We can talk about polling bounces all we want, but what we should be seeing in today’s polls is a bit of distortion in favor of Romney because he just held a three-day convention and the Democrats have not yet responded with their own convention. In other words, we should expect Romney’s current polling numbers to be artificially inflated.
We do see that Romney got a little bounce out of his convention which is reflected in both state and national polls, and which also is seen in measures of approval and likability. But if you look at other metrics, like 538.com’s Electoral College and popular vote projections, or Romney’s chance of winning (see sidebar on right), his numbers have been going in the wrong direction since roughly the moment the Republicans convened in Tampa. In fact, Romney is currently in his weakest position under those metrics since at least the beginning of June.
The highest plot on the Electoral College distribution probability chart is now north of 350 votes. Part of this is because Nate Silver takes account of economic data, and there has been some modestly good news lately on that front. Yet, for whatever the reasons, Obama’s numbers are trending up in the aftermath of the Republican convention.
Let’s take a look at where Romney stood on August 27th, when the RNC convened and where he stands eight days later.
Chance of Winning
August 27: Obama 69.3% Romney 30.7%
September 4: Obama 74.8% Romney 25.2%
Electoral College Projection
August 27: Obama 299 Romney 239
September 4: Obama 308.2 Romney 229.8
Popular Vote
August 27: Obama 50.6% Romney 48.2%
September 4: Obama 51.1% Romney 47.8%
Obviously, Nate Silver isn’t infallible. But his model is pretty robust. I didn’t expect Romney to lose ground in those key metrics after his convention. I expected the opposite. And I’m an optimist.
It’s the little things that sometimes begin to count the most:
From RandomActsofChaos
Republicans: Slavery was 150 years ago, black people need to get over it. It’s not like I ever owned slaves.
Republicans: I know what it’s like to be an immigrant from the hardships my ancestors faced 150 years ago.
Even for my brother, it’s getting a little thick. His only defense now is “everyone of them is a liar. Romney’s just better at it”. I’m not even sure my brother believes the better at it part anymore.
My brother truly believes that KS Gov. Brownback is a center left governor.
Boo, I think it’s important to note that Nate’s model does take into account an expected convention bump of about 4% for Romney right now. So because it is turning out to be more in the 1-3% range it is effectively giving Obama that extra 1-3% in the vote which is what is driving up Obama’s numbers.
If Obama also has a weak convention we would expect the chance to win to drift back towards that steady 69%-70% for Obama (if you look at the “now cast” you see that the bump is treated as real, and Romney is at his strongest to date) we’ve seen for a while now. I don’t think things will go bad, I expect it to stay near 75% and maybe even improve.
I get that impression too, but I’m a little concerned by the apparently growing view that Nate’s projections are the be-all and end-all of electoral projections (I like to head over to Princeton Election Consortium occasionally so that I can get a more basic view of the race). The fact is that in addition to the hard numbers, he does add a fair amount of his own “secret sauce” in crafting his projections, and that can of course inject human error into the analysis. I don’t have any particular reason to question his methodology, but there is always the possibility that other factors either inside or outside of his model will play a larger or smaller role on the election than he anticipates.
You’re certainly correct (and as a Twins fan the early 2000s Twins occasionally defeated the SABR predictions) on accuracy but at this point I have to seriously question whether his “model” actually leaves ANYTHING out.
What he leaves out is that anything can happen. Some campaign slogan could catch fire for no apparent reason. Somebody could say something stupid or wrong enough to discourage turnout for their side. Economic numbers could turn good or bad (and they were bad today). There could be some natural or unnatural disaster or some expose of somebody’s malfeasance.
I really dislike efforts like Silver’s and his ilk. They may not intend to embed a pattern in voters’ minds, but that’s what they end up doing. They draw attention from issues that need to be addressed to obsessing on tea-leaf reading. I think we’d all be better off without them.
I think that if we didn’t have them it would be much easier for people like Fox to drive the narrative and for election theft to happen. I don’t think the removal of efforts like Silver’s would do anything to focus attention on the issues. Quite the contrary actually, as that lack of something resembling an outside check on what moves the numbers would make it that much easier for the people with the money to pretend the issues that they want to spend money on are the important one. On a more personal note, having that data to look at helps me to ignore the wurlitzer and keep from falling into despair over the state of our elections.
I think this is badly mistaken. For one thing, economic numbers are taken into account. He might be wrong, but he’s not missing much that we can possibly know at any given moment. Of course “anything” can happen, but that doesn’t meant it will.
Whether issues need to be addressed or not, there’s only so much you can write about how the GOP proposes to end medicare for instance.
The danger in reading Nate Silver’s projections is reading probabilities as the percent of the popular vote. Even if you don’t make that mistake explicitly, the thought colors how you interpret the data.
A 72% probability of an Obama win means that there is a 2 in 7 chance of Obama losing and Nate’s projection of victory of being wrong. That’s a pretty wide margin for random events to intervene.
I haven’t been watching Nate closely this time around, but I remember that from last time. He uses a technique of running probability scenarios which are great for identifying the most likely possible outcome but tend to overstate the possibility of unlikely events, and understate the possibility of likely events.
For example, in the week before the 2008 election he gave McCain something like 4 to 6% chance of winning. That was wrong – at that point, with the date that was known, he had less than 0.000004% chance of winning. However, 4% of Nate’s probability scenarios showed that result, so that’s what he reported.
Nate’s always learning and improving. For example, in 2008 he created, in mid-August, a model that predicted the polls for the remainder of the time leading up to the election based on the factors known at that time, plus generic factors based on history for convention bounces and the impact of debates. His projection almost perfectly matched what actually happened (and this was before 99.99% of us had heard of Sarah Palin and a month before the stock market collapsed – events that many have attributed to the shifts in the electoral polls). I see that this time around he’s built that historical data into his model – assuming convention bounces and accounting for them.
Actually, keep in mind that Romney hasn’t been able to touch his massive fundraising warchest until he formally accepted the Republican nomination Thursday night. That means a massive blitz of new, coordinated TV ads, as well as a huge wave of new field offices in key states and massive organizing efforts on his part.
So I actually think this is Obama’s high water mark. It’ll dip back down to the high 60’s after the convention due to Nate’s adjustments, but then I would expect it to slowly erode over the next couple of months to the point that it’s near 50-50 come election day.
Will there be a wave of new field offices and organizing? Romney has been an inept campaigner so far, and he’s a top-down kinda guy. I’m guessing that attitude will bite him in the ass.
“There is some noise in this data: Mr. Romney gets everything from a seven-point bounce (in a Public Policy Polling survey of Michigan) to what technically is an ‘anti-bounce’ …”
More remarkable is the advantage of Obama in the swing states. Romney did not impress the independent voter nor the working class people in states like Indiana, Ohio, Michigan and Pennsylvania. Isn’t that the reason the popular vote is close but the Electoral College gives Obama a good lead. It’s his to lose. The loonies aren’t helping Romney, he is desperate and there seems to be less cohesion in the Republican message.
Factchecker: Romney’s ’12 million jobs’ promise and creating jobs after the Great Recession of 2007-2009.
Optimist isn’t the word I’d use.
Anyhow, even if Obama is doing somewhat well a look at the polling shows the Senate becoming razor thing once again. $10 million might not change the presidential race, but we can’t compete with that kind of money in the Congressional racers.
Why the Obama campaign is cautiously confident in victory