Rich Ulrich wrote: > - by the way, I have not yet seen this post (on sci.stat.edu) > that dfb has cited.
Nor have I. ---- snip ----- > There are references in my stats-FAQ, and here is an > important comment concerning the classical test by Hotelling, > in comparison to other tests. (I expect that the test above should > be in the Meng article, but I don't remember that detail.) > > * * * from my stats-FAQ. Citation, and other comments. > > Meng, Xiao-Li, Rosenthal, Robert; and Rubin, Donald B. (1992). > "Comparing correlated correlation coefficients." > _Psychological Bulletin_ , 111, 172-175. > > Hotelling's solution is included in Ferguson's textbook. > > [June 17, 2002: vol., above, corrected to '111'. > Comment: Hotelling's is an exact test of a particular hypothesis, > one that tests positive correlations against a *residual* > of error variation in the criterion. The articles > I have read have not made clear that one test is proper when > the other is not. > > === Communication to me, 2002, from Paul von Hippel > " ... if you read the appendix > and related articles you realize that they're confining > themselves to the case where the regressors are random variables. > If the regressors are fixed, as in an experimental design, > then Hotelling's test is appropriate. Hotelling (1940) was > quite explicit about this, so what Meng, Rosenthal, & Rubin > are really criticizing is the mistaken practice of using > Hotelling's test with random regressors. > > "Williams (1959) adapted Hotelling's test to the case of > random regressors. In simulation studies Williams' test > has held up quite well against the alternatives described > by Meng, Rosenthal, & Rubin. This is all in the papers cited > in MR&R's bibliography." > === end of communication ]. > * * * end of extract from my stats-FAQ. > > p.s. I've been asked to compare correlations when the > two variables were supposed to be 'diagnostic predictors' > and they differed from each other by the addition of an item > or two.... so the two predictors were correlated with each > other at about 0.95 or more. That is a place for looking > directing at the value of the "change" rather than performing > a test on correlations. In "Statistics for Psychology" (1997, 4th Ed), Dave Howell describes the Williams (1959) test as "better" than Hotelling's (1931) procedure--I don't know on what grounds. Williams' procedure is the one used in Jim Steiger's MULTICORR program, apparently. MULTICORR can be downloaded from Steiger's website: http://www.interchg.ubc.ca/steiger/homepage.htm Cheers, Bruce -- Bruce Weaver [EMAIL PROTECTED] www.angelfire.com/wv/bwhomedir/ . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
