2016 Postmortem
Related: About this forumNate Silver's 538 Blog for 10/5/12
Is he saying that he'd "bet on Mr. Romney now" to win???
[link:http://fivethirtyeight.blogs.nytimes.com/2012/10/05/oct-5-day-after-debate-strong-swing-state-polls-for-romney/#postComment|
ksoze
(2,068 posts)intaglio
(8,170 posts)Obama still at an 83% chance of winning
regnaD kciN
(26,044 posts)...which is what the 85-15 projection equates to. He's saying it would be a good longshot bet.
What you need to keep in mind is that Nate predicted a 2.2% bounce after the debate. Even yesterday, as he notes, that was looking too low; based on polling that's come out today, as I noted earlier, it looks like we're in the middle of an utterly-unprecedented bounce of 12% or higher. If those numbers hold up the rest of the way, Romney wins, period. The question is, will they? As this bounce appears unprecedented, there is likewise no precedent to see if such a move is a permanent "closing the deal," or will fade over the next week (or month). We shall see.
boxman15
(1,033 posts)I'm seeing 4-5 points at the most, bringing us to a virtual tie nationally. Where does 12% come from?
regnaD kciN
(26,044 posts)...since they don't reveal the day-to-day numbers, the only change you can calculate is the "new day" over the "day that just dropped-off" figure. Since tracking polls are averaged over the number of days it runs, that means that the difference between the "new day" and the "old day" is, roughly, the change in the reported average times the number of days the poll runs.
In this case, the polls have been remarkably consistent.
Friday, the four-day Ipsos poll reported a three-point shift to Romney. 3 (the shift) x 4 (the number of days) = 12, which means that there was a 12% shift to Romney between the day that came on (Thursday-Friday) and the day that dropped off (Sunday-Monday).
Today, the three-day Rasmussen poll showed a four-point move to Romney. Once again, 3 x 4 = 12.
In today's Gallup poll, you saw a two-point swing to Romney. Since Gallup is a seven-day tracker, that means 2 x 7 = 14, or, roughly, a 14% swing to Romney.
I say "roughly" in each of these cases because of the rounding factor -- a 4% shift in the average could, theoretically, mean anywhere from 3.5% to 4.49%. That's information none of the polls divulge, but can probably be resolved by taking the result of several different polls. In this case, averaging all three, it would appear that the "new day"/"old day" difference for all three is slightly more than R+12.
At this point, we're only seeing 2-4 point gains for Romney in the topline numbers, but that's because, in most of the polls, half or more of the data points come from before the debate. As those days come off, if the numbers for the new days coming in match the day just past (and, once again, there's no guarantee they will), you'll see a 12-point or so swing to Romney in all three once all if the data points are post-debate. Given the state of the trackers before the debate, that would give Romney a 6-8 point lead in all of them...which would be an utterly unprecedented comeback for a trailing candidate in such a short time, and so close to an election.
IMPORTANT UPDATE: As I wrote that, Ipsos released its current results, which show no change in the topline numbers. This means, once again, that the Friday-Saturday polling results were pretty much equal to those in the day that dropped off (Monday-Tuesday). Now, we have no way of knowing what the latter numbers were; however, unless Monday-Tuesday was an exceptionally weak Obama polling day, the obvious conclusion is that Romney's big edge on Friday might be a transient "blip" rather than a long-lasting shift. (Or, alternately, that the positive job numbers offset the debate bounce.). In any event, time, and the other polls, will tell. The first big clue will be tomorrow's Rasmussen numbers. If Romney's bounce continues, he should increase his (currently 2%) lead substantially, to around 6% or higher. If he doesn't -- even if he maintains, but does not substantially increase, his current lead -- it will be another sign that this was just a short-term bump, which is itself likely to disappear as the immediate post-debate days, too, drop off the polls.
woolldog
(8,791 posts)is that the odds the site is giving understate Romney's strength right now. So basically instead of a 85% chance Obama only has an 75% chance (just an example).