In article <[EMAIL PROTECTED]>, [EMAIL PROTECTED] writes (stuff trimmed here, partly for space, and partly because it's becoming unclear who said what) > >Why not do something even simpler: Iff all the nulls are true, (and >all distributions are continuous), then all the p-values have a >uniform(0,1) distribution. Make a QQplot of your p-values against the >uniform distribution. > >Kjetil Halvorsen > (trimmed here too)
1) I like numbers as well as graphs because while a graph can be wonderfully illuminating it doesn't always tell me whether I am seeing an insight or an artefact. 2) Neither Holm-Bonferroni nor the quantile tail significance require independence of the various p-values, whereas the shape of your QQplot could be strongly influenced by this. -- A. G. McDowell . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
