In national polls, the race has swung back three points since the Presidential debate to a narrow Obama lead. This return has been steady over time, and so the role of the VP debate is unclear. Combined with state polls, the data suggest that the effect of Mitt Romney’s performance was an instantaneous jump of 5.5 points, which has now subsided back to where polls were in August. The decline in the state poll meta-analysis has been blocked by Ohio. Today, President Obama’s November re-elect probability is 84% – still a Russian-roulette situation for the Democrats.
Andrew Sullivan is staring so hard into what he thinks is an abyss that he has not noticed actual conditions.
Biden does seem to have reversed the speed of Obama’s free-fall but not the decline itself….It’s not going away by itself. That is: not a bounce.
It is hard to prevent desires/fears from coloring one’s reading of data. Pundits and journalists are often expected to comment even when there’s not enough information to do so. I would rather they pull a Ludwig.
His pronouncement about a “resilient jump” is ironic. In four national polls with mostly post-VP-debate data, we have a median of Obama 2.0 +/- 1.3% (Oct. 10-14). Compare that with the Era Between The Debates, Obamney 0.0 +/- 0.9% (Oct. 6-10, n=4) and the days immediately following the debate, Romney +1.0 +/- 0.9% (Oct. 4-7, 4 polls). So this recovery started before the VP debate. Like I said, Biden’s the rooster, not the sunrise.
The Presidential debate bounce = 5.5 points. Within the span of a single day, Romney’s debate performance got him a large gain that was apparent immediately in national polls and some state polls. The most exact measure available is based on the Meta-Analysis, which measures the true margin viewed through the Electoral College. The estimated swing is precise to within <0.5%.
The Meta-Analysis gave the appearance of a steady slide. However, state polls take a while to trickle in, creating a false impression. Just like the Ryan-VP bounce and the Palin-VP bounce, it took about 12 days to see the whole change unfold. Because of its slowness, it probably missed a brief period around Oct. 4-6 of Romney being ahead.
One hint of the suddenness of the Romney debate bounce is how the EV estimator has moved in clear steps corresponding to states falling like dominoes: VA, then CO, then FL, then a little NH domino at the end.
Enthusiasm has been a major factor in this change: Romney voters getting more enthusiastic, and Obama voters becoming less enthusiastic. Not many people got off the fence: post-debate, undecided voters were at 6.0 +/- 0.6% (n=18), compared with pre-debate 5.0 +/- 0.8% (n=21). A few voters have changed their minds.
The “debounce.” In national polls, some return has been apparent over a period of 7 days. Tentatively, it looks like it took one week for opinion to come back by 3 points, i.e. about halfway. This week we’ll get a better measure as we watch national polls. State polls will help, though more gradually.
Where things are, and what to look for in the weeks ahead. I expect high feelings among Romney voters will fade, as well as low feelings among Obama voters – though in both cases, not all the way. I view recent events as the equivalent of a major market correction. Do not be surprised to see the Meta-Margin reverse direction this week.
Here is what the automated calculation is showing for today – and the future. I emphasize automated – I follow its output, just as you do.
The red zone is like a hurricane strike zone. It corresponds to a 1 standard deviation interval, or “1 sigma.” Outcomes should fall in that range about 2/3 of the time (68% to be exact), above and below 1/6 of the time each. Today the strike zone’s edge is almost exactly at 269 EV. You can figure out very easily the re-elect probability for President Obama: 84%.
The yellow zone includes pollster-to-pollster variability plus “2 sigma” of drift. If the race goes out of that range, I’ll drink a bottle of Tabasco. (I’m in New Orleans at the moment.)
Two weeks ago, the calculation dictated that Obama would fall from a high of +6.2%. Today, it predicts the converse. Let me review the contributing facts.
- The President’s been ahead all season. I said the Meta-Margin should float around Obama +3.0 +/- 2.2%. This means that as of today, he’s in a place I said he’d be about 14% of the time.
- We now know that the debate spike was large – but not long-lasting.
- In the 2004 and 2008 races, the race’s snapshot on any day was not predictive of where it would be 3 weeks later. Closer than 3 weeks, it gets slowly more predictive. The movement from today to Election Day should be about 1.8% (1 standard deviation), perhaps more in extreme cases.
- We don’t know which way things will move next. However, the people surveyed in June-September are the same people being surveyed in October. Movement from an extreme is likely to be toward the equilibrium. Think of the Presidential race as a swinging pendulum.
Of course, there is a chance that this will not happen, and Mitt Romney will win the Presidency. It’s like hurricanes: storms go to unexpected places.
As I said, I think Romney’s win probability is about 16%. To a Democrat, that’s a six-shooter with one shell labeled “R-outcome”. To a Republican, it’s loaded with five shell labeled “D-outcome”. Yet in comments, the Republicans are the giddy ones. This says so much to me about both sides.
Thanks to Michale Fee for the idea for the domino graph.