Takeaway #1

Despite Democrats’ appraisal of themselves as the “pro-diversity, anti–white supremacy” party, Joe Biden’s coalition depended, above all, on affluent, white suburban voters.

Biden improved over Hillary Clinton in highly educated, mostly white, large metropolitan counties by at least twice the rate seen in other county demographics.

We looked at how much of Biden’s improvement over Clinton in key battleground states came from large metropolitan, high-education counties compared to counties with large populations of white, black, and Latino working-class voters. We found that, in three of the five states examined — Georgia, North Carolina, and Pennsylvania — Biden’s improvement over Clinton depended most heavily on affluent, white suburban counties.

The key exception was in Texas, where Biden’s increase over Clinton accounted for a slightly larger share of the 2020 vote in counties with large Latino populations compared to affluent suburban counties. Biden’s increase over Clinton in counties with large black and white working-class populations accounted for a small percentage of the 2020 vote in all five states.

When we account for a range of other factors that might have affected Biden’s improvement over Clinton, we see that education is the strongest positive predictor. We also observe, however, that Biden made advances over Clinton in progressive strongholds (many of which are also high-education counties), such as counties where Bernie Sanders performed well in the 2020 Democratic primaries, and counties that experienced increased unemployment between 2016 and 2020 (these tend to be young, low-poverty, low-homeownership, and relatively urban counties). In other words, Biden did better than Clinton not just in wealthy suburbs, but also in gentrifying urban areas.

Within large metropolitan areas, Biden did better than Clinton in nearly every county with a large share of professionals in the workforce, an above-average median income, and an above-average level of educational attainment. And his margins over Clinton improve as the share of professionals, high-income individuals, and highly educated individuals grows.

Takeaway #2

Working-class voters shifted to the right across all racial groups.

Despite being helmed by a right-wing president fond of racist invective, the party of angry old white people got less white and more working class in 2020. We see attrition from the Democratic Party not just in areas with large white working-class populations, but also in areas with large black and Latino working-class ones.

After accounting for a range of factors that could have affected increased support for Donald Trump in counties with large working-class populations, we find that factors associated with economic insecurity and low population are positive predictors across all. These include counties that are more rural, with higher rates of poverty and home­ownership (strongly associated with rural, white, and older communities).

Takeaway #3

Turnout was up across the board, but weakest in working-class black areas.

Both Joe Biden and Donald Trump earned more votes than any other presidential candidates in American history. However, the increases in turnout were not even across the board. Counties with large white populations had the highest overall turnout, followed by counties with high Latino populations and counties with high black populations.

We see a similar pattern when we look at changes in turnout rates between 2016 and 2020, with the greatest voter enthusiasm coming from counties with large Latino populations, followed by those with large white and black populations.

Why do we see such large turnout increases in counties with high Latino populations? Our analysis suggests that this is due to a number of factors, including strong turnout among younger voters, higher-income voters, lower-education voters, and big increases in the South, Southwest, and West. This could also have been the result of targeted efforts by campaigns to mobilize voters in these particular areas.

Among counties with large proportions of black voters, wealthier counties saw consistently high turnout increases, while lower-income counties saw much weaker turnout increases (almost zero in counties with very large black populations). We see a similar, if less pronounced, difference in turnout changes between counties with large black populations and education levels above and below the median. Interestingly, however, we see no such trends among counties with high Latino or high white populations, where turnout increases were consistently high across class demographics.

After accounting for a range of factors that could affect voter turnout in counties with large white, black, and Latino populations, we find that income is the strongest positive predictor of turnout in each. This indicates that, despite important differences described above, the determinants of turnout increases were relatively uniform across class. This is consistent with historical precedent, and it suggests that class was a key predictor of voter turnout in 2020 — the more affluent the county, the more Democratic.

To get a sense of the practical significance of these turnout patterns, we calculated the percentage of the 2020 presidential vote that was attributable to turnout increases in key battleground states across county types.

We find that underwhelming turnout increases in counties with large black and working-class populations across the five key battleground states examined (Florida, Georgia, North Carolina, Pennsylvania, and Texas) could have significantly impacted Biden’s chance of victory in each. Turnout increases in those counties accounted for a smaller share of the total statewide vote than increases in all other county types across all five states (with the partial exception of Georgia, where increases in black and working-class counties made up a larger share of the vote than in high Latino and working-class counties, though less than in white working-class or large metropolitan, high-education counties).