Sept. 8: Conventions May Put Obama in Front-Runner's Position
fivethirtyeight.blogs.nytimes.com/2012/09/08/sept...entions-may-put-obama-in-front-runners-position/
nytimes The New York Times
FiveThirtyEight: Sept. 8: Conventions May PutObama in Front-Runner's Positionhttp://t.co/W3P26Jk4Sept. 8: Conventions May Put Obama in Front-Runner’s Position
By NATE SILVEROn Friday, we began to see reasonably clear signs that President Obama would receive some kind of bounce in the polls from the Democratic convention.
Mr. Obama had another strong day in the polls on Saturday, making further gains in each of four national tracking polls. The question now is not whether Mr. Obama will get a bounce in the polls, but how substantial it will be.
Some of the data, in fact, suggests that the conventions may have changed the composition of the race, making Mr. Obama a reasonably clear favorite as we enter the stretch run of the campaign.
On Saturday, Mr. Obama extended his advantage to three points from two points in the Gallup national tracking poll, and to four points from two in an online survey conducted by Ipsos. He pulled ahead of Mitt Romney by two points in the Rasmussen Reports tracking poll, reversing a one-point deficit in the edition of the poll published on Friday.
A fourth tracking poll, conducted online by the RAND Corporation’s American Life Panel, had Mr. Obama three percentage points ahead of Mr. Romney in the survey it published early Saturday morning; the candidates had been virtually tied in the poll on Friday. (The RAND survey has an interesting methodology — we’ll explore it more in a separate post.)
The gains that Mr. Obama has made in these tracking polls over the past 48 hours already appear to match or exceed the ones that Mr. Romney made after his convention. The odds, however, are that Mr. Obama has some further room to grow.
The reason is that the tracking polls are not turned around instantaneously. The Gallup poll, for instance, now consists of interviews conducted between Saturday, Sept. 1, and Friday, Sept. 7. That means that many of the interviews in the poll still predate the effective start of the Democratic convention on Tuesday night.
That Mr. Obama has made these gains in polls that only partially reflect the Democratic convention suggests that his bounce could be more substantial once they fully do so. Mathematically, Mr. Obama has to have been running well ahead of Mr. Romney in the most recent interviews in these surveys to have made up for middling data earlier in the week.
In fact, it is possible to reverse-engineer an estimate of what Mr. Obama’s numbers look like in the postconvention part of the tracking surveys. Specifically, I will be looking to infer Mr. Obama’s numbers from interviews conducted after Bill Clinton’s speech on Wednesday night, which in my view was the pivotal moment of the convention.
Let’s use the Gallup tracking poll as an example. Mr. Obama now leads in that survey by four percentage points. Conversely, he led by one point in the version of the poll published on Wednesday afternoon, ahead of Mr. Clinton’s speech. What must Mr. Obama’s numbers have looked like in the interviews since the Clinton speech in order for him to make those gains?
This can be determined with a little algebra if we know what percentage of the interviews in the Gallup survey reflect post-Clinton data. Fortunately, this calculation is fairly straightforward.
Gallup’s tracking poll is reported over a seven-day window, and roughly the same number of people are polled each day. The interviews Gallup conducted on Saturday, Sept. 1, and then on Sunday, Monday, and Tuesday, predated the Clinton speech. The interviews it conducted on Thursday and Friday post-dated it. The only day about which there is any ambiguity is Wednesday itself. But, since Mr. Clinton’s speech was made late Wednesday night, only a small fraction of the respondents in the poll would have had the chance to watch it by the time that Gallup called them — probably just the last round or so of interviews that Gallup conducted on the West Coast. (We’ll assume that 20 percent of the Wednesday interviews did reflect Mr. Clinton’s speech, although the fraction was probably a little lower than that in practice.)
Over all, that means that only about 30 percent of the data from the Gallup poll post-dated Mr. Clinton’s remarks.
If you do the math, it implies that Mr. Obama must have been leading Mr. Romney by 10 or 11 points in the minority of the poll conducted since Mr. Clinton’s speech for him to have gained three points in the survey over all.
In the table below, I’ve run through the same calculation for the other tracking polls. The results imply that Mr. Obama has run about nine points ahead of Mr. Romney in the portion of the Ipsos poll conducted since Mr. Clinton’s speech, about eight points ahead in the RAND poll, and about four points ahead in the Rasmussen poll.
On average between the four polls, it appears that Mr. Obama must have held about an eight-point lead since Mr. Clinton’s speech in order to have gained so much ground so quickly.
This method is not perfect — the only way we would know exactly how well Mr. Obama had been doing is if the polling firms published day-by-day results, which none of them do.
