2016
Poll: 'Shy Trump' voters are a mirage
A new study suggests an army of hidden Trump supporters is unlikely to materialize on Election Day.
By
STEVEN SHEPARD
11/03/16 08:58 AM EDT
Donald Trump has insisted for months that polls are failing to capture the breadth of his support because some of his backers won’t admit it to a pollster over the telephone.
He’s wrong.
Story Continued Below
According to
a POLITICO/Morning Consult study conducted by Morning Consult this past weekend and released Thursday, a hidden army of Trump voters that's undetected by the polls is unlikely to materialize on Election Day.
The study — which was composed of interviews with likely voters conducted over the phone with a live interviewer, and other interviews conducted online without a personal interaction — showed only a slight, not-statistically-significant difference in their effect on voters’ preferences for president.
That’s despite an emerging disparity in the public polling: Hillary Clinton leads Trump by 4 percentage points in national, live-interview phone polls, according to HuffPost Pollster’s average as of Wednesday night — but only by 2.6 points in online and automated phone polls.
Trump asks supporters to 'pretend we're down'
By
STEVEN SHEPARD,
LOUIS NELSON and
NOLAN D. MCCASKILL
As Trump closes the gap with Clinton and moves into striking distance of the White House, a systemic polling error of only a few points could tip the scales. And while the overall difference was negligible, between completing the survey with complete anonymity and talking to a live interviewer over the phone, there were important variations with key demographic groups that underscore the cultural divides that have defined this historic election.
“In the primaries, we conducted a study that sought to get at why Donald Trump was performing better in online surveys than on telephone polls,” said Morning Consult co-founder and Chief Research Officer Kyle Dropp. “The study showed that a polling mode effect called ‘social desirability bias’ — respondents didn’t want to tell a live interviewer they were voting for Trump — was at play. Our new study to assess the general election shows that not to be the case overall, although there are likely still small pockets of shy Trump voters.”
ADVERTISING
inRead invented by Teads
Overall, the POLITICO/Morning Consult study released Thursday showed Clinton with a 5-point lead among voters interviewed over the phone, 52 percent to 47 percent.
The race was modestly closer in online interviews: Clinton’s share of the vote ticked down a single point, to 51 percent. Trump’s support inched up a point, to 48 percent.
The “shy Trump” theory centers around a basic premise: Given Trump’s controversial candidacy — which has featured attacks on electoral norms he derides as “political correctness,” provocative policy proposals like a ban on Muslims entering the U.S. and vulgar comments about women — some voters may perceive it is socially undesirable to admit support for the Republican nominee.
It is similar to a phenomenon seen in Great Britain, where Conservative Party candidates have sometimes outperformed their poll numbers because voters did not want to admit they would back the Tory in their districts. Many cited a similar occurrence earlier this year, when British voters decided, in a nationwide referendum, to leave the European Union.
In American politics, the 1982 gubernatorial election in California is often cited as an example of social-desirability bias skewing the polls. The race is so legendary — then-Los Angeles Mayor Tom Bradley, an African-American, led in the polls but lost to Republican George Deukmejian on Election Day — that this phenomenon is known in the U.S. as the “Bradley effect.”
If social-desirability bias were to boost Trump next week, it would be the reverse of what some contend befell Bradley: Rather than voters saying they will vote for Bradley and then supporting Deukmejian, Trump’s campaign is betting that voters are saying they will vote against him now, but will cast a vote for him in the privacy of the voting booth.
CNN polls: Clinton leads Trump in Pennsylvania and Florida but trails in Nevada and Arizona
By
MADELINE CONWAY
Trump has called himself “Mr. Brexit” on numerous occasions, and he has cited the Bradley effect as evidence that he will outperform his poll numbers significantly on Nov. 8.
“He was supposed to win by 10 points, and he lost by 5 or something. So it’s a certain effect,” Trump said back in June, alluding to Bradley without mentioning him by name. “Now, I have — unfortunately, maybe fortunately — the opposite effect. When I poll, I do fine. But when I run, I do much better.”
“In other words, people say, ‘I’m not going to say who I’m voting for,’” Trump added. “And then they get it, and I do much better. It’s like an amazing effect.”
Kellyanne Conway, Trump’s campaign manager and a longtime Republican campaign pollster, subscribed to this same theory in an interview with a British television outlet this summer.
“Donald Trump performs consistently better in online polling where a human being is not talking to another human being about what he or she may do in the elections,” Conway said. “It’s become socially desirable, especially if you’re a college-educated person in the U.S., to say that you’re against Donald Trump.”
The POLITICO/Morning Consult study suggests Conway — who didn’t reply to a POLITICO request for an interview on Wednesday — is at least half right. While the overall results found completing the survey online didn’t boost Trump significantly, he did run stronger among more-educated voters.
In phone interviews, likely voters with a college degree said they support Clinton by a 21-point margin, 60 percent to 39 percent. But online, that margin shrunk to just 7 points, 53 percent to 46 percent.
Similarly, among likely voters in households earning more than $50,000 annually, Clinton leads by 10 points over the phone, 54 percent to 44 percent. The candidates run neck-and-neck among these voters online, however: 50 percent for Trump, and 49 percent for Clinton.
A previous
Morning Consult study late last year found Trump ran about 6 points better among Republicans and GOP-leaning independent voters in online interviews than he did over the phone.
But that same effect is not carrying over to the same extent in the general election.
The November presidential electorate differs from the smaller universe of primary voters and caucus-goers. And while Trump mobilized working-class white voters in the primaries, those voters are — on the whole — still better-educated and higher-income than the general electorate.
Donald Trump’s soaring spirits
By
SHANE GOLDMACHER
"Seeing a measurable difference in those with higher education and higher income is consistent with our study during the primaries,” Dropp said. “These more ‘elite’ groups seem to be the most apprehensive when it comes to telling others they would vote for Trump. That being said, they do not seem to be affecting overall levels of support.”
While many have looked at the disparity between online and phone polls, there are key differences between those surveys that could drive the variance in results. Instead, the POLITICO/Morning Consult study leaves most elements of the survey constant — including the sampling method.
Respondents all began the survey online. Randomly, some respondents were prompted to call an inbound call center and complete the survey over the phone. Others continued with the online survey.
The study did find some “mode effect” — that is, a difference in how respondents answer based on how they completed the survey — on nonpolitical questions.
“We usually see a social desirability mode effect with questions that seem judgmental,” said Dropp. “In this study in particular, we asked respondents to indicate the satisfaction with their family life. On the phone, 68 percent of respondents said they were very satisfied compared to 57 percent online, an 11-point difference.”
The study was conducted Oct. 27-30, surveying 2,330 registered voters, including 2,075 likely voters. Both the online and phone samples were weighted separately based on age, race/ethnicity, gender, educational attainment and region. Overall results have a margin of error of plus or minus 2 percentage points, and the online and phone results both have margins of error of 3 percentage points.
Read more:
Poll: 'Shy Trump' voters are a mirage
Follow us:
@politico on Twitter |
Politico on Facebook
