Sen. Elizabeth Warren’s campaign has said she has drawn crowds of up to 15,000. Meanwhile, former Vice President Joe Biden has not exactly been packing them in, even as he continues to lead by a healthy margin in most polls of the Democratic presidential primary. So could Warren’s big crowds be picking up on something that the polls are missing?
The short answer is: No. While the ability to generate big crowds is certainly nice — it may signal enthusiasm among highly engaged voters or produce favorable media coverage — you should ignore any candidate, surrogate or media outlet that tells you that large crowd sizes mean that the polls are underestimating a candidate’s support. It’s just spin; polls are much more accurate at forecasting elections than crowd-size estimates, which don’t tell us all that much.
For every example like 2008, when then-Sen. Barack Obama’s huge crowds seemed to reflect real enthusiasm for his campaign, there is one like 2012, when former Massachusetts Gov. Mitt Romney won his primary despite drawing noticeably small crowds on the campaign trail. Or take what happened in 2016. Despite a lot of hay being made about crowd sizes during the 2016 campaign, that cycle also was an argument against crowd sizes being predictive. Although now-President Trump did often draw large crowds at his primary rallies, Hillary Clinton reportedly beat him out for largest crowd of the 2016 campaign, 40,000 to 30,000. And at roughly this point in the Democratic primary in 2016, Sen. Bernie Sanders was outdrawing Clinton!
One of the many problems with crowd-size estimates is that they can be extremely rough, and they’re subject to reporting bias. (If Warren says she drew 10,000 people to a college campus, but the university says the crowd was only 5,000, whom would you believe?) There are also a ton of factors other than enthusiasm for the candidate that can affect crowd sizes: Where is the event being held? (Is it in New York City, or in a small town in Iowa?) How frequently does the candidate hold events? (If candidates are frequent visitors to an area, there is perhaps less urgency for voters to attend any one rally.) Are there other draws besides the candidate? (For example, that Clinton rally that drew 40,000 also featured performances by Bruce Springsteen and Jon Bon Jovi.)
Like Google search data, crowd sizes may also be more of a measure of curiosity about a candidate than actual support. Voters don’t just go to rallies to cheer on their candidate; they also go to learn more and maybe allow themselves to be persuaded. In the 2020 primary, it would therefore make sense that known commodities like Biden might get smaller crowds, while novelties like Warren might get bigger ones.
And like small donors, the people who attend campaign events may skew toward a certain demographic — disproportionately upper-income, well-educated and white. For instance, those who have the time and financial flexibility to attend a campaign event are probably relatively affluent. And according to a 2018 poll from the Pew Research Center, 35 percent of people with college degrees, and 41 percent of those with postgraduate degrees, said they had attended a political rally or event in the past five years; only 20 percent of those with no college education said they had. Liberal Democrats were also much more likely than moderate or conservative Democrats to have attended a campaign event. And given that many political events are held in the evening, that might preclude a lot of people who work evening shifts, who are disproportionately black, Hispanic or low-income, from attending too.
In other words, crowds at political events are self-selected. By contrast, polls are scientific instruments that use proven sampling techniques and statistical weighting to ensure that they are reflecting a representative population. As such, the former should never outweigh the latter, or else you’ll end up overestimating the standing of plugged-in whites’ preferred candidate.
And this actually might go a long way toward explaining the difference between Warren’s massive mobs and Biden’s smaller crowds. Polls say Warren performs especially well among college-educated, white, liberal voters, while Biden’s support is largely concentrated among groups that might not show up at rallies as often — black voters, moderates and working-class voters. So the next time you see a story about crowd size in the Democratic primary, remember that crowd sizes can mislead you if you let them. Don’t let them.