It may be a bit fruitless to spend too much time worrying about the Wednesday afternoon FiveThirtyEight forecast when Wednesday night’s debate had the potential to change the election landscape. But for the sake of continuity, here goes.
The FiveThirtyEight forecast had Mr. Obama gaining slightly on Wednesday, estimating that he had a 86.1 percent chance of winning the Electoral College on Nov. 6 — up from 84.7 percent in Tuesday’s forecast.
This came despite the fact that it appeared there actually had been a modest shift back toward Mitt Romney in the polls even before the debate. In our “now-cast” — an estimate of what would happen if an election were held immediately — Mr. Obama’s projected margin of victory in the national popular vote had fallen by about one percentage point between Sunday and Wednesday.
Our Nov. 6 forecast, however, had already anticipated some decline for Mr. Obama, and so has been less sensitive to the shift.
In addition, there is a particular Electoral College outlook that is becoming problematic for Mr. Romney. As of Wednesday, our Nov. 6 forecast had Mr. Obama winning the popular vote by 4.1 percentage points. However, his advantage was larger than that — at least 4.9 percentage points, in 22 states (and the District of Columbia) — totaling 275 electoral votes:
I highlight New Hampshire in yellow on this map because, although it is one of the states where Mr. Obama’s lead now exceeds 4.9 percentage points, it is neither necessary nor sufficient for him to win the Electoral College votes in this configuration.
New Hampshire is not necessary because you could remove its 4 electoral votes from Mr. Obama’s column and he would still have 271, a winning total. It is not sufficient because if you removed any competitive state but New Hampshire from Mr. Obama’s column (for example, Nevada) he would at best achieve a 269-269 tie.
Really, a great deal of this comes down to Ohio. Historically, Ohio is about two percentage points more Republican-leaning than the country as a whole. This year, however, it has polled as being Democratic-leaning by one percentage point or so.
I ran an alternate version of our simulation on Wednesday in which Ohio was in fact polling two points more Republican than the country as a whole, as it has, on average, in the past, while leaving all other states unchanged. That change alone boosted Mr. Romney’s Electoral College winning chances to 19 percent from 14 percent.
The Predictive Value of Instant-Reaction Polls
In a way, however, it’s good news for Mr. Romney if his team needs to worry about the Electoral College. Why? Because this stuff only matters if the vote is very close; mathematically, it is extremely unlikely for a candidate to lose the Electoral College if he wins the national popular vote by more than about two percentage points.
Ohio is not a great problem for Mr. Romney to have — but it is a better one than being in a losing position all over the map. After the debate on Wednesday night in Denver, the chances are better that Mr. Romney will narrow his deficit in the national polls — and Mr. Obama will need his electoral firewall.
A bit of caution, however, about the predictive power of those polls showing Mr. Romney having clearly won the debate on Wednesday. As I mentioned after the debate, I had not come across a study on the relationship between instant-reaction debate polls and the eventual effect on the horse race polls. So I decided to do a quick one myself.
The chart below reflects the candidates that debate watchers deemed to be the winners or losers in CNN polls in 17 presidential debates between 1984 and 2008. I compare this figure with the change in national head-to-head polls before and after the debate.
(Estimates for the horse race change after the first debate in each cycle, and all debates in 1984, come from my own database; those for the rest of the debates are from Thomas Holbrook).
The estimates are framed from the perspective of the incumbent party candidate, who is designated with the letter ‘I’ and the color purple in the chart; the challenger is designated with ‘C’ and the color orange.
The clearest win for any candidate in the CNN poll came in 1992, when Bill Clinton was deemed the winner of the second presidential debate by 58 percent of respondents, as compared with 16 percent for President George H.W. Bush and 15 percent for H. Ross Perot. That town-hall-style debate, in Richmond, Va., included this famous moment in which Mr. Clinton displayed his capacity to empathize with voters:
Indeed, Mr. Clinton increased his margin over Mr. Bush by about four percentage points in head-to-head polls immediately after that debate. Although Mr. Clinton lost some of that advantage during the remainder of the campaign, this is still a favorable precedent for Mr. Romney, since his 38-point margin of victory in CNN’s poll on Wednesday night is similar to Mr. Clinton’s 42-point win in that 1992 debate.
In several other cases, however, the instant-reaction polls did not correlate with the change in head-to-head polls. Before Wednesday night, the second-clearest margin of victory for a challenging candidate in the CNN poll came in 2008, when Mr. Obama was declared the winner of the third presidential debate by a 27-point margin. However, his opponent John McCain actually gained slightly in the polls instead just after that debate.
In 2000, CNN instant polls deemed Al Gore the winner of both the first and third presidential debates. But he lost about three points to George W. Bush in head-to-head polls after each of them. In 1996, Mr. Clinton was declared the winner of the first presidential debate in the CNN poll, but Bob Dole gained slightly in the head-to-head polls after that.
Over all, the relationship between the winner of the instant-reaction poll and the change in head-to-head polls is positive, although not statistically significant.
But for what it’s worth, the historical data would project a gain of 2.2 percentage points for Mr. Romney in the head-to-head polls by this time next week.