Quote from trefoil:
There were a few good theories in the blog post shortie linked to, but I first read about this in Barron's ages ago, and it was supposed to be due strictly to the hedging activity of options market makers: given that more puts than calls are usually traded, closing them out would give op ex week a bullish bias, and the week after, when new puts would be opened, would force the market makers to hedge these with shorts on their underlyings, and so would wind up being net bearish.
We have a lot of sophisticated options folks here, so maybe one of them will come in and tell us if what I read as the reason why was true or not.
I think the caveat in the blog post about small sample sizes would apply to any month-by-month breakdown of this bias. Our children and grandchildren will have better info on this than we ever will.