There are very many ties in the data. Kendall's rank correlation treats ties
in a more systematic way and would be preferred here. The large sample size,
however, means that it will make very little difference.
Back to question.
This page maintained by Martin Bland.
Last updated: 8 April, 2004.