The Kaiser–Meyer–Olkin measure of sampling adequacy compares the correlations and the partial
correlations between variables. If the partial correlations are relatively high compared to the correlations,
the KMO measure is small, and a low-dimensional representation of the data is not possible.