What is the alternative to falsifiability? I don't ask this as a challenge. Rather, as a person who does a great deal of empirical work there are times when I run a regression and it gives me results different one I expected or the results are statistically significant. Then, I proceed to torture the data in a number of imaginative ways. Yet, often after the most brutal torturing the data will not confess - it stubbornly provides statistically insignificant results or results that are counter to what I expected. At that point, I typically consider that my a priori expectations were wrong or in need of modification. This has happened in every empirical paper that I've written. Surely, modern philosophers of science are not saying that I should abandon checking my theoretical assumptions against the data? What is the appropriate empirical criteria for evaluating a model? Another practical problem. There have been 1 trillion empirical studies on the economics of racial and /or gender inequality. There are 1 million theories of racial and gender inequality. The theoretical and empirical studies roughly fall into camps: market discrimination is possible v. market discrimination is not possible. In the absence of falsification, is there an objective way of determining whether the data supports or rejects discrimination? peace, patrick l mason At 09:11 PM 10/19/00 +0100, you wrote: >It's annoying to have to put in a good word for this vastly-over-rated (not >least by himself) thinker -- but falsifiability is *quite* a good idea as a >test of whether a set of notions is, so to speak, advanced in good faith. >(And his anecdote against Adler, which he uses to illustrate his >disillusionment with confirmationism, is certainly both funny and >well-directed). > >It's just a pity that it's logically disabled as a prescription for practice >(the Duhem-Quine problem); moreover Popper should have known this (since >Duhem preceded him by two or three decades) and the fact that he doesn't >deal with this objection is evidence of either ignorance or deception. > >I note that Nicky (hi -- good to hear from you again) asks for justification >of the claim that falsifiability is the *only* criterion for scientificity; >this would certainly be a hard claim to justify, and even Popper never >claimed this about his stock-in-trade. > >In fact, I seem to recall that he went to some lengths to point out the >curious and unfortunate consequence of the falsifiability criterion that >preposterous theories (e.g. "the moon is made of green cheese") are -- being >highly falsifiable -- highly scientific. > >In his #4166 Jerry questions the usefulness of what he describes as "the >empiricist method" in comprehending capitalism: to try and nip in the bud >any discussion at cross-purposes, could Jerry clarify whether he means >"empiricist" or "empirical"? > >Id' certainly agree with him about the former; if the latter, I think I'd >want to say that empirical investigation helps us to check whether our >would-be comprehension of capitalism is actually engaging with the reality >of it. > >Julian
This archive was generated by hypermail 2b29 : Tue Oct 31 2000 - 00:00:10 EST