[Statistical interpretations in biology using a coefficient of independence].

From tables of contingency, chi 2-test is frequently used to decide whether two classes of events may be considered as independent or not. After a preliminary choice of a confidence level, the results of this method (rejection or non-rejection of a probability law of independence admitted as our null-hypothesis) are practically very difficult to soften, despite the fact that it would be often interesting to point out some further important informations. We are however able to deduce easily, from these same tables of contingency, a coefficient mu, characterizing the independence and belonging to [0, 1]: when the unit value can be reached by mu, a complete independence appears possible but, for mu < 1, we can also specify the proportion of independence law which is hidden within the complete law of probability associated with our random sample.