Comparison of data by Pearson product-moment correlation coefficients is prone to outliers. The problem can be alleviated by normalizing data with outliers before computing the Pearson correlation coefficient. The sample provides such normalization by optimal power space transformation.
