 |
 |
Hi Murray There is a whole family of them. The grandaddy is dQ/M i.e. the interquartile range divided by the median. But instead of the 25/75 percentiles you could use 10/90 or whatever you choose. (These all go back to Francis Galton, 1820-1910.) Regarding assymmetry, why is this a problem? You could try the power transform that brings them to symmetry (i.e. so that M is in the middle of dQ). But I don;t know the properties if you do this, so I should stick with the original data. Good luck! JOHN On 16 January 2012 14:34, Murray Doyle On 16 January 2012 14:34, Murray Doyle <[log in to unmask]> wrote: > Hi All - to compare dispersion between a number of different distributions > with quite different average values, I was intending to use Coefficient of > Variation. > > However only some of distributions are symmetric - some are skewed with > long tails - is there a median equivalent to the Coefficient of Variation > (sigma/mu) to deal with this asymmetry? > > Many thanks. > > Murray Doyle > > > > You may leave the list at any time by sending the command > > SIGNOFF allstat > > to [log in to unmask], leaving the subject line blank. > You may leave the list at any time by sending the command SIGNOFF allstat to [log in to unmask], leaving the subject line blank.