Revealing Statistical Independence of Two Experimental Data Sets: An Improvement on Spearman's Algorithm

A high effective statistical independence test procedure derived from Spearman’s Rank Correlation Test is presented, applicable to all kind of continuous variables (normal or not, even of unknown probability law). Some relevant practical signal processing test examples as well as a Monte Carlo performance comparison with Spearman’s Rank Correlation Test capabilities are explained. Besides describing the test procedure algorithm, the paper reveals, from an engineering point of view, some significant aspects concerning the understanding (perception) of the important and not simple concepts, i.e. testing dependence versus statistical independence.