On Wed, 7 Mar 2001, Hatcher, Joe wrote:
> Concerning one-tailed tests, I like the following idea,
> which I ran across somewhere I cannot remember where. One
> should only use a one-tailed test when one is making a
> prediction the opposite of which would not be interesting or
> useful <snip>
I agree very strongly with this. Not too long ago, in fact, I
protested to _Science_ about a study by Maurer et al (1999) which
claimed that cataract removal in infants improves vision after as
little as one hour of patterned visual input. Their study
received wide publicity in the press, and was the subject of an
approving essay in _Science_ itself. They had reached their
conclusion by using what I regarded as "creative" (i.e.
inappropriate) use of one-tailed tests. _Science_ rejected my
letter.
In my comment, I quoted Keppel (1991) as stating "Most journal
editors question or refuse to publish research that is based
largely on directional tests". But not _Science_, apparently.
I also have a useful editorial (Levitt, 1994) discussing the
issue, published in, of all places, the International Journal
of Clinical and Experimental Hypnosis.
-Stephen
References
Maurer, D. et al (1999). Rapid improvement in the acuity of
infants after visual input. Science, 286, 108--
Keppel, G. (1991). Design and Analysis: A Researcher's
Handbook (3rd ed.), Prentice-Hall, p. 123
Levitt, E. (1994). The one-tailed test: a statistical editorial.
International Journal of Clinical and Experimental Hypnosis,
vol [Roman numerals, which I can't decode], 4-6.
------------------------------------------------------------------------
Stephen Black, Ph.D. tel: (819) 822-9600 ext 2470
Department of Psychology fax: (819) 822-9661
Bishop's University e-mail: [EMAIL PROTECTED]
Lennoxville, QC
J1M 1Z7
Canada Department web page at http://www.ubishops.ca/ccc/div/soc/psy
Check out TIPS listserv for teachers of psychology at:
http://www.frostburg.edu/dept/psyc/southerly/tips/
------------------------------------------------------------------------