Philip Bump | May 10, 2021
Since polls closed Nov. 3, there’s been a lot of focus on the extent to which Black and Hispanic voters seem to have been more open to President Donald Trump than they were in 2016. Preliminary exit polls showed Trump doing better with non-White voters than he had four years prior — a shift that certainly defies perceptions of Trump’s relationship with those communities.
Sophisticated new analysis of the election results, combining polling data with a look at the voting records of millions of Americans, helps explain why that shift occurred and, more important, how it relates to the eventual outcome.
The data firm Catalist explored the 2020 electorate in a lengthy report published Monday. It breaks out the voter pool along several demographic lines, including race and education. In particular, the analysis splits up the White electorate into those with and without a college degree, a divide that emerged as essential in 2016 and 2018.
That yields a rough estimate of who turned out and how they voted that looks like this. The bars are scaled vertically relative to the percent of the electorate of each group. (Whites without college degrees were 44 percent of the electorate, for example.) The blue and red bars show the support for Joe Biden and Trump within the group. (Whites without degrees heavily preferred Trump.)
What should stand out on that chart is how White voters without college degrees are the only group that preferred Trump to Biden. Whites with degrees and Black, Hispanic and Asian (identified as AAPI: Asian Americans and Pacific Islanders) voters all preferred Biden to some degree.
One upshot is that Biden’s support was much more diverse than that of Trump. About 6-in-10 Biden voters were White, including a third who are White and have no college degree. More than half of Trump’s support came from that category alone, with 85 percent of his support coming from White voters overall.
This isn’t surprising, of course. The Democratic Party has consistently been more diverse than the Republican Party; it was actually more diverse 25 years ago than the GOP is now. But given the attention paid to Trump’s improvements with non-White voters, it’s important to point out that Biden got twice as much support from White voters without college degrees as Trump did from voters who weren’t White.
It’s also important to contextualize that shift to Trump with the overall change in the electorate. 2020 saw a decline in the density of White voters without college degrees relative to 2016 and 2012 before that, according to Catalist. The density of Hispanic voters increased, while the density of Black voters remained about the same, down from 2012 when President Barack Obama was seeking his second term. (We’ve excluded AAPI and other demographic groups from these charts for the sake of clarity.)
The level of support earned by Biden (measured as a share of votes for either the Democrat or the Republican since 2012) dropped among both Black and Hispanic voters while it increased substantially for Whites with college degrees and even ticked up slightly among White voters without a degree.
All of that focus on the declines obscures a key point elevated by Catalist: When combined with the increase in turnout, the net result was more votes for the Democrat.
If we roughly apply Catalist’s estimates for the density of each group to the number of votes cast in each election, you see that there were more voters in each group than there were in 2016. Yes, Whites without a college degree made up less of the electorate than four years prior, but turnout of those voters increased 10 percent.
Think about what that means. If in 2016 there were 1,000 Black voters who backed the Democrat by a 75-to-25-percent margin, that’s a net of 500 votes in favor of the Democrat. If the margin among Black voters drops to a 72-to-28-percent split but the number of voters increases to 1,200 — the Democrat nets 28 more votes. The divide between the Republican and the Democrat widens.
At some point, the increase in turnout wouldn’t be enough to offset a significant drop. That’s likely what happened with Hispanic voters: The drop in support was large enough that even with increased turnout, it meant a net loss of votes for Biden relative to 2016. The same pattern also probably worked against Biden with Whites who don’t have a college degree. Trump appears to have gained in his vote margin with those voters simply because of the increased turnout.
But among Black voters in particular, Catalist estimates that the increase in turnout was enough to not only offset the drop in the Democratic margin but, in fact, to win two key states. In Georgia, for example, 200,000 more Black voters cast ballots than in 2016 in a state Biden won by about 12,000 votes. In Arizona, an additional 30,000 Blacks voted, more than twice the state’s 10,000-vote difference.
There are two key questions that emerge. The first is who those new voters were. The second is what the long-term trend in non-White voting might be.
To answer the first question, Catalist estimates that about 14 percent of 2020 voters had never cast a ballot before. About that same number had never voted in a presidential race in their states before, meaning that they had either voted in 2018 or were casting new ballots in states to which they’d moved. (This comports with the finding that the density of new voters was higher in states that have seen bigger increases in their populations.) About two-thirds of those new voters were under the age of 30, reflecting to some extent young people who were able to vote for the first time.
That newer voters tended to be less densely White might also help explain the shift in Black and Hispanic vote margins since 2016.
“As with Latino voters,” the firm’s Yair Ghitza and Jonathan Robinson write, “it is plausible that new marginal voters entering the electorate for the first time naturally drove the overall support numbers slightly down, because these new voters are less ideological than longtime voters.”
This overlaps with analysis conducted by the firm Equis Labs. Its research found that Trump helped drive turnout particularly among more conservative Hispanic voters that helped him eat into Biden’s advantage. Its findings mirrored Catalist in another way, finding divergences among Hispanic voters depending on national origin.
Answering the second question — the one about the long-term trend — is far trickier. To some extent it depends on the ability of future candidates to inspire turnout in the way that Trump did in 2020 (both on the left and the right).
The Catalist data hints at one answer.
“[T]here are still major gains to be made in voter turnout writ large,” it reads. “This is particularly true among communities of color, where nonvoting rates are substantially higher.”
If Hispanic voters continue to make up more of the electorate than White voters without college degrees — a group that preferred Biden by 25 points gaining and a group that preferred Trump by 25 points declining — there are certainly some straightforward assumptions one might make.