Dr Chaos <[EMAIL PROTECTED]> wrote in message news:<[EMAIL PROTECTED]>... > [...] > is this the same as the Wilcoxon matched-pairs test?
No. The Wilcoxon would replace the absolute values of the L-R diffs by their ranks, attach the original signs, and then proceed as you described. Without the ranking, you have a randomization test of the difference between means for paired data. IIRC, the t-test gained acceptance in such situations partly because it gave results that were close to those given by the randomization test, but with much less work, provided the sample size was not too small. . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
