Mary pointed out that there are some indications that polls could be overstating Harris' weakness with Black voters due to the difficulties in getting a representative sample among those voters. I wanted to add that polling is challenging in general, and polling Black voters is especially so.

After an election, pollsters calibrate for the future based on actual election results. But individual votes are of course private, so there's no gold standard that tells us how voters from certain demographic groups — including Black Americans — voted. And when pollsters do recalibrate after an election, they are far more likely to focus on the more numerous white voters.

It's also harder to recruit representative samples of Black voters in the first place. I recently co-authored an article, "Getting the Race Wrong," discussing the online and in-person surveys we conducted here in Philadelphia during the 2023 Democratic mayoral primary. In that race, our polls mirrored the public polls, getting the levels of support for Rebecca Rhynhart (the city's controller, who had significant support from Center City’s liberal neighborhoods) and Helen Gym (an at-large city councilmember who campaigned with Bernie Sanders and drew support from Philadelphia's most progressive neighborhoods) close to right.

But like the public polls, we significantly understated the support for the eventual winner, Cherelle Parker, a former city councilmember and state House representative. A big part of the reason why is that Parker dominated in Philadelphia's majority Black neighborhoods, in a city whose Democratic electorate is majority Black. Even when we upweighted the Black respondents who participated in our survey, our polls understated her support, in part because our Black respondents were significantly more likely to have a college degree than the general population of Black voters here in Philadelphia.

Throughout this year, we've been tracking how both Trump and Harris have been performing with key demographic groups. And we've seen a common trend emerge over the last few months: In polls, Harris appears to be improving upon Biden's 2020 performance among older voters and white voters, and losing ground among younger voters and voters of color. The latest round of polling before the election appears to continue the trend.

Harris is polling, on the margin, 6 percentage points better than Biden's 2020 performance among white voters, while she's lost support among Black voters and Hispanic voters. But white voters make up a much higher percentage of the electorate than other demographic groups: 67% of voters in 2020 identified as white, according to the 2020 exit polls. And in the key northern battleground states of Michigan, Wisconsin and Pennsylvania, the white share of the electorate is even larger — 81% in Michigan and Pennsylvania and 86% in Wisconsin. This may explain some of the trends we've seen in the polls over the last few months, where Harris performs better in these three states than in the other swing states.

However, there is some indication that the national polls could be overestimating Harris' relative weakness among voters of color. We've seen some polls conducted specifically among Black and Hispanic voters that show less movement towards Republicans. Polls like these can sometimes be more reliable, as they are able to get a more representative sample of voters that may be difficult to reach for national surveys.

-Mary Radcliffe

After voting for the Democratic candidate in each presidential election since 1992, Pennsylvania narrowly flipped into Trump's camp in 2016, only to swing back to Biden in 2020. But Biden won by only 1 percentage point, and the state once again looks to be extremely close this time around, with 538's final forecast of the state showing essentially a tied race.

Not coincidentally, Pennsylvania is the most-polled state in the 2024 election cycle. For one thing, its 19 electoral votes make it the largest of the seven core swing states. And because of its size and competitiveness, Pennsylvania is also the most likely "tipping-point" state in 538's presidential election forecast.

Like many of the key states this cycle, one of the main dividing lines in the Pennsylvania electorate is education, particularly among white voters. Pennsylvania's overall population is whiter than the country's as a whole: about 73% non-Hispanic white compared with 57% nationally. Within the white population is a large and notable chunk of the state's electorate: white voters without a four-year college degree, who accounted for 45% of Pennsylvania's voters in 2020, per exit polls. Nationally, this GOP-leaning group has shifted further to the right in the Trump years.

Conversely, Democrats have made gains nationally among white voters with at least a four-year college degree, and the same is true in Pennsylvania. For instance, Philadelphia's four suburban collar counties are fairly white and have nearly 30% of the state's population that is white with a college degree. Not coincidentally, the quartet has collectively trended more Democratic than anywhere else in the state in recent years, backing Obama by 10 points in 2012 and Biden by 19 points in 2020, even as the state as a whole moved to the right over that time span.

As the state's largest city, Philadelphia forms a key part of the state's electoral math. But while the city's collar counties have moved left, the city proper actually has shifted somewhat to the right, as Democrats' substantial edge there declined by about 4 points in each of the last two presidential cycles. And Black voters are a big part of the equation here: While only about 1 in 10 of Pennsylvania's voters identified as Black in 2020, nearly half the state's Black population lives in Philadelphia. And while Democrats appear to have gained in the more well-educated and affluent parts of Philadelphia, predominantly Black precincts have experienced at least a small decline in Democratic support and, also importantly, turnout rates. Recent polling suggests this could be a big concern for Harris: An average of polls in the state found her ahead 79% to 17% among Black voters, well down from Biden's advantage in 2020 of 92% to 7%, according to exit polls.

And there's more to watch in Pennsylvania than just the presidential race. First, a critical Senate race between three-term Democratic incumbent Sen. Bob Casey and Republican Dave McCormick will be key to Democrat's slim chances to hold the chamber. Casey is favored in our final election forecast, winning 72 in 100 simulations, but that still leaves a real chance for McCormick to pick up a surprise win in the state.

There's also at least three highly competitive House seats, in the 7th, 8th and 10th Districts. All of these feature incumbents representing areas that have been trending away from their party over the last few cycles. And finally, the contests for attorney general, auditor general, treasurer and control of the state House are all competitive.

If you just read Geoff's post about the most likely tipping point states and are curious to take a closer look at how those states stack up, take a gander at our snake chart! According to 538's final election forecast Pennsylvania's segment sits right on the 270 electoral vote threshold, meaning that as each candidate builds their coalition, the Keystone State could be make-or-break for the electoral college outcome.