Okay, I just ran the numbers this morning, and here is what I came up with on one of my rigs:

The hash rate for the rig was similar to the reported has rate in the live view.  I will have to assume that the reported fees collected for the hash rate are accurate for the coins that were being mined (I have no way to determine the "live value" of the contract being fulfilled)

When the hash rate is compared to the reported statistics (once again assuming that the $US is being reported correctly) over a 30 minute period where my miners were working on the same currency for the entire time with not breaks of service or re-benchmarking, the live assessed value breaks down to 4.3666 cents US per 10 minute reporting period, while statistics reporting for the same period breaks down to 3.884 cents per 10 minute reporting period (over a 22% difference).

I specifically chose a time period without large fluctuations, limited my mining to 2 currencies, averaged over 3 reporting periods, and verified hash rates and expected pay rates once a minute for 30 minutes starting at the reported start time for a reporting period.  I do not know how to make this any more of a legitimate test of hash integrity than this.

I will also say that the payout amount does match the total amount on the statistics page with obvious rounding to the nearest cent only on the final payout amount (not every reporting period like some organizations do to pad the bottom line).

I will gladly reverse this post and fully support Winminer if there is a reasonable explanation for this, but for now what my log and the live reporting are very close to what my computer is telling me it is doing.  And the idea that the live reporting is in error (even if my computer is telling me that it is doing the work that is being reported) and the numbers on the stats page are correct (even though they don't even come close to what is being done by the rig), and trust them they are doing the right thing for us, I am having a hard time believing that.  Especially after the NiceHash and the MtGox incidents that I have worked through.  I don't begrudge a ring from a little skimming to aid in the founders retirement plan (after all they did provide the services that you are making money on), and I don't bitch about a 3% fee for transfer of funds (the price of doing business).  I do think that 22% skim is going a bit too far to look legitimate to me.  What I believe is the live reporting is reporting the correct mining, but they have not figured out how to add (or subtract in this case) the skim program into the live version, so the discrepancy is being blamed on the live reporting being inaccurate.  I might be paranoid (doesn't mean that they are not out to get you all the time), but I know what my equipment should be doing, I know what work my equipment has done, and I know that the amount being reported is not correct.

Please let me know if you have any questions about this analysis.  If anybody has any reasons that would explain this discrepancy that I have not listed or discussed please let me know.  The truth is the ultimate sanitizer of all theories.