[ topic: Dennis and I figure that a z-score is what you get when you 'normalize' the sample by the mean and SD, for most any distribution; whereas Donald had asserted that the references cite Population parameters. ]
On 25 Aug 2003 13:18:36 -0700, [EMAIL PROTECTED] (Wuensch, Karl L) wrote: > Might Donald be thinking of z scores in the context of hypothesis > testing, that is, the position of a sample mean in a normally distributed > sampling distribution? > Someone confused me that way just a few weeks ago in the SPSS group. What I found with google suggested that a "z-test" is *almost* always called a z-test, and not a z score. (I saw just one or two sites where I concluded, eventually, 'Oh, this person is confused, too.') Similarly, every useful reference to a T score (capitalize T) used a mean of 50 and SD of 10. Most of those sites also talked about z scores. A z test does make sense for Donald's comments (I think), because you do want the "known variance" for the z. Most of the z-tests that I see applied (clinical research) are from 'non-parametric statistics' -- the rank-order test, or the test of proportions, reduces to a 1-d.f. chisquared, or the z-score that is the square root of the chisquared. The variance is "known" because the variance is wholly determined by the N, which (thus) implies the sum of ranks, and sum of squared ranks. -- Rich Ulrich, [EMAIL PROTECTED] http://www.pitt.edu/~wpilib/index.html "Taxes are the price we pay for civilization." . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
