Audio

[various audio recordings to come...]

 

Philosophy

Herbert Feigl and Paul Feyerabend: "Psyche & Soma" [mp3]

    Discussion of the mental and the physical, the mind/body problem in philosophy, before a Minnesota Philosophy of Science Center audience, November 26, 1962. Feigl speaks first, followed by Feyerabend (at 33:28); then begins (at 55:33) a back-and-forth discussion between them. (Paul Meehl may have been in the audience, but he is not heard on this recording. Starke Hathaway asks a question at the end of the discussion.)

 

Talks

Critique of Null Hypothesis Significance Testing [mp3]

Center for Cognitive Studies (CSS), Psychology Department, University of Minnesota, 01/30/2003 [Handout_and_notes]

    In this last talk he delivered, Meehl distinguishes the weak and strong uses of tests of statistical significance (TOSS) in hard sciences (making numerical point predictions) and in soft sciences (predicting only differences between groups). He covers null hypothesis statistical testing (NHST) in psychology; mentions why Skinner didn’t need significance tests; his own early research to test his theory of schizophrenia; the impact of Karl Popper’s Logic of Scientific Discovery and definition of a scientific theory as one that makes predictions that could refute it (and that Meehl was not a “disciple of Popper”); Bayes theorem; David Lykken’s “crud factor” (In soft psychology everything is correlated with everything; in large samples the null hypothesis is almost always false and correlations are sizeable); valid and invalid logical structure (most scientific research is structured as an invalid syllogism: “If p then q; observe q; infer that p must be true” is not a valid argument because other p-theories could also explain q); the power function; additional factors involved when testing a theory (auxiliary theories, ceteris paribus clause, instrumental auxiliaries); why meta-analysis (intended to evaluate interventions) is not appropriate for appraising theories; NHST is appropriate when substantive theory and statistical hypothesis are essentially equivalent, but they are not equivalent when appraising theories in soft areas.