I discounted Page splits/sec because page splits are also caused by INSERT statements, and therefore if you wanted to measure the rate at which UPDATE statements cause data to be moved, you may get some misleading results.

Have I missed something, or did I misread the question?

John