JEDDAH: Much of the world woke up last Wednesday to the surprise that Donald Trump is to become the 45th president of the United States.

 And if it was a surprise to the public, it must have been a shock to global polling companies — most of which forecast Hillary Clinton would win the race.

 On the day before the election, all but two of the 21 polls tracked by RealClearPolitics were predicting a win for Clinton, with one by NBC News/SM forecasting that the candidate had a 7-point lead. Bloomberg, Reuters/Ipsos and even Fox News all forecast that the Democratic nominee was ahead.

They were, of course, wrong. And as was the case with other recent polls that proved off the mark — including those around the 2015 UK general election and this year’s Brexit referendum — the spotlight now falls on the pollsters.

As Sam Wang, who is behind the Princeton Election Consortium blog, put it in quotes to various media: “This failure of polling was the largest in a presidential election year in decades. It’s going to take some time to sort out what went wrong.”

 

So what did go wrong? 

 Stephan Shakespeare, the Chief Executive of polling company YouGov, acknowledged that most polls got the outcome of the US election wrong. His own company’s final estimate put Clinton in the lead by about 3.8 percent.

But he said that the final margin of error was not as big as many media reports suggest. 

“The polling consistently underestimated Trump, but not by nearly as much as people are saying,” Shakespeare told Arab News. “I think pundits are massively exaggerating what’s been happening.”

The polling expert pointed to the result of the "popular vote" — the number of votes received by each candidate, irrespective of the electoral college. Those votes are still being counted, but Clinton is expected to win by a margin of more than two million.

“While we should have done better, you do need to remember that Hillary Clinton won the popular vote, and will end up with about a 1.5 percent lead,” said Shakespeare.

“Polls do not attempt and cannot attempt to predict the particular structures of the electoral college. What we do is predict the popular vote. By the time the final votes are counted, we will have been just over 2 [percentage points] out, which is not good, but not awful.

“This was a case where a small difference in the popular vote… spread between different states, created a big difference in the electoral college.”

Despite that, the fact remains that the polls were out. So why was the public support for Trump so underestimated?

Dr. Steven Anderson, Chair of the Republicans Overseas in the UAE, told Arab News just before the US election day that many voters would quietly vote for Trump without publicly declaring it. 

“A lot of people I’ve spoken with… are not interested in pollsters. There’s a vast, silent group of people who will stand up and vote for Trump,” he said at the time.

“Quite frankly, there’s a lot that don’t want to admit voting for Trump. So they’ll publicly say one thing — and then when they’re in the voting booth they’ll say ‘yeah I want to vote for this guy’.”

Shakespeare said he did not see this as influencing YouGov’s poll results, which he said are “online and very anonymous”.

But he said that there was an “unusually large” number of undecided voters in the recent US election, which helped sway the poll results.

“They were probably people — we can only guess at the moment — who would be put off by Trump, and didn’t think of themselves as Trump voters. But when it came to voting, decided that they preferred him,” Shakespeare said.

So is there a future for polling in what seems like an increasingly polarized world?

Shakespeare, understandably, sees it as having a key role. He pointed to polls like the Scottish referendum and London mayoral election, the result of which he said YouGov forecast “precisely right”. But he acknowledges that the way polls are conducted may have to be tweaked.

“Up until now, in all of the last elections put together, we had an average error of about 1 percent. In the last couple of big public elections, that went to 2 percent. And that’s not acceptable. And clearly we are looking to adjust techniques a little bit.”

Changes or not, expect the media airwaves to be buzzing with opinion poll findings in elections to come. Because — unless all election results are to be a surprise — there is simply no alternative.

“Polling is not a technique that can be reasonably expected to hit the bulls eye every time,” said Shakespeare. “But it’s a hell of a lot better than mere guesswork.”