A few hours after Monday night's presidential debate, I wrote, in
this space, "I don't think Obama won the debate by enough to reverse the momentum that Romney has had for most of the last few weeks, but he may have
finally stopped it." Now that post-debate tracking poll numbers are starting to roll in, it's possible to start testing that hypothesis. And, if I do say
so myself, it's holding up fairly well.
Yesterday the daily tracking polls reported the first results that included polling done after the debate, on Tuesday. Today New York Times
polling guru Nate Silver looked at those numbers for the eight tracking polls he follows and showed that, on balance, they had moved in Obama's favor since
the last pre-debate poll, on Monday. He wrote, "We can debate
whether Mr. Obama has a pinch of momentum or whether the race is instead flat, but it's improbable that Mr. Romney would have a day like this if he still
Since Silver wrote that this morning, another day's worth of numbers have come in, and if, based on those numbers, I had to choose between the "pinch of
momentum for Obama" scenario and the "flat race" scenario, I think I'd go with "flat race." Before explaining why, let's look at the numbers Silver was
working with. (The column labeled "Wed. Poll" refers to polls released on Wednesday but conducted on Tuesday.)
OK, now all eight of those polls have been updated to reflect polling done on Wednesday. I list the results below, but the upshot is this: Obama suffered a
little backsliding in Wednesday polling, but only a little. So Wednesday's numbers, like Tuesday's numbers, were better than the last pre-debate numbers,
if not by quite as much as Tuesday's numbers. Specifically: Whereas Tuesday's numbers yielded an average gain for Obama of 0.9 points relative to Monday's
numbers (as reflected in the lower right-hand number on Silver's chart), Wednesday's numbers gave Obama an average gain of 0.6 points relative to Monday's
So I'd say neither candidate now has any momentum to speak of. If the race stays that way--more or less flat--that's probably good for Obama. Though the
national polls on balance give Romney a fraction-of-a-point lead, Obama seems to be, as a practical matter, in the lead, for two reasons:
 Obama's numbers in swing states are running ahead of his numbers nationally. When the national polls were moving in Romney's direction, this gap may have
been partly due to the fact that, because swing states polls were being done less often than national polls, swing state polls were lagging indicators. But when, as now, national polls
are flat, and swing state polls are being conducted more and more often, that ceases to be a plausible explanation for the difference.
 The polls, especially in swing states, may underpredict Obama's election day numbers. These polls count only the responses of "likely voters"--a subset
of the "registered voters" the pollsters interview. Obama tends to do better with the latter than the former. And some people think that, because Obama's
"ground game" is better than Romney's, more Obama voters whom pollsters put in the "registered but not likely" category will wind up voting.
These two factors explain why, though Romney is slightly ahead in national "likely voter" polls, Obama is a clear favorite in the betting markets. As I
write this, Intrade gives him a 62 percent chance of winning.
Caveat: These are multi-day tracking polls, so each day's number is the average of several recent days' numbers (7 days in the case of Gallup, 3 in the case of
Rasmussen, etc.). And since the pollsters don't release each day's numbers individually, the running averages they do release are a bit opaque. If, say,
Obama's Gallup number goes up by 1 after polls done Oct. 24, you can only say two things for sure: (1) the Obama average for Oct. 18-24 is higher by one
point than the average for Oct. 17-23; and, therefore, (2) the Obama number of Oct. 24 is higher than the number for Oct. 17. The shorter the cycle the
tracking poll is on, the more precisely we can isolate the impact of specific events like debates. So, since Rasmussen is a three-day tracking poll, we can
say for sure that the Obama number two days after the debate was higher than the Obama number the day before the debate.
OK, for the record: Here are those numbers that have shown up since Silver posted this morning. These are from polling done on Wednesday and published on
Thursday, and the number for each poll represents the change relative to polling done on Monday--the last pre-debate polling. Gallup: Obama +2; IBD/TIPP: no
change; Ipsos/Reuters: Romney +2; PPP: Obama +3; RAND: Obama +2.1; Rasmussen: Obama +1; UPI/C-Voter: Obama +1; Washington Post/ABC: Romney +2.