The initial +11 poll had a very small sample size for a national poll (approximately 450 LVs, I believe?), so this is pretty much statistical noise. With the Pussygate video, the debate, and now Trump going nuclear on the Republican Party all happening in just a few days, we're in a very fluid situation. Voters are still absorbing all of this, and it will take a bit more time for the polls to fully "price in" all the information people have received.
Overall, I think we can say that Hillary is in a fantastic position, and I eagerly wait further high-quality polling to see how strong her lead is.
Based on standard polling conventions and statistical practices, the Clinton +9 3 day average is the only number that should have ever been reported for this poll. One-day polls of a presidential race are statistical malpractice.
I agree. I suppose NBC saw an opportunity to release an additional "poll" that was perfectly timed to capture the "response" to Pussygate. But I think the better thing to do was wait until they had the full poll, and if they want to reveal the single day post-debate subsample, then fine.
