NHST and all that

A few weeks ago I was involved in an initially interesting and then increasingly bad-tempered exchange about p. values, confidence intervals, null-hypothesis testing and so forth on the quantitative_methods_teaching mailing list. This probably was not the right forum to discuss the issue in, but in my defense I was provoked...Anyway, the QMT mailing list is a good and useful thing when used properly to disseminate information about the teaching of quantitative methods.

The initial provocation was some crowing about the now well known move by the B (C?) string psychology journal  Basic and Applied Social Psychology to "ban" much of the apparatus of standard frequentist statistical inference (or at least the reporting of it). You can see the original editorial here and the most recent clarification here. For what it is worth, my view is that this is a case of a well meaning but misguided editor collaring the wrong suspects. I doubt it will have much impact in psychology but ripples are already spreading to adjacent disciplines and there is a danger that a lot of ink will be spilled in regaining a sense of proportion as the one-eyed encourage the blind to firmly grasp the wrong end of the stick.

All of which is just a preliminary to saying that if you are really interested in the issues at stake you should read Stephen Senn's guest post  and the ensuing comments on Deborah Mayo's blog. The exchanges between Senn, Colquhoun and Mayo are particularly enlightening (and good tempered). 

You see, people that know what they are talking about can have a discussion without bellowing down a megaphone.