Power analysis for non parametric tests (sample calculation)

[mousme]

Hi, I am preparing a little presentation on power calculation for parametric tests and it occured to me that I don't know how it works for non-parametric tests. I remember reading somewhere that you basically use the same sample size as for the equivalent parametric test and multiply it by 2 or 3 ... but I am not sure. Any idea? Thanks!

[Karabiner]

Multiplication by 2 or 3 cannot be true.

http://stats.stackexchange.com/quest...etric-analysis
http://www.marinespatialecologylab.o...r-analysis.pdf
http://www.psycho.uni-duesseldorf.de...r-tutorial.pdf
(Search the pragraph Nonparametric Tests in that paper)

or http://tinyurl.com/9utye5t

With kind regards

K.

[mousme]

Thanks a lot, this is really helpful,
A.