Thom Baguley <[EMAIL PROTECTED]> wrote in news:3DABEC4F.CBCAC8A7@;lboro.ac.uk:
> It will depend on the journal to which you are submitting the research. > I would go for ii) and iii) provided appropriate N and summary stats > (means and s.d. or whatever are given somewhere in the article)*. > > The use of z or T largely depends on whether the large sample normal > approximation is used. > > Thom > > * means and s.d. (or equivalents) are preferred summary stats because > they can be readily used in secondary analyses - not because a Wilcoxon > T is reported. > > Jan Malte Wiener wrote: > >> hi, >> i am trying to find out how to quote a wilcoxon signed rank test >> correctly in a publication. i have found that (i) only the p-value is >> reported, (ii) the p- and a T-value are reported or (iii) that the p- >> and a Z-value are reported. >> what is the correct way to quote wilcoxon signed rank tests? >> thanks for any answer, >> jan > > OPINION: It is redundant to publish both the p-value and the z statistic. Likewise it is redundant to publish both the p-value and a T-statistic. The people who know this will probably assume that you don't know it, and will adjust their estimate of your knowledge accordingly. You should publish the groups' sizes, their medians, interquartile ranges, and your choice of p- value or inferential statistic (Z or T). David Winsemius. . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
