We make an effort to be disciplined about the way that we characterize the polls each day. Generally speaking, if the FiveThirtyEight forecast moves appreciably toward President Obama, the tenor of the article is going to be favorable to him - and likewise for Mitt Romney, if the same is true for him.
If the forecast remains unchanged, we won't try to emphasize the smallest ticks of movement.
Sometimes, though, a day's worth of polling falls into the awkward middle ground between being a good day for one of the candidates and revealing little new information about the state of the campaign.
Friday was one such case: a reasonably strong day in the polls for Mr. Obama, but nothing that ought to alter our impressions of the race.
Mr. Obama's chances of winning the presidential election are listed at 76.9 percent by the forecast model, an incremental improvement from 76.1 percent on Thursday.
The trend over the last three days is clearer: Mr. Obama's forecast is up from a 72.9 percent chance of winning the Electoral College on Tuesday. However, he remains off his highest point in the forecast early last week, when he topped out at 80.8 percent.
Emblematic of Mr. Obama's good-but-not-great polling day were a set of polls from the firm Purple Strategies, which had him ahead in four of the five swing states: Ohio, Virginia, Colorado and North Carolina. However, Mr. Obama still trailed Mr. Romney by one point in Florida, according to the poll.
Purple Strategies had polled all of these states but North Carolina previously, and Mr. Obama's standing improved on average by three percentage points from the polls they condu cted in August.
This is consistent with the post-convention bounce that we've seen for Mr. Obama on the whole. The FiveThirtyEight “now-cast†estimates that if an election were held today, Mr. Obama would have a 95 percent chance of winning it. Additionally, he is projected to win the national popular vote by almost five points â€" up from about two points before the conventions. The three-point gain is the same as in the average Purple Strategies poll.
Our Nov. 6 forecast continues to be more conservative, however, as we still need to account for the possibility that Mr. Obama's numbers are inflated by the aftereffects of his party's convention. By this time next week, it will be safer to conclude that Mr. Obama's gains are permanent, and the forecast will move toward Mr. Obama if Mr. Romney does not make some tangible improvement.
The Purple Strategies polls, which were conducted by automated script, are also consistent with the recent data in another way. While they are decent for Mr. Obama, they are not as good for him as polls that use live interviewers and call cellphones along with landlines, which have had him ahead by about six points on average in swing states.
As the overall race has changed, however, one thing largely hasn't: the states that look as though they'll be the most important to the outcome.
The table below represents a comparison between the forecast that we ran on Aug. 28 (I chose this date because it was the last one before any polls would have been affected by the Republican convention) and the one on Friday. It shows Mr. Obama's projected margin of victory or defeat, now and then, in the 11 swing states that the campaigns and the pollsters have concentrated most of their efforts on.
Mr. Obama's projected margin of victory in the national popular vote has improved by 0.9 percentage points since Aug. 28, from 2.4 points to 3.3 points. (Note, again, that this reflects the expectati on that Mr. Obama's polls will probably recede some between now and Nov. 6; he'd be a favorite to win by more than 3.3 points in an election held today.)
In most of the swing states, Mr. Obama's forecast has improved by almost exactly that margin. His projected margin of victory in Ohio and Virginia has improved by 1.0 percentage points since that time and by 0.8 percentage points in Florida. His numbers have improved by 1.1 percentage points in Iowa and by 0.6 percent in Pennsylvania.
There are two states where Mr. Obama's forecast has actually declined: New Hampshire and North Carolina. But the decline has been modest, and might not easily be differentiated from statistical noise. New Hampshire, in particular, has not been polled as much as the other states, especially by the polls that use industry-standard methodology. It's easy enough to imagine a CNN or NBC poll coming out next week that puts Mr. Obama seven or eight points ahead there, which might be enoug h to reverse out the trend.
Mr. Obama's numbers have also been slightly weaker in Nevada and Colorado than in other swing states recently. But his outlook has still slightly improved in those states since August, and the change has probably not been enough to appreciably impact campaign strategy.
Mr. Obama has technically made his largest gains in Michigan, where his forecast improved by 1.7 percentage points, and in Wisconsin, where it has grown by 1.6 points. Still, these shifts are only slightly different from the overall national trend.
Mr. Romney, in my view, must continue to contest Wisconsin. It is ranked fourth in the country on our tipping point list, which ranks the states that are most likely to provide the decisive votes in the Electoral College in a close election.
If there's one state where the latest polls might merit a substantive change in campaign strategy, it could be Michigan. That's because Michigan was only marginally competitive before, and with the latest polls having shown strong numbers for Mr. Obama there, it now seems extremely unlikely to be a part of Mr. Romney's strongest comeback plan. The forecast estimates that Mr. Obama has a 96 percent chance of winning Michigan on Nov. 6.
A similar argument might be made about North Carolina, another state whose importance to the electoral math we have questioned in the past.
The numbers in North Carolina polls have been a bit mixed, and Mr. Obama could certainly win it. But he has little reason to bother with it when his bounce has manifested itself more clearly in most of the other swing states, and when those states sufficiently provide a myriad number of ways for him to reach 270 electoral votes.
But none of these Electoral College calculations will matter unless Mr. Romney can improve his overall standing in the race. If the national popular vote is within one or two percentage points, a split outcome is possible, but it is exce ptionally unlikely otherwise.
For the time being, instead, most of the value in swing state polls is that they provide useful information about the state of the national campaign. In general, the gains that Mr. Obama has made in the state polls have been quite consistent, both with one another and with the national trend.
The FiveThirtyEight model uses state polls alongside national ones to calculate the current trend in the race. Doing so often lends more stability to the forecast than using national polls alone.
We have added nearly 100 state polls to our database over the course of the past week. If you can't infer some useful information from those, and instead insist on calibrating your expectations of the race from one or two tracking polls, you're going to have a warped perspective on where the contest stands.
No comments:
Post a Comment