Stephen Downes’ OL Weekly last Friday (1-24) contained a bit of a rant on Karl Popper this week. It was an aside under the heading Free the Facts! (Free the Fact (the article) advocates open access to research journals, a very worthy cause, but SD took exception to the view of science expressed.) I disagree totally.
I admit that Popper, like many philosophers, tends to be obtuse; writing more for other philosophers not scientists or the public. But, I believe the gist of his argument (at least for scientists) is that: you can’t support a theory or proposition based solely on one or two studies. You can prove that you specific proposition is false (at least usually with 95% certainly), but you can never be certain that your proposition really captured the actual cause in the correct fashion. Many studies seem to act as if confirmation is true, but confirming propositional claims is a complex and broad-based task. Confirmation should be based on bodies of work not individual studies. It can also be supported with effect sizes in meta-analysis, with power reporting and with validation studies. (See Brualdi, 99, for a complete idea of what the concept of validity can entail.)
This viewpoint does many the translation of research to practice difficult, but it really was easy, we would have solved most of our pressing problems long ago. Assuming an easy confirmation process is not helpful. It makes the process even longer and erodes confidence in research efficacy. (Think fashion in educational research as one example.)