n-wise correlation among binary vectors




Please forgive the naivete of this question; I realize that it's
Stats 101 stuff.

I want to know the name of the types of analysis one would use to
solve the following generic problem.

I have a collection of M items and N properties, and an M x N matrix
of 1s and 0s representing which items have which properties.

The problem is to find clusters of properties (i.e. column vectors
in the matrix) that are strongly and/or significantly correlated
(I realize these strength and signficance are different; I'm
interested in both, together and individually).

I know of methods to cluster the vectors hierarchically (I think),
but I have no basis for deciding the cutoff distance for these
clusters.

Thanks for any pointers!

kj

--
NOTE: In my address everything before the first period is backwards;
and the last period, and everything after it, should be discarded.
.