Posted by 
With Democrats so enthused about Vice President Kamala Harris’s campaign of late, you might have gotten the impression that she has a solid polling advantage over former President Donald Trump.
That’s not the case. Harris currently leads national polls by about 3 percentage points on average, but the Electoral College is what actually determines the outcome. That means to win the presidency, Harris and Trump need to pick up as many swing states as possible. And as of midday Tuesday, state polling averages from publications like Silver Bulletin, Real Clear Politics, FiveThirtyEight, and the New York Times point to hair-raisingly tight contests in the most important swing states.
Wisconsin and Michigan are currently looking best for Harris: All four polling averages mentioned above agree she is narrowly ahead there, by about 1 to 3 percentage points. But those states would not be enough to give Harris an Electoral College majority.
The state of play is murkier in Pennsylvania, Georgia, Nevada, and Arizona, where the averages vary. Though some averages of these states show Harris very slightly ahead (by about 1 percentage point or less) in each of those, others show tied races (in Pennsylvania and Nevada) or very narrow Trump leads (in Arizona and Georgia).
In the remaining swing state, North Carolina, Trump leads polling averages by a very tiny margin of less than 1 percentage point.
The only reasonable takeaway from polling like that is: the contest looks very close and neither candidate has gained a clear advantage.
The Electoral College nitty-gritty
The margins in all these swing states are extremely tight. Still, as one scenario for how election night could play out, let’s assume that, as the averages currently suggest, Harris wins Wisconsin, Michigan and the other lean-blue states; Trump wins North Carolina and the lean-red states; and Pennsylvania, Georgia, Arizona, and Nevada are too close to call.
This would place Harris at 251 electoral votes — 19 away from the 270 she needs to win. And she’d have two paths to get over the threshold.
Path one would be to win Pennsylvania. Its 19 electoral college votes would be just exactly enough. (That’s why it’s the most important swing state.)
Path two would be to win Georgia plus either Arizona or Nevada. Winning only the combo of Arizona and Nevada would not be enough, as that would provide Harris with 17 electoral votes. She’d need Georgia too.
That’s why it matters that all of these states — Pennsylvania, Georgia, Arizona, and Nevada — look so close. Tiny shifts in them could well determine the outcome. And neither candidate looks to lead comfortably in enough of them so far.
Is Harris’s lack of a big convention bounce understandable or ominous?
Polling averages tell you what the polling currently shows, but election forecasts, in theory, are supposed to do something extra: to anticipate via modeling how the race may change between now and Election Day.
That has proven contentious in an election year that has, in the past few months, veered off into uncharted territory with Democrats’ late Biden-for-Harris swap.
FiveThirtyEight — under new management since founder Nate Silver departed last year — came under much criticism earlier this summer, first due to its forecast model’s optimistic projections about Biden’s chances despite his bad polling, and second because the model remained offline for a month after Biden quit.
G. Elliott Morris, who now runs FiveThirtyEight for ABC News, republished the model last week, acknowledging he had to make some significant changes to it. The old version of the model gave Biden a significant advantage for being the incumbent president and seemed to give that greater weight than new polling showing Biden struggling. Critics questioned whether it truly made sense to assume the unpopular, elderly president would get an “incumbent advantage” toward the end of the race.
Silver is now publishing his own independent forecast at Silver Bulletin, and for most of August, the model aligned with polls, showing Harris as a slight favorite.
But beginning last Thursday, Silver Bulletin showed Trump taking a slight edge. As of midday Tuesday, it gave Trump a 56.7 percent chance to win the election. That’s despite Silver Bulletin’s polling averages showing Harris with narrow edges in most swing states.
Again, a forecasting model is not just a crude polling average, it’s also trying to anticipate how the race might change between now and the election. As Silver explained in a recent post, the model is dinging Harris in part because presidential nominees typically get a polling bounce after their convention and lose some of that advantage later on.
Harris hasn’t gotten much of a bounce since the Democratic convention and she’s barely leading now, so if we assume she’s likely to lose a bit of ground later, that means Trump would have the edge.
But again, is that a safe assumption? Harris optimists could make the case that she already got her “convention bounce” with the enormous press and public attention on her since Biden dropped out and she emerged as Biden’s replacement, so we shouldn’t necessarily have expected her polling to improve further after the convention.
Harris pessimists could turn that logic around, saying that yes, her campaign has been in a “honeymoon” phase since she entered the race, but that the honeymoon is bound to end and she’s sure to experience more contentious periods of the race going forward.
In any event, the next big moment that can reshape the election — and the polls — is coming soon: the Trump-Harris debate is on September 10, just one week away. And no one has a crystal ball for how that will go.
