Q-4: Due April 3, 2014. Reading question 4 requires synthesizing some material from two readings. Use either #1 or #2 as an impetus to a well-written essay of ~3 double spaced pages showing you understand key aspects of the issue. Use quotes from the text. (If you wish to include one of these questions as part of your “short” essay, do the other for Q-4, or talk with me about what you would like to do.)
- Both the criticisms raised by Howson and that raised by Achinstein (in the optional paper) essentially raise the same concern: a hypothesis H (e.g., student S is “college-ready”) may pass severely on my account, and yet by assigning a sufficiently low prior probability to H, the posterior probability for H may be low. So it would seem I am claiming there is strong evidence for a hypothesis even though it is highly improbable (on a Bayesian computation). The identical issue arises more generally for statistical tests of hypotheses, or so Bayesians allege. Analyze the main issue, focusing on whatever interests you, and some part of Mayo’s response (either to Howson or Achinstein). Explain how this relates to the general issue of competing goals: highly probable vs highly well probed.
- From the Meehl reading from 3/27
I suggest to you that Sir Ronald has befuddled us, mesmerized us, and led us down the primrose path. I believe that the almost universal reliance on merely refuting the null hypothesis as the standard method for corroborating substantive theories in the soft areas is a terrible mistake, is basically unsound, poor scientific strategy, and one of the worst things that ever happened in the history of psychology.
It is true that the substantive alternative hypothesis might be said to entail (or make it probable) that the null will be rejected, but taking a rejection of the null hypothesis as evidence corroborating substantive theories is a fallacy of rejection (of the first kind*). This is discussed in Mayo and Spanos 2011. Yet this very famous criticism of tests is standard. Analyze and discuss aspects of this issue, keeping close to the readings for 3/27 and 4/3 (Meehl, Mayo and Spanos).
*(A fallacy of rejection of the second kind infers a discrepancy larger than that warranted. That is why Meehl complains that improving the test with increased sample size actually subjects the alternative to a less risky test.)
(Aside: Meehl is one of the best known original critics of significance tests in psychology. He was a Freudian, as well as a Popperian, and while Meehl accepted Popper’s criticisms of the ability for psychological accounts (i.e., Freud, Adler) to interpret any result in keeping with their theory, he was also irked by the practice of corroboration of a substantive psychological theory by merely rejecting a null hypothesis. Yet those attempts were regarded as scientific, and Freudian data merely anecdotal, or less than fully scientific.)