[Correlation between two variables in a populations truncated with regards to one of them (author's transl)].

This paper concerns the comparison of the correlation coefficients between two variables X and Y, a) for a whole population or b) in the sub-population in which X is greater than xo. We show a simple relationship between these two coefficients which depends only on the ratio between the sub-population variance and the whole populations variance. The corresponding graph is drawn. The observed differences between these two correlation coefficients can be large and must be taken into account for the interpretation of such data.