I found this on Wikipedia:
I have a dataset that is log-normally distributed and I would like to know how to accurately calculate the CV and Standard Deviation. To get the mean (or rather the geometric mean) I first transform the data, calculate the mean, and then un-transform the data. It is my understanding, however, that the same can not be done for the CV and STD. Do you have any suggestions?
Thanks for the link. Any thoughts on the CV?
Wild guess: geometric std dev divided by geometric mean?
Advertise on Talk Stats