[R] Appropriate measure of correlation with'zero-inflated' data?

[Douglas Bates]

I have created some slides on the Bowl Championship series data for my
class.  Copies of just that part are available in
http://www.stat.wisc.edu/~bates/BCS.zip

Two interesting things about the analysis are that a parallel
coordinate plot shows the differences in agreement quite clearly and
that axes on the biplot from a principal components analysis are
readily interpretable.

It's a great example for those teaching multivariate analysis.