The sample correlation coefficient using the distances at which a pair of observations are joined in a dendrogram and the dissimilarity (or similarity) values for that pair. It is a measure of how successful cluster analysis has been in partitioning the data.