Arrgnce; opinions;challenge II

[From Bill Powers (930327.0700)]

Greg Williams (930326 - 2) --

I guess I tend to read between the lines some, and what I get
is the "flavor" that modeling individuals is much more
important for Bill, Mary, and you than other methods of science
(especially statistical).

I think that's true; it's true that you tend to read between the
lines some, quite often correctly, and that I at least prefer
modeling to statistical methods. I don't prefer modeling,
however, just because I like modeling, and I don't hesitate to
use statistics when there is a use for it just because it's
statistics. My reason for preferring modeling is that I like
explanations that fit the observations as much of the time as
possible, preferably all of the time. This means that when an
explanation (of whatever kind) fails to fit any valid
observation, I want to know why and I generally attribute the
reason to a failure in the explanation, not to some innate
barrier to discovery such as an allegedly inherent variability of
natural systems. if modeling didn't help to produce explanations
that work, I wouldn't like modeling much, either.

This is what leads to the aim of explaining individual behavior.
The exceptions to generalizations about behavior always turn out
to be some individual acting in a way that doesn't fit the
generalization. If one believes, as I do, that we should reserve
the term "knowledge" to mean explanations with no known
exceptions, then even a single individual case that contradicts
an explanation means that the explanation is still inadequate. We
can live with inadequate explanations, but doing so doesn't mean
we should pretend that they are more adequate than they are.

I think that most scientists actually have the same attitude.
Nobody likes to have exceptions staring one in the face. One way
to avoid exceptions is to back off and make more general
statements, statements about populations instead of individuals.
You can't say that a certain piece of advertising will influence
Greg Williams of Gravel Switch, KY to rush to the store and buy a
widget, but perhaps you can say that in a population of a million
people reached by the ad, at least 100 people are all but certain
to rush to the store. You can't prove that your reasoning about
why these people are influenced is correct, and you don't know
which 100 people will respond, but at least you can say that
events haven't proven your reasoning wrong. After some experience
with making this sort of prediction, you can become pretty
confident that your advertising theory will produce at least 100
customers per million, with no exceptions.

The difficulty arises when the client says "That's all very fine,
but what I want is a return of 1000 customers per million, and
furthermore, I want to be sure that these 1000 people will
actually pay their bills." If you then give in to temptation and
promise what the client wants, you're going to start experiencing
failures of the theory at a significant rate, because all you
really _know_ is that at least 100 persons per million will buy
the product. You start promising more than your knowledge can

This, it seems to me, is the bind in which people who use
statistics in lieu of knowledge about individuals find
themselves, more often than not. In the medical profession, for
example, we have all kinds of claims about what pills, diets, and
physical regimines will do for you, all based on statistical
studies of populations. One result is that hundreds of billions
of dollars are spent each year by people for pills, diets, and
medical attention that do almost all of them no good at all. Yet
the medical profession can point to the statistics and show that
in fact, all these treatments have had a clear statistical effect
on the population. They point out that taking aspirin every day
reduces the chance of getting a heart attack by 40 percent. They
don't say what the chance was initially, or how small a chance
there is that you would have had a heart attack without taking
aspirin. Aspirin is cheap. But what about the effects of taking
expensive drugs that produce even smaller effects in a
population, on problems with even lower incidences?

We have the same problems almost everywhere that statistics is
used in a way that affects people's lives, whether it be in
psychotherapy or scholastic tests or job screening tests. The
knowledge is used in situations where it is actually invalid and
in ways that are actually harmful. That I what I object to,
loudly and strongly. You can call that arrogance if you wish.

There are plenty of situations in which statistical knowledge is
used correctly, in such a way that it could easily pass my
criterion that there be no, or insignificantly few, exceptions.
Those are the situations in which the knowledge is applied to the
same population from which it is derived, and is used only to
make predictions about the population.

Even when the knowledge is limited, it can still be used
honestly. The advertiser could easily meet my criteria by
continuing to claim that using his knowledge of responses to
advertising claims, at least 100 people per million will respond
to his ads, but also making it clear that the chances of getting
1000 respondents are 100-to-1 against. The doctor could meet my
criterion by telling his patient that these $20-each pills, taken
every day for a year, have a 0.2 percent chance of alleviating
whatever is causing the patient's current complaint, and a 25%
chance of not causing something worse. The psychotherapist could
meet it by telling the parent that this test for Attention
Deficit Disorder in the child is 85% accurate, but that the
incidence of ADD is only about 2%, so the chances are that a
positive indication of ADD has only about one chance in 7 of
being correct -- and Ritalin, which is indicated for the
condition, not only seems to help about half (or whatever) of
children taking it, but also causes side-effects of dullness and
other things that children don't like.

Unfortunately, avarice, self-importance -- and yes, arrogance --
usually triumph over honesty. One must, after all, make a living,
and what's wrong with making a profit?.

One manifestation FOR ME of arrogance is assuming that others
should ascribe the SAME importance to one of these as you do.

