Sept. 27: The Impact of the ‘47 Percent’
By NATE SILVER“Ninety percent of ‘game-changing’ gaffes are less important in retrospect than they seem in the moment,” I wrote.
But was this one of the exceptional cases? A week and a half has passed since Mr. Romney’s remarks became known to the public — meaning that there’s been enough time to evaluate their effect on the polls.There’s a case to be made that they did damage Mr. Romney’s standing some.
In the chart below, I’ve tracked the progress of the national popular vote in the FiveThirtyEight “now-cast” over the past five weeks. The “now-cast” reflects our best estimate of what would happen in an election held today, based on a combination of recent national and state polls. Unlike our Nov. 6 forecast, the “now-cast” does not account for economic measures, and it does not adjust for the effect of the party conventions. This makes it a little bit more straightforward to interpret in terms of tracking the progress of the polls in real time.
In the chart, I’ve highlighted the dates of what are probably the four most important political news events of the last month: the Republican and Democratic conventions; the deaths of four Americans in an attack on the consulate in Benghazi, Libya; and the release of the “47 percent” tape.
The first of these events, the Republican National Convention, did not produce much in the way of a discernible change in the “now-cast.” My view is that Mr. Romney probably did receive a bounce in his polls, but it was small and short-lived, since the Democratic convention began only a few days after the Republican one ended. The “now-cast” is trained not to overreact to modest changes in the polls, and so it had trouble distinguishing any convention bounce for Mr. Romney from statistical noise.
The Democratic convention, however, did produce a clear (if quite middling by historical standards) bounce in the polls for President Obama. He went from being 1.4 percentage points ahead in the “now-cast” popular vote when his convention began, to four points ahead a week after it ended, once there had been time for it to work its way through the polls.
It’s not clear whether the Libya attacks had any impact on the polls, despite the news media judging Mr. Romney’s reaction to them very harshly (while spending less time scrutinizing a potential security lapse on the part of Mr. Obama’s administration).
By Sept. 17, the date when the video of Mr. Romney’s remarks was released and received widespread attention, the momentum from Mr. Obama’s convention appeared to have stalled (although not necessarily reversed itself). Mr. Obama led in the popular vote by 4.1 percentage points on that date, according to the “now-cast.”
Since then, however, Mr. Obama has gained further ground in the polls. As of Thursday, he led in the popular vote by 5.7 percentage points in the “now-cast,” a gain of 1.6 percentage points since Mr. Romney’s remarks became known to the public.
It’s hard to tell whether this recent gain for Mr. Obama reflects the effect of the “47 percent” comments specifically. But the most typical pattern after a party convention is that a candidate who gains ground in the polls cedes at least some of it back.
Instead, the more pertinent question seems not whether Mr. Obama is losing ground, but whether he is still gaining it.
Thursday’s Polls
What we can say with more confidence is that Mr. Romney is now in a rather poor position in the polls. In three of the four national tracking surveys published on Thursday, Mr. Romney trailed by margins of six, seven and eight percentage points. He also trailed by five percentage points in a one-off survey published by Fox News. The exception was Thursday’s Rasmussen Reports tracking poll, which showed the race in an exact tie, although that was improvement for Mr. Obama from a two-point deficit on Wednesday.
The state polling published on Thursday was more of a mixed bag. Mr. Obama led by seven points in an NBC News-Marist College survey of New Hampshire, a strong but not extraordinary result for him. He also led by two points in a Marist poll of North Carolina, continuing a streak of stronger polling for him in that state. For the first time all year, Mr. Obama is listed as the favorite in North Carolina in the “now-cast” — although he still trails slightly in the Nov. 6 forecast, which expects his numbers to decline so
me between now and Election Day.
However, Mr. Obama got middling results in a Suffolk University poll of Virginia, which put him ahead by 2 points, and in the Marist poll of Nevada, which also had him up by 2. Perhaps it’s damning Mr. Romney with faint praise to describe swing-state polls in which he trailed as constituting “good” news for him — but these surveys were a little bit better for Mr. Romney than other swing-state polls in recent days.
There were also a series of partisan-tinged polls released on Thursday: in Mr. Romney’s case, a poll by Voter Consumer Research in Iowa for the Web site The Iowa Republican, which showed him leading by one point in that state; and for Mr. Obama, a series of polls conducted by Public Policy Polling on behalf of the NRDC Action Fund in Ohio and other swing states.
We have taken a rather inclusive attitude toward which polls are included in the forecast this year — excluding only those conducted directly on behalf of the campaigns, or by “super PACs” very closely associated with them like Priorities USA Action. The philosophy here is that persistent bias in these polls will be corrected for by our “house effects” adjustment, and that there is little merit in making overly fine distinctions about which polls qualify as partisan and which don’t. There are nominally nonpartisan polls that have strong house effects — and arguably partisan ones that normally play it pretty straight. Nonetheless, I wouldn’t recommend that you rely on these borderline cases to tell you much about the momentum in the race.
The overall story line, however, is fairly clear: Mr. Romney is at best holding ground in the polls, and quite possibly losing some, at a time when he needs to be gaining it instead. Further, it’s increasingly implausible for Mr. Romney to attribute the numbers to temporary effects from the Democratic convention. Mr. Obama’s probability of winning the Electoral College advanced to 83.9 percent in the Nov. 6 forecast, up from 81.9 percent on Wednesday.
This post has been revised to reflect the following correction:
Correction: September 28, 2012
An earlier version of this post incorrectly referred to the organization on whose behalf Public Policy Polling conducted a series of polls in Ohio and other swing states. It was the Natural Resources Defense Council, not the National Resource Defense Council.
This post has been revised to reflect the following correction:
Correction: September 28, 2012
Earlier versions of this post incorrectly referred to the organization on whose behalf Public Policy Polling conducted a series of polls in Ohio and other swing states. It was the NRDC Action Fund, not the Natural Resources Defense Council or the National Resource Defense Council.
No comments:
Post a Comment