Since Donald Trump won the presidency Tuesday night, many headlines have declared his victory “stunning” or an “upset.”
The disconnect between national polls that overwhelmingly showed Hillary Clinton with a lead and the final result that delivered Trump a decisive electoral win, if not the popular vote, surprised even veteran political observers.
At a conference Thursday hosted by Capitol Weekly, California pollsters offered theories about how the surveys got it wrong: an enthusiasm gap for Clinton that depressed Democratic turnout, a herd mentality that discouraged people from seriously considering the few polls that showed Trump ahead, not enough polling of battleground states like Michigan that many assumed were solid blue.
“We had a very volatile election, two deeply unpopular candidates,” political strategist Rose Kapolczynski said. “Some voters could have been reluctantly in Hillary’s camp and in the end decided to either switch or just not vote at all.”
Digital Access for only $0.99
For the most comprehensive local coverage, subscribe today.
She said there was a “massive and collective” error in the turnout models that pollsters used to weight their results.
“Those are judgment calls that pollsters make,” she said. “There’s always going to be judgment calls.”
Mark DiCamillo, director of the Field Poll, pointed out that Clinton appears to have won female voters by about 12 percentage points, far less than the 20 points or more she led by among women in pre-election polls.
He compared it to the so-called “Bradley Effect,” the theory that many Democratic voters were reluctant to tell pollsters they didn’t support Tom Bradley, a black Democrat, in his 1982 gubernatorial bid, resulting in his narrow loss despite holding a significant lead in polls.
Some Trump voters may have been unwilling to publicly express their support, DiCamillo said. “This may be a case where social desirability effects were in play.”
Pollsters now face questions over how to fix their methods going forward. Some have begun calling cell phones instead of landlines, a more expensive and difficult process, or using internet panels, a strategy criticized by others for drawing a more self-selecting group of respondents.
“We have a serious issue with people not answering their phones,” Jonathan Brown, president of Sextant Strategies & Research, said. He suggested that the voice of Trump supporters might have been missed if conservative voters, more distrustful of and angry at the media and other institutions that rely on polling, were less willing to take the surveys.
“If those kinds of peoples don’t respond to polls,” Brown said, “they’re not going to be represented in polls.”