Someone please post the derivation of (plus minus)square root of (plus minus)2C-N/N, where C is the number of positive observations and N is the total number of paired observations minus 1. This is actually the formula of finding coefficient of correlation by concurrent deviation method.

Your formula that you provided is not quite correct. Underneath the square root sign it should read +- (2C/N - 1) or +- (2C - N)/N - where the sign of the correlation coefficient takes the same sign as what is underneath the radical (i.e. so that it does not have negative value underneath the square root sign.

Anyway, it would seem to me that the derivation

might be related to the following article:

Cahan, S. (2000). Closeness to identity: A conceptual framework for the definition and interpretations of correlation.

*Journal of Educational and Behavioral Statistics*, 25(4), 437-441.

This article - I'm not sure - may provide you some assistance because the nature it is generality.