Trump may have a reverse Bradley effect

As I’ve written a few times, polls that rely on their respondents to self-report whether they are registered and likely to vote could be overestimating Donald Trump’s support. But what if racism among college-educated Republicans is leading those same polls to underestimate it?

At least, that’s one possibility suggested by an analysis published yesterday by Morning Consult.

The polling firm collected a large (roughly 2,500 person) sample of self-reported registered voters who identified as likely Republican primary voters. All respondents began the survey online, but only one third stayed online the whole way through. One third completed the survey through an interactive voice response (IVR) system, while another third completed the survey via live-caller phone conversations.

The survey found that Republicans and Republican-leaning independents who completed the survey online were roughly six percentage points more likely to support Donald Trump than those who completed the survey with a live caller — the difference jumped to nine percentage points when the sample was limited to likely voters. Similar gaps were not observed for other candidates, and IVR responses tracked more closely to the online results than the live-caller results.

Donald Trump, screenshot via YouTube

Donald Trump, screenshot via YouTube

What’s more, the difference between Trump’s online and live-caller support was more pronounced among respondents with higher levels of education. Respondents whose education topped out at high school were equally likely to support Trump online versus over the phone, while college graduates were much less likely to tell an interviewer that the supported Trump than they were to click his name on the web.

Usual caveats about polling samples and random chance aside, the simplest explanation for this finding is that college-educated Republicans are exhibiting some degree of social desirability bias: A significant number of them support Trump, but are embarrassed to tell a stranger. Unlike their less-educated peers, the argument goes, their public support for Trump comes with a social cost, making them less likely to state their true attitudes when an actual person is listening. When push comes to shove and they’re given a secret ballot, however, we should expect these voters to show up for Trump.

This is the opposite of the Bradley effect, a term coined in 1982 when Los Angeles Mayor Tom Bradley lost the California governor’s race despite leading in the polls heading into Election Day. The term is usually invoked to describe white voters exhibiting social desirability bias to say that they’re either undecided or planning to vote for a black candidate, only to turn out to vote for their white opponent. In this case, college-educated whites seem to be exhibiting the same kind of bias, only this time they’re saying that they’re either undecided or not planning to vote for a borderline white nationalist, when in fact they would support him if the election were tomorrow.

Interestingly, this reverse Bradley effect isn’t mutually exclusive with the registration gap I noted above. Live-caller polls that rely on self-reported voter characteristics could be both underestimating Trump’s support among college-educated Republicans and overestimating turnout among Republicans without a college degree, who are more likely to tell the interviewer that they support Trump. Perhaps as interestingly, the effect of this registered voter gap has been shown to be even larger than the reverse Bradley effect Morning Consult found yesterday.

All this is to say that when you see flurries of polls showing wild fluctuations in the topline support for various candidates — especially in primaries — methodology matters. A lot. When push comes to shove and Republicans actually start casting ballots, we’re going to find out just how accurate, and inaccurate, the assumptions these pollsters are making about the electorate turned out to be.

Jon Green graduated from Kenyon College with a B.A. in Political Science and high honors in Political Cognition. He worked as a field organizer for Congressman Tom Perriello in 2010 and a Regional Field Director for President Obama's re-election campaign in 2012. Jon writes on a number of topics, but pays especially close attention to elections, religion and political cognition. Follow him on Twitter at @_Jon_Green, and on Google+. .

Share This Post

© 2018 AMERICAblog Media, LLC. All rights reserved. · Entries RSS