IV-DV; Challenge 2

[From Bill Powers (930329.1830)]

What a day!

Greg Williams (930329-5) --

I think if we perused the refereed behavioral science
literature together, we'd find a lot of cases where
correlations are found between two population measures, and no
generative model for IV-DV variables of individuals is

You have better access than I, although I'll look, too. Let me
know what you find, when you have time.

I seems that any case in the field, where ongoing disturbances
cannot be known by the experimenter, might be "hard" for
succeeding with The Test.

True. When the experimenter isn't there to ask questions and
observe what's happening, it's pretty difficult. However, with
enough verbal transaction and cooperation, it's not impossible. I
think that you and I will manage to find some of your controlled
variables even under the present circumstances. You're not that
hard to figure out.

You to Mary:

Maybe Bill can convince me that there aren't any problems with
the kind of introspection he wants me to use.

It sounds as though you don't want to try a method of
introspection that you see as having some problems. Is this true?

Or maybe he will come up with a different way of attempting to
meet my challenge without relying on my own impressions.

I trust your impressions. Don't you?

Shouldn't the Test work even on non-verbal creatures? In that
case, such access would be moot.

Pretty tough to apply the Test for control of nonverbal
perceptions over the net.

to me:

Maybe it would be better for YOU to be the judge?

Oh no, I would have won already. That would be too easy. I prefer
hard cases, if I can find any.

... what conclusion would you draw about the possibility of
understanding human behavior by ANY means?

It all depends on the dynamics of disappearing and reappearing,
I guess. Having access to the detailed dynamics some way
(electrical probes?) might in principle be able to reconstruct
individual motivations backward in time, if chaos isn't
problematic (and it might not be, if the dynamics were governed
by limit cycles).

An informative answer. Let me try some test questions now, for
which I think I know what the answers will and will not be.

1. Do you want me to go on questioning you in this same way?

2. Do you want this process to end up with a valid understanding
on my part of one of your controlled variables?

3. Do you sense that I will fail or succeed in meeting the

I predict that your answers will be, very approximately:

1. That's up to you, I don't know what I want to happen, there's
no evidence of any desire one way or the other.

2. I have no idea how it will end up; I can't see evidence for
any preference.

3. My senses don't lead to any prediction; I'm just waiting to
see how this will turn out.

I predict that the answers will NOT be

1. Either Yes or No.

2. Either Yes or No.

3. Either Yes or No.

Having revealed my predictions ahead of time, I expect you to
play fair. Note that I still haven't said what I think you're
controlling for. I'm still gathering evidence, without a lot of
help from you.


Bill P.