[math-fun] Generalized spanning trees in hypergraphs

Can someone give me a reference to the result of Kalai and Linial (?) generalizing Cayley's formula to spanning "trees" in hypergraphs? I mean the result in which the right-hand-side is something like n to the power of n-choose-k, counting generalized spanning trees of the complete k-hypergraph on n vertices. Thanks, Jim Propp