Harris and Trump are tied in 538's new polling averages
Good news, polling fans: 538 now has polling averages for the new presidential matchup between Vice President Kamala Harris and former President Donald Trump. As of Friday at 10 a.m. Eastern, our average of national polls says Harris has the support of 45.0 percent of voters, while Trump garners 43.5 percent.
That 1.5-percentage-point lead is within our average's uncertainty interval, which you can think of as a sort of margin of error for our polling averages. We also have new Harris versus Trump polling averages for the key swing states of Georgia, Michigan and Pennsylvania, and the races there are within the uncertainty interval as well. (We will add polling averages for additional states as soon as we get enough polls in them.*)
538's averages are slightly better for Harris than the ones you'll find from other public aggregators. There are a few reasons for this. The most important is that our average only considers polls that began interviewing voters after President Joe Biden dropped out of the presidential election and endorsed Harris to be the Democratic nominee. (It is long-standing 538 policy not to include horse-race polls with candidates that aren't actually running in our models.) Because those earlier polls were asking about a then-hypothetical matchup, voters may have been thinking about the candidates in theoretical ways that are no longer relevant to today's very real contest. Indeed, by our estimate, Harris has performed about 3 points better nationally in polls conducted after Biden's withdrawal than before it.
Another reason our averages are better for Harris is that 538 adjusts polls that were conducted among registered voters (marked as "RV" on our polls page) or all adults ("A") toward the results of likely voter ("LV") polls. We make this adjustment because we know for a fact that not all adults or even all registered voters will vote in the election, and the types of people who vote differ from nonvoters in predictable ways.
Specifically, there is evidence that likely voters are now a bit more Democratic than the general population; for example, likely voter polls of Trump versus Biden tended to look better for Democrats than polls of that matchup among registered voters or all adults. Our average makes this adjustment by looking for systematic differences between likely voter and registered voter/all adult polls after controlling for other factors, such as the pollster who conducted the poll, the mode it was conducted with, the time it was conducted, etc.
Without this adjustment, Harris and Trump would be tied at 44.0 percent nationally (based on polls conducted since July 22 and released by Aug. 1). But Harris's margin is about 0.2 points higher in likely voter polls than in registered voter polls and about 0.6 points higher in likely voter polls than in polls of all adults. Controlling for this alone increases our estimate of Harris's margin by 0.5 points.
538's polling averages also look for differences between polls that do and do not give voters the ability to say they support third-party candidates, such as independent Robert F. Kennedy Jr. (who currently averages 5.7 percent support nationally). We make this adjustment because most voters will have more options than just Harris and Trump on Election Day (depending on which candidates qualify for the ballot in which states), so head-to-head polls are not asking about a real election either. We find that, when Kennedy is included in polls, Harris's margin over Trump grows by about 0.5 points.
Finally, Harris gets another boost in our national average because of how well she is doing in the few state polls that have been conducted since Biden's withdrawal. On average, in 2020, Biden beat Trump by 3.8 points in the states that have been polled for Harris versus Trump so far. Her average polling lead in those states today is 1.1 points. If you apply that 2.7-point shift toward Trump to the national popular vote (which Biden won by 4.5 points in 2020), you get a predicted national environment of Harris+1.8 today.
All in all, Harris is polling better against Trump than Biden, who was trailing by more than 3 points nationally on the day he dropped out. (You can still access our old Biden-Trump averages by clicking the link in the banner at the top of our national polling average page or at the top of any state polling page that used to have a Biden-Trump average.) But it's an open question whether that will last. As the new nominee, Harris may be enjoying a honeymoon period, and an amped-up Democratic base may be more likely to respond to surveys than Republicans — a phenomenon known as differential partisan nonresponse. It's also possible that the initial boost of enthusiasm she has received will fade over time. You can bookmark the interactive page for our new averages and keep up with the latest polls as soon as they come in.
Footnotes
*As explained in our polls policy, we do not publish polling averages of presidential general elections in a given state until we've aggregated at least five polls of that state from at least three pollsters. In this unique circumstance, we are further limiting that to polls conducted entirely after Biden dropped out. For aesthetic reasons, we also wait a few days after meeting this threshold to publish the average, so that the chart doesn't just have one point and one day of a line on it. That's why Arizona and Wisconsin, which have already met the five poll/three pollster threshold, don't have averages yet; we plan to publish those by Monday.
