Posted to Karl and to edstat. An interesting distinction, but one I myself have not found necessary to make. Thanks for the reference!
"Null" is an adjective and "nil" a noun, at least in ordinary English: so the phrase "null hypothesis" can be said to make grammatical sense as it stands; while "nil hypothesis" must be construed as a kind of shorthand or abbreviation for "a null hypothesis that specifies the value of the parameter of interest to be nil". And I would point out that not all hypotheses of "zero effect" specify a value of zero for the parameter of interest: the idea of "zero effect" may well be represented by the value 0.5 for a population proportion, for instance. On Wed, 25 Feb 2004, Wuensch, Karl L wrote: > Jack Cohen [The Earth is Round (p < .05), American Psychologist, > 1994, 49, 997-1003] made a distinction between a NULL hypothesis and a NIL > hypothesis. The NULL hypothesis is that which is being directly tested. > With the so-called "parametric" tests, this is the hypothesis that specifies > an exact value for the tested parameter (such as mu = 10, mu <= 10, or mu >= > 10) rather than not that (such as mu NE 10, mu > 10, or mu < 10). The NIL > hypothesis is a null hypothesis that specifies a zero difference or zero > effect, such as (mu1 - mu2) = 0, rho = 0, phi = 0, eta = 0, and so on. < snip, the rest > Cheers! -- Don. ------------------------------------------------------------ Donald F. Burrill [EMAIL PROTECTED] 56 Sebbins Pond Drive, Bedford, NH 03110 (603) 626-0816 . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
