Suppose I have a theory (or model) T which makes an assumption A (A is necessary for T). Is a test of T a (indirect) test of A as well? That is, does confirmation of T entail confirmation of A?
Suppose T makes several assumptions, A, B, C, D. Now suppose we test T. Is that test a test of A, B, C, and D? Also, does confirmation of T spread equally over the set {A,B,C,D}, assuming that confirmation of T is also confirmation of its logical consequences?
The main worry is whether a test of a theory is a test of all its parts, the parts necessary for making a prediction.
This is quick, but I'm just heading out the door. Intuitions please.
Subscribe to:
Post Comments (Atom)
4 comments:
What is this, experimental philosophy on the cheap? You might need a bigger sample size.
I'm with Duhem and Quine on this one, I'm a holist baby. Though at the same time different theories might make the same assumptions but different predictions. Similarly, when there are many assumptions, some are more "entrenched" than others, closer to the core as Quine would say, and therefore are in principle tested along with observable predictions they allow to be made. But most theories share those "core" assumptions so they will be confirmed in most theories despite the fact that those theories might be making different predictions about observable things.
But you already know all this stuff. So why ask? Are you really interested in intuitions?
wow I feel really dumb trying to understand that Brian. Thanks
So here's my concern. certain models in psychometrics make the assumption that psychological traits have a quantitative stucture. One guy says that it is bad that psychometricians do not test this assumption. Borsboom says that the assumption is tested (albeit indirectly) when the models are tested. So the assumption is never itself tested, rather the theory which presupposes it is tested.
Now the question is whether when one finds confirmation of the theory (model) one is thereby accruing confirmation for the assumption that psychological traits are quantitative. I want to say no. I want to say that there are assumptions that, though they are necessary for generating predictions, are not tested in hypothesis testing.
Take some hypothesis in physics. any one of them seems to entail that there is an external world, or that there are physical objects, or even that physical properties are quantitative; however, no one takes the confirmation of a theory in physics or quantum mech. to be confirmation that there is an external world. Evidence for the theory does not accrue in favor of that assumption even if the tested theory is confirmed and assumes that (auxiliary) hypothesis.
Thoughts?
I agree Brian. I don't think that confirmation of a theory entails confirmation of an assumption. Like hyponyms in linguistics. For example, all sows are necessarily pigs, but that doesn't mean all pigs are sows. Wait...I don't really know what you are talking about.
Post a Comment