Additional cautionary notes about the Pearson's correlation coefficient

In this paper we draw attention to two usually-neglected problems with the use of the Pearson's correlation coefficient for ordinal scores. We refer to the questions of substantial differences between the marginal distributions of the two variables to be correlated, and the number of ties on each one of the distributions or on both. Finally, we propose a correction factor for the Pearson's correlation coefficient when the number of ties is large. Some empirical examples are given in order to demonstrate these points.