r sqaured values for non parametric correlations

[deleted_user]

I am trying to generate r squared values, and their standard deviations or errors, for non-parametric correlations. Any ideas how to do this? Any help is greatly appreciated.

[Doc_Duke]

Standard Deviations are not defined for non-parametric tests. SD's and r-squared statistics assume a normal distribution.

If you are looking for a variability measure for the correlation, your best bet is to do a bootstrap on the correlation and quote the 25th and 75th percentiles.

Doc Muhlbaier
Duke