The Insignificance of Statistical Signigicance Testing
An introductory paper on the insignificance of statistical significance testing brought Douglas Johnson the The Wildlife Society Award for Outstanding Publication in Wildlife Ecology and Management, in the Article Category, 13 September 2000, in Nashville, Tennessee.
President of The Wildlife Society, Nova Silvy (right), presents Doug Johnson (left) with the Outstanding Publication Award.
The paper is widely cited online. It’s not deep or technical; I don’t recall a single equation in it. For a paper like this to receive a Best Paper award, it says two things: (a) the field of Wildlife Ecology was new to the problem of significance testing, and (b) the paper was published before everybody’s career was established on the basis of the p-value.
Too bad for Psychology., when the last 4 generations of psychology graduate students were drilled to use t-test, ANOVA, and alike. First to adopt, last to pull out.
Apparently UIC Economist Deirdre McCloskey wrote a similarly titled paper a few years before “The Insignificance of Statistical Significance,” Scientific American (Apr 1995): 32-33. She/he has a whole list of articles on the rhetoric of significance testing and econometrics.
June 17th, 2005 at 2:23 pm e
I think there is still nothing better on this topic than Paul Meehl’s 1978 paper, which is available online here http://www.tc.umn.edu/~pemeehl/113TheoreticalRisks.pdf
June 17th, 2005 at 2:40 pm e
I just re-read the paper. I forgot how dense it is, but it does make a number of very useful points if you can plow through the rest. Sadly, these aren’t new issues.