12 Days: Trackers tease a big debate bump for Obama. Maybe.

I’ve been saying for days that Friday is the day you want to watch for polling, and only trackers, and only PPP and Rasmussen. Those two firms have the shortest sample frame which makes them simultaneously the most sensitive to change and most subject to noise. Right now, today, October 24th, both PPP and Rasmussen only have a single day of post-debate reactions in their average. Their surveys began on the Sunday the 21st and ended on Tuesday the 23rd, but Tuesday has no debate reactions because it’s likely that all of that data came long before the 9pm debate began.

Gallup won’t have a full frame until next Tuesday at the earliest, but given the frame size of seven days, you can get a real feel for what the result is once they’ve got 4/7 debate reaction days under their belt, which would be Saturday’s report.

That said, we’re already seeing meaningful movement in the trackers towards Obama, and the size of these movements combined with how quickly they are showing could indicate a rather large bump coming for President Obama this week. Perhaps as big as Mitt Romney’s, putting the race out of reach for Romney. It’s also possible that Obama simply had a very good couple of days right before the debate.

You can’t and won’t know which it is until some of these trackers get more data. But here are things you can tell right now.

– PPP and Rasmussen each have a single day of the debate in their three-day survey average. Rasmussen (shockingly when you keep reading) didn’t move at all. PPP moved from Romney up by 2 to a tied race, a +2 movement towards Obama.

– RAND, which doesn’t use a random sample, but instead repeatedly samples the same group of ~3500 people on a rotating basis (1/7th of the group is polled per day) which is a really cool experiment that could show true trends and *no* noise, but not necessarily an accurate absolute result, showed Obama moving from +1.9 to +3.8, a +1.9 point gain. That’s a result that completely covers the debate.

– Ipsos/Reuters which does a six day sample moved from a tied race to Obama +1, a movement of 1 point towards Obama.

– Gallup, which does a seven day sample and only has 14% of its average covered by the debate, showed Romney losing two points, from +5 to +3, the worst showing for Romney with Gallup since October 16th. It takes a lot for Gallup to move two points. It hasn’t moved two points since the 10/16 – 10/17, when it inexplicably moved from +2 to +6 for Romney when nothing happened during those days.

– IBD/TIPP, which does a 6-7 days, moved from Obama +2.6 to Obama +3.

– UPI-CVOTER, which appears to do a 7-8 day sample, moved from a tie to Obama +2.

To make this more clear, here are the trackers as a difference from the previous day, yesterday and today:

Although Mitt Romney leads the trackers by an average of 0.7 points, and the median is tied, the conclusion is indisputable. Obama has gained an average of 1.3 points across seven trackers with 6/7 moving in the same direction today, and is moving forward. Rapidly. Perhaps only for a day, but today for certain.

The question is why.

You can eliminate the possibility of one really good polling day for Mitt Romney falling off the average, dragging the average and median towards center, because we’ve got enough trackers with different time frames that five of the eight lost different days off their average. For example, PPP and Rasmussen aren’t counting Saturday the 27th any longer, but Gallup, Ipsos, IBD, and UPI still are. If the 27th was a very good day for Mitt Romney, above average and an outlier day, then that could explain the movement in PPP’s poll without Obama having gained anything on his own from the debate. (But Rasmussen’s non-movement would contradict that finding, anyway.)

Likewise, you can discount a small and positive bump for Obama earlier in the week (say Friday or Saturday) that may be pushing Gallup and the longer trackers towards Obama, because the short trackers would have discarded those days by now.

Too many trackers covering different lengths of time all moved in the same direction by roughly the same amount, with the only commonality being the addition of Tuesday’s results to their averages. And as I noted, to move PPP and Rasmussen with a single day is relatively easy, which is why the signal-to-noise ratio (SNR) is low (bad). It takes a much larger single day performance to move Gallup, IBD, and UPI.

My conclusion — you are free to form your own but you better have some solid explanations for them — is that Obama had a very good polling day on Tuesday the 23rd, big enough to move the seven-day trackers, but not so big that it blew up the three-day trackers. I can’t explain that, but I don’t need to. All I need to do is what you have to do, which is consider what I’ve written here today, mull it over, and wait for more data.

FYI, there haven’t been any new national polls out today as of publishing. I may look at state polls later, but probably tomorrow.

There are a few new state polls out today. Rasmussen has Obama up by 2 in Nevada and I really don’t think that state is a tossup. Four new polls out of Ohio are Obama +3 (SurveyUSA), Obama +2 (Lake Research), Obama +5 (Time Magazine), and Tie (Rasmussen). Ras is looking like an outlier in Ohio, here are the last 10 polls, oldest to newest:

Obama +1, +3, Tie, Obama +5, +1, Tie, Tie, Obama +3, +2, +5, Tie.

Ohio may be somewhat of a tossup, but it is *not* tied. Obama is leading by an average of 2 points in the last ten polls and last five, as well. Mitt Romney can’t lose Ohio and win the election at this point. Even with Florida.

Obama may be pulling ahead in Virginia as well. Two polls out today show Obama up by 1 (Mellman) and 3 (JZ Analytics/Newsmax). Obama has lead 4/5 recent polls there, and 5/7.

Finally, Romney got two decent polls out of New Hampshire, +2 from ARG and Rasmussen yesterday. But NH only has 4 electoral votes. Ohio has 18, Virginia 13, and Nevada 6. See all the data for yourself.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.