Every election forces us to re-examine many of the assumptions we've been making about voting behavior. The 2016 election brought a recognition of the 'education gap' among white voters — those with a college degree preferring Democrats and those without voting overwhelming Republican. But, even as pollsters and pundits admitted that not weighing on education was a significant source of error in polling in 2016, there was little effort made to examine whether this same dynamic was showing up among voters of color.
In digging through the 2020 voter data provided by the Democratic data firm Catalist, Third Way's Aliza Astrow found that even as Biden was able to slightly improve on Clinton's showing with white, non-college voters, "Democrats endured a sharp drop-off in support" from non-college voters of color. In 2016, according to the data from Catalist, Clinton took 81 percent of the vote from non-college voters of color. In 2020, Biden took 75 percent among this group, a 6-point drop.
While it's hard to characterize a 75 percent showing as 'weak,' Democrats' heavy reliance on voters of color means the party can't afford to see more slippage among this demographic group in upcoming elections. Democrats' long-term viability in Sun Belt states like Arizona, Texas, Georgia and North Carolina require more than just winning over white suburbanites and not losing any more ground with white, non-college voters. They have to continue to run up the score with voters of color.
As with white voters, there is a decent gender gap among voters of color, both among non-college and college-educated voters. But, the drop in support among women (both college and non-college-educated) for Democrats between 2016 and 2020 was significant. For example, while Biden did 5 points worse among non-college men of color and 2 points worse among college-educated men of color than Clinton, he performed five points worse among college-educated women and seven points worse among non-college women. While many suspected that Trump’s appeal was unique to male voters of color (some attributed it to Trump’s direct appeal to ‘machismo’) Astrow’s analysis shows that he gained among women voters of color too.
Astrow's analysis also revealed that, unlike white voters, the “education gap” among voters of color shrunk - not expanded - over the last eight years. During the Trump era, the conversation about growing support for Democrats from white, college educated voters was discussed ad nauseam. And, to no one’s surprise, Biden carried white college-educated voters by a 17-point margin (54 percent to 37 percent); an 11-point increase from Obama's 2012 showing.
But, among voters of color, Democrats once small advantage with non-college voters disappeared in 2020. And, while Democrats lost some ground with college-educated voters of color as well, it was not as steep as it was with non-college voters (3 points versus 6 points). However, what worries Democrats like Astrow is the fact that non-college voters of color "are a growing share of the electorate." In 2020, non-college voters of color outnumbered college-educated voters of color by a 2-1 margin.
But identifying this gap — and understanding why these voters are voting the way they do — is only as good as keeping tabs on these voters in the future. And, that is going to be a lot harder — and more expensive — than tracking white voters.
Voters of color are a smaller share of the electorate, and as such, they make up a smaller percentage of respondents in a traditional poll. The smaller the size of the demographic group, the harder it is to break that group into smaller chunks (like by education or specific racial categories). This challenge is evident even in this analysis by Third Way, which grouped all non-white voters into the catch-all "voters of color" category instead of breaking it out into Latino, Black, Asian, etc.
To understand what it would take to better capture shifts in opinion by voters of color, I checked in with Ashley Kirzinger, a pollster with the Kaiser Family Foundation (KFF). The Cook Political Report and KFF teamed up last year for two surveys of key battleground states in the upper Midwest and the Sunbelt.
She told me that "our current sampling strategy in Arizona yielded 1,298 voters including 219 Latino voters in the state. We had already provided additional incentives for Latino voters and the survey materials (and interviews) were conducted in English and Spanish. Using this sampling strategy, we got 160 Latino voters with less than a college degree and 59 that are college plus. To have large enough sample sizes to look at Latino voters by college degree, it would require increasing our sample by 600 individuals (regardless of race)."
Adding another 600 voters to a poll will take time — and money. It's not a cheap endeavor.
But, we can't have a rigorous or serious analysis of how voters of color are interpreting the candidates and the campaigns without real data.
This will be especially important next year as both sides battle over majority non-white CDs like south Florida, the Rio Grande Valley in Texas, and Los Angeles/Orange County in California, where Trump and/or GOP House candidates performed unexpectedly well in 2020. Was this increase in GOP support just a short-term, Trump-centric bump? Or, given what we have seen over the last 10-15 years among white voters, are non-college voters of color coming to identify more closely with the values and ideology of the Republican Party? Are future elections going to be marked by increased polarization by education and decreased polarization by race?
At the national level, and in key battleground states in the Midwest, white non-college voters remain the largest share of the electorate. Democrats can't afford to lose more ground with these voters. But, this group is also (slowly) shrinking while voters of color without a college degree continues to grow. Given how close races in states like Georgia, Arizona, North Carolina and Florida will continue to be in the near future, understanding the intentions and motivations of both types of voters will be critical for both parties to win elections.
Our subscribers have first access to individual race pages for each House, Senate and Governors race, which will include race ratings (each race is rated on a seven-point scale) and a narrative analysis pertaining to that race.