UPDATED: Leading into Tuesday’s election, only one national poll — the USC Dornsife/Los Angeles Times “Daybreak” poll — consistently showed Donald Trump with a lead in the U.S. presidential race.
After Trump’s surprise victory, one might expect the creators of the poll to be crowing in vindication. They aren’t.
Instead, poll adviser Dan Schnur said the electorate should learn lessons from the poll — principally that it is important to measure the intensity of voters’ commitments, along with which way they intend to cast their ballots.
“People should look at a poll like this one as a supplement to traditional polling, not a replacement for traditional polling,” Schnur said Wednesday morning. “Hillary Clinton had a larger number of supporters, but Trump backers were more intense and more fervent in their support and that appears to be exactly what we observed on Election Night. The white male working class voters who had not voted in particularly high numbers in last few elections turned out heavily for Trump.”
The final USC Dornsife/L.A. Times Poll before the Nov. 8 vote showed Trump ahead by three percentage points. It was alone among 10 national surveys in showing Trump with a lead. The Real Clear Politics average of all major polls showed Clinton with a 3.5% lead nationally.
With Clinton and Trump essentially finishing in a dead heat nationally (Clinton ahead fractionally, with most ballots counted) the final outcome ended up being somewhere in between the findings of the USC survey and the others.
“So [the USC poll] was no more precise than the other polls,” said Schnur, a long-time political adviser to Republican office holders and now director of USC’s Jesse M. Unruh Institute of Politics.
In an op-ed piece in the L.A. Times the day before the election, Schnur noted the inaccuracies that had cropped up in many polls recently, most in the U.S. and abroad. He said that showed the need for exploring alternative polling methods.
The USC Dornsife/L.A. Times poll asked respondents, sampled online, to rate their level of commitment to their candidate on a scale from 1 to 100. That measure helped reveal the greater fervor among Trump voters, Schnur said.
While most major polls are based on telephone surveys, the USC / L.A. Times poll relied on online sampling. There is an argument among some pollsters that computer-based sampling limits the “social desirability” factor from a sample — the tendency of respondents to tell pollsters what they think they want to hear. The USC poll may have limited that tendency and therefore detected a bit of the “hidden” Trump vote, but this tendency was not a huge determinant in the poll’s findings, Schnur said.
Regardless, the poll’s standing as an outlier, opposed by multiple polls showing Clinton in the lead, meant that the USC survey took a battering. Schnur noted that the poll was called “absurd,” “reckless,” and “useless.”
Schnur said that in the future he hopes such polls will be measured for their methodologies and not demonized for the outcomes they project.
“Scientific advancement comes as a result of inquiry and experimentation, both successful and otherwise,” Schnur wrote in the Times. “The most appropriate response to the Daybreak poll should not be to dismiss it because it reveals something unfamiliar and potentially disagreeable, but rather to study it and learn from it so the science of public opinion research can move forward.”