Bear in mind that many third-party producers of questions contract their question writing to recent university graduates (read: not MBAs), many of whom are underpaid (read: unmotivated) and haven't even taken the GMAT.
While trying fresh CATs is a good idea, I advise all clients only to refer to GMATPrep and Official Guide materials when trying to correctly gauge their performance with respect to the actual exam.
Furthermore, no released test works with the exact algorithms of the real test; otherwise, some clever person might reverse-engineer it. Certain among these tests have shockingly primitive algorithms.
In short, take the GMATPrep score as indicative, not the PR score