I believe in the stats programs I've written.
I'd suggest that many to most stats reports are valid... just that not everyone reports the same values in the same way. It's tough to do an apples-to-apples comparison when everyone has a different methodology for determining what consitutes a 'hit'.