Does this prevent you from trying to persuade others to your
point of view? You can acknowledge that each person selects goals
privately for private reasons, and has every natural right to do
so, without passing a law that prevents one person from trying,
by nonviolent means, to get others to see an advantage to
themselves in adopting a different goal. If people believe that
certain goals and ways of achieving them will impress others, is
there anything wrong with letting them know that you, at least,
are not impressed? If someone expresses an opinion about
something of general importance, is there anything wrong with
stating that you have a definitely different opinion? I get the
impression that you interpret PCT as saying that people should
adopt a strictly hands-off attitude toward other people's goals,
attitudes, and opinions. I don't read the principles of PCT that
way at all. I haven't noticed others reading them that way when
it comes to arguing with me.

When you say something like "this is what you should want to
do, and doing something else isn't as good," you are simply
saying (arrogantly) that your own goals are better (i.e., "more
important" or "real knowledge") than the goals somebody else

Finding the right attitude isn't always easy, and one doesn't
always manage to live up to the attitudes that seem right. I try
to put the "if"s in my arguments as appropriate, but sometimes I
forget to say "If you want knowledge that fits the data..."
Sometimes I make the mistake of just assuming that everyone would
prefer their knowledge to be supported by experience instead of
being contradicted by it. It still seems rather silly to have to
say that all the time. Why would anyone who prefers theories that
are the worst at making predictions be listening to or
participating in this conversation?

Are you suggesting that the concerns of people who deal with
populations are in some sense unseemly, because "a few"
individuals generally get stepped on?

Oh, yes. Definitely. I think they are doing harm. Unless they are
willing to take real responsibility for the harm they cause, and
take steps to prevent it, they should be doing something else for
a living. That is my opinion and I will defend it.

Are you suggesting that these people could use PCT to do a
better (to you) job in some way?

Perhaps, if they would actually try to use it. It's also possible
that the job doesn't have to be done, or should be done in an
entirely different way that doesn't cause harm.

Perhaps your preference is for the population dealers to simply
quit dealing?

For the most part, yes. That is what I would prefer, if they're
unable to make the effects they have on others commensurate with
their actual knowledge. And before you ask if I would prefer that
doctors stop treating illnesses, please think more deeply about
what I mean. I ask only that doctors live up to their own credo,
which begins, "First, do no harm." And I hope you are keeping
your own (apparent) principle in mind: that it is arrogant to
tell someone else what to prefer.

Personal human goals are the only ones we've got. You seem to
be implying sleaziness or selfishness again here. Do you really
know that much about the people who use population statistics
to say that what they (some? many?) are doing is "bad"

Yes. Many of them. The ethical "badness" is, of course, in terms
of my own ethics. Many of them do things which, if I did them,
would make me feel ashamed; which I would want to present to
others in a better light so they would not know that I am
ashamed, or look down on me. I would be ashamed to deny medical
insurance to a person who is suffering from effects of a previous
car accident, and then turn around and sell the same person
medical insurance at three times the cost, as was done to Mary by
Blue Cross, which happens also to run the uninsurable pool in
Colorado. I would be ashamed at dictating the future course of
another person's life on the basis of a test that I knew had a
significant chance of misjudging that person. I would be ashamed
at doing most of the things that I have seen done by the people
who rely on poor statistics to make important decisions about
other people's lives. I wouldn't do those things, and don't. I am
disgusted about such practices and see no reason to keep quiet
about it. Perhaps people who have a good opinion of my judgments
in other matters will take my attitude as something to be
considered when they form their own opinions. Perhaps not. That's
beyond my control.



OK, the challenge becomes: show me you understand my behavior
and what it is being used to control. But first, YOU tell ME
what a passing grade at such "understanding" would consist in.

My own ethics tells me that if I go around issuing challenges, I
can't very well turn one down.

Passing grade? Why not make it as high as possible? Let's say
that I will have met your challenge if you agree that I
understand some behavior of yours and what it is being used to
control. I'm not claiming that I can get a passing grade, but why
settle for a lesser goal

A couple of questions first. Will you be trying to help me
understand a behavior of yours and what it is being used to
control? The easiest way to find out what a person is controlling
for is, of course, to ask. If the person's intention is to reveal
the controlled variable, then the problem shouldn't be too hard
to solve. On the other hand, if the intention is to make it
difficult, that intention shouldn't be too hard to discern,
either, nor the means of carrying it out. The only really
difficult case would be the one in which a person uses deception,
trying to give the impression of controlling for something when
there is no actual control of it, with the intent not only of
concealing a real controlled variable, but making it appear that
something else is being controlled.

That, however, would be difficult to achieve, because to give the
impression of controlling for something without actually
controlling it becomes impossible when the spurious controlled
variable is disturbed. Either one must control it, thus revealing
it, or let it change under the disturbance, in which case it is
clearly not controlled.

You should be certain that you want to issue this challenge,
because I see no way for the Test for the Controlled Variable to
fail. I trust that you don't mean the challenge as saying that I
must lay out the entire structure of all your controlled
variables from bottom to top and from now to the end of your
life, a project that would require rather a long time and rather
intimate interactions on a continuing basis, not just through
words but through extended personal contact. I think I would have
to admit that I couldn't meet that challenge. I assume, however,
that you have something more practical in mind.


You have issued a challenge to me to explain what variable a
behavior of yours is being used to control. Issuing the challenge
is clearly a behavior of yours; what is not immediately apparent
is the perception in you that is being controlled by issuing this
challenge. So let me try the easiest way first:

What is the effect you intend to produce by issuing this

Bill P.