It's been pretty commonly noted around here that RealClearPolitics has a certain... let's just say, an interesting slant in what polls it chooses to include and what polls it doesn't. (Read: All Republican schill pollsters count, only PPP counts from the Dem side and then only sometimes)
Today though, it's gotten particularly blatant in a desperate effort to shore things up for Romney.
After Ras this morning, RCP showed Obama at +0.4 nationally in their polling average which included every poll from Ras (11/2-11/4) way back to NPR (10/23 - 10/25) Many, included myself, were anticipating that this would increase once they had to update Gallup and replace the old (ending 10/28) Gallup poll showing Romney +5 with the new one showing Romney +1.
Except that they updated it, and it didn't change. Why? Because they arbitrarily decided that even though this morning they were including polls which finished on 10/25, now - just, by shocking chance, as they had to stop using the Gallup outlier - they were going to cut things off at 11/1. Which, completely coincidentally, sliced out Obama's +5 result from the National Journal and brought his average back down to +0.4.
Not that we really needed any more examples of it, but if you did, there you go; RCP completely arbitrarily changed the cut-off date in their national average as soon as they couldn't use the old Gallup anymore so that they could cut out other pro-Obama results and continue to artificially try and hold down the President's average.
Hacks at the start, hacks at the end. Nice try, guys, but we'll see what happens tomorrow.