Throughout Donald Trump’s presidency, a persistent media story-line held that Democrats were out of touch with the true sentiments of a certain type of economically populist white voter, particularly those concentrated in the industrial Midwest.

Coming after Trump shattered the “blue wall” states in 2016, his grip on his non-college white base through displays of presidential incompetence, malevolence and extraordinary corruption further entrenched that story line. When Trump outperformed Democratic expectations in 2020, the elite Democratic self-flagellation continued, even in victory.

The leading Democratic polling firms are undertaking a systematic examination of what went wrong with their polling in 2020, and it increasingly looks as if this self-criticism is being taken seriously.

I spoke to several Democratic pollsters involved in this self-assessment, and they helped clarify what it’s supposed to accomplish. Non-college white voters are at the center of it.

Last month, five top Democratic polling outfits released a memo announcing an effort, one unusual for its coordination among competitors, to assess why Democratic polling had underestimated Trump’s strength in some parts of the country.

The memo outlined some early conclusions: Democratic pollsters had failed to connect with low propensity Republican voters, who are disproportionately rural and white non-college Americans.

The result was that Democratic polls consistently misconstrued the makeup of the low-propensity electorate across the board, especially in conservative parts of the country. Those low-propensity voters ended up voting Republican at a vastly higher rate than Democratic polls had registered.

Now these Democratic firms have drilled more deeply into why this might have happened. As their pollsters told me, one possible explanation is that the non-college white vote is divided roughly into two segments. Though this is preliminary, these segments can roughly be grouped along a spectrum of how activated they are by Trump himself.

Molly Murphy, a partner at ALG Polling — which does polling for President Biden — says that among non-college whites, the dividing line might be between more traditional Republican voters, and those who found themselves far more energized by Trump himself than by party loyalty.

“One operating theory is there is this group of voters who are turned on and activated by Trump," Murphy told me. “They are not generally political participants. They are not big believers in institutions.”

“They are not as willing to participate in polls and talk to strangers about their opinions,” Murphy continued, suggesting this might be why those voters proved hard for pollsters to reach.

One possibility that Democrats are exploring, Murphy said, is that these latter voters might be driven by “a larger sense of social distrust. Trump’s rhetoric and willingness to take on institutions spoke to the anger at the establishment they were feeling.”

This dovetails with a suggestion from polling analyst Steven Shepard, that Trump might have exacerbated this effect with his attacks on polls as deliberately rigged against Trump and the media as out to get him.

Still, Murphy also cautioned that this shouldn’t be looked at just in racial terms. She noted that Trump did better among non-college voters across the racial spectrum, and said Democrats need to look hard at why they missed those voters as well.

Nick Gourevitch, a partner at Global Strategy Group, says one way of understanding the more Trumpy of these non-college white demographics might be through their openness to Trump’s attacks on institutions.

Throughout 2020, we had a “national debate about how to respond to Covid,” Gourevitch told me. “People who were less likely to answer polls probably sided with Trump against the strong majority that followed guidance on how to respond to the virus.”

“The traditionally Republican voter who did buy into mask-wearing and social distancing was more likely to support Biden,” Gourevitch continued, adding that while these voters helped Biden win, they “might have been over-represented in polls.”

One big question concerns the role of the economy. It’s often taken on faith that Trump-enthralled white voters tend to be economically struggling. But Gourevitch suggested many non-college white voters in the less Trumpy and more gettable camp, ones Biden might have made inroads among, might also be in that struggling category — in other words, there are struggling voters in both camps.

As this self-assessment continues, these Democratic firms will be considering experimental efforts, such as door-to-door polling, to try to reach these under-represented voters more effectively. But wherever this self-assessment ends up, the stakes in getting it right are high.

“What’s good about this project is that pollsters are getting together to talk to these critical voters,” Jefrey Pollock, president of Global Strategy Group, told me. “For the Democratic Party to succeed long term requires a broad coalition. That includes doing better with these non-college voters. We cannot cede them forever.”

Read more: