R31802 run in Production vs Test Discrepancy

Hi,

We backed up our production data a few hours prior to our nightly run of  R31802.  We restored that data to the test environment and ran the R31802 there.  We expected the test number to be close, but probably a little less than the production run executed shortly after the backup.  What we found in two of our runs is the test run on the backup data had significantly HIGHER totals than the production run.  This of course makes no sense. We obviosuly verified that there was no other activity going on that could have affected the results.  I am led to beleive