[From Bill Powers (939329.1415 MST)]
Greg Williams (930329-4) --
... the "IV-DV approach to studying behavior" also includes, as
I've been trying to say recently, deriving statistical measures
of population characteristics, with no commitment to an IV-DV
generative model.
I'm not sure it doesn't. If a population study finds a
correlation between taking aspirin and a decrease in heart
attacks, doesn't this carry the implication that if you take
aspirin, you'll have a decreased risk of a heart attack? It's
certainly publicized that way.
It's hard for me to think of a circumstance in which, once an IV-
DV relationship has been found, the IV isn't then used as a
predictor of the DV. Maybe the prediction is just "yes-no", but
that's a generative model. Not one of your GREAT generative
models, you understand.
You to Rick:
You are currently just ASSUMING it will, not having tried, I
think, to apply the foundational method of testing for
controlled variables to "hard" cases in the field. Want to join
Bill Powers and try it on me?
You should have said that to me. Are you trying to present
yourself to me as a "hard" case? You're not a hard case.
···
----------------------------------------------------------
Best,
Bill P.
--------------------------------------------------------------
FROM MARY:
[from Mary Powers 9303.29]
Greg Williams
What's going on? You want PCT to explain your behavior when you
yourself don't have access to your own motivational control
structure? If Bill gives you an analysis, correct or incorrect,
you're saying you aren't equipped to confirm or deny. So why are
you asking?
Mary