Republican presidential candidate Donald Trump speaks during a campaign rally at the Westgate Las Vegas Resort & Casino on Dec. 14, 2015 in Las Vegas, Nev.
Photo by Justin Sullivan/Getty

What if the polls are underestimating Trump’s support?

Updated
We’ve grown quite accustomed in recent months to looking at Republican polls and finding the same result: Donald Trump is doing very well in the GOP presidential race. But look a little closer and an unexpected detail emerges.
 
FiveThirtyEight published a piece earlier this month noting that in traditional polls – a human being calls respondents and asks scripted questions with multiple-choice answers – Trump’s support is lower than in polls conducted online or through automated, robo-call surveys.
 
And while some subtle variances are easily overlooked, the FiveThirtyEight analysis found a sizable, statistically significant difference.
 
So, what’s the explanation for this? Should we discount online polling that has no real track record in primary races? Should we assume this is just a temporary fluke that will sort itself out once the primaries begin in earnest?
 
It’s worthy of a conversation, but Bloomberg Politics ran an interesting piece yesterday raising a possibility that hasn’t generated much attention.
Polls may be underestimating Donald Trump’s support, according to intriguing new research that says the Republican front-runner benefits from a “social desirability bias” – some people who plan to vote for him are too embarrassed to admit it.
 
The study, published Monday by Morning Consult, found that Trump fares about six percentage points better among likely Republican voters in online polls than when a pollster is speaking by phone to a live human being. Moreover, the report indicated, the higher an individual’s educational attainment, the greater the likelihood that the respondent wouldn’t admit on the telephone to supporting Trump.
Now seems like a good time to pause and talk a little bit about the “Bradley Effect.”
 
In California’s 1982 gubernatorial campaign, polls showed then-Los Angeles Mayor Tom Bradley (D) comfortably ahead in the statewide race. The polls, however, were wrong – and Bradley lost in a very close contest.
 
Some social scientists argued soon after that the pre-election data were off for a very specific reason: voters lied to pollsters about their intentions towards Bradley, an African American, who ran against a white rival. It came to be known as the Bradley Effect.
 
The Wikipedia description is as good as any: the theory “posits that the inaccurate polls were skewed by the phenomenon of social desirability bias. Specifically, some white voters give inaccurate polling responses for fear that, by stating their true preference, they will open themselves to criticism of racial motivation. Members of the public may feel under pressure to provide an answer that is deemed to be more publicly acceptable.”
 
The theory has plenty of critics. Your mileage may vary.
 
What does this have to do with Trump? Because the Republican frontrunner does noticeably better in polls in which there is no human-to-human interaction, some believe we may be seeing a sort of reverse-Bradley Effect – instead of having a candidate voters feel awkward about opposing, we see a candidate voters feel awkward about backing.
 
In other words, there’s speculation that some Republican respondents may be embarrassed to voice support for Trump when an actual person calls their home, but feel less awkward in online or robo-call polls, which might explain the gap.
 
And if that’s true, Trump’s support in traditional, independent, historically reliable polls may be understated a bit, because some of his supporters feel awkward about admitting it out loud.
 
I’m not necessarily endorsing the idea, though I am curious about the difference in his support based on the kind of poll. It’s something to keep an eye on.

Donald Trump and Polling

What if the polls are underestimating Trump's support?

Updated