or content. But in
the long message from Bill, I had understood that he at least had come to
see what I had been talking about, in part. And I agreed with most of
his posting. So I was amused that what I perceived as his change in tone,
now that he understood what I had been talking about, was ascribed to
my own change in tone and understanding.
The course of true communication never runs smooth, eh?
Or should I say:
Wud some Guid the giftie gie's
Tae see oursens as ithers see's.
(Burns, more or less)
-----------------------------
On the issue of the use of group data for studying individuals, I always
did say that Bill's study was valid in demonstrating the impropriety of
using "between subjects" results to impute "within subjects" effects. I
haven't changed that. Neither have I changed my opinion that group
averages of within subject values can be useful in guiding research, and
sometimes (as in the case of the unfortunate Ms Y with breast cancer) in
guiding actions in respect of individuals.
Granted, I initially (Martin Taylor 971120 15:40) pointed out quite
different issues related to the paper, and those issues haven't gone
away. But they are less important, in my view, than the "between/within"
issue.
And I grant Bill that many psychological papers do make this very error.
I maintain that I was right to say [Martin Taylor 971122 21:40]:
What Bill showed is that it is improper to take the slope of a set
of averages and say that it is the average of a set of slopes.
and
I grant you that there are some pretty
unmathematical people in psychology, who might not see this fact as
obvious, so the demonstration is useful.
But I was wrong to say, between these two quotes:
But most people knew that already
as I acknowledged in a following message.
Now Bill says (Bill Powers (971125.0706 MST)), in my view correctly:
The most common assumption is that if one can find a correlation between X
and Y in population data, then that correlation exists in each person in
the population, at least to some extent
They do, and they shouldn't. Bill's study makes that glaringly obvious,
as he says, expanding my comment about "slopes":
The assumption that accompanies all studies of this kind is that the group
trend signals an individual trend of the same kind. Of course it does no
such thing: the individual trends can all be different and can include
trends in the opposite direction. I showed an extreme case, where the
individual trends are _all_ in the opposite direction, which we happen, in
that case, to know is a fact because we have inside information about how
the "people" actually work.
What I never could understand was why I should be cricitized on the grounds
that I said earlier what Bill said later, in different words.
However, criticism is not insult, nor derision.
---------------------
I think I honour Bill's work as highly as that of any scientist. I think he
makes mistakes, and he doesn't know _everything_. When I think something
he says is mistaken on the basis of something I know about, I'll say so
if it is germane to PCT (and perhaps if it isn't). If I get frustrated,
I sometimes say things I regret later, but they've been said, and can't
be unsaid. Sometimes Bill takes them amiss, as I do some few of his comments.
But there we are. That's life.
And I labour under the impression that Bill treats me likewise.
Lastly, I don't think either of us believes PCT or HPCT to be a finished
monument to the genius of Bill Powers, immune to defacement by well-meaning
enthusiasts. Bill often says that he wants PCT to be improved, but only
by well justified experiments. Therein we differ, since I give more weight
than he does to well justified theoretical modifications. Perhaps we
differ on the criteria for "well justified." But that's OK.
----------------------
Bottom line. Unlike Rick Marken, I would be very sorry if Hans followed
through in withdrawing from the discussion. I have found his arguments
thought-provoking, even if he comes from a basic viewpoint a little
different from that of most contributors (or perhaps because of that).
All my academic life I've been a proponent of what used to be called
"the Bionic approach," according to which one conceives of a technical
problem and then looks at animals to see how they solve it (bat sonar
is the classic example), and then looks at technological solutions to
see why animals are doing weird and wonderful things (such as bat
frequency sweeps). The technological and the physiological go hand-in-hand
much more powerfully than either goes alone. And I think the Blom-Powers
stand-off reflects a failure to appreciate this.
Martin