Take another look at polls

The provincial byelection in Nanaimo last month was a lifesaver for the NDP. Had the party’s candidate lost, the seat tally in the legislature would have been tied and a general election might have followed. The B.C. Liberals did manage to increase their share of the vote. But it wasn’t enough to overcome the huge majority won by the NDP in 2017.

The question that remains, though, is why the Liberals lost by nine points when a poll shortly before the election had them 12 points ahead. That’s a 21-point swing in just a few days.

One possibility is that the survey methodology was flawed. There do appear to have been oddities in how the sampling was constructed.

Another is that byelections are notoriously difficult to predict because voter turnout is usually low. Three thousand fewer ballots were cast than in the 2017 general election.

Then again, there have been numerous occasions over the past few years when polling companies completely misread the public’s mind. During the provincial election of 2013, we were assured the NDP was comfortably ahead. In the event, the Liberals won easily.

And on the eve of the 2016 presidential election in the U.S., almost all of the surveys had the Democratic candidate, Hillary Clinton, leading her opponent, Donald Trump, by a wide margin. Yet Trump won in the electoral college.

But another possibility exists. Perhaps the Nanaimo poll altered the outcome by changing voter intentions.

On this theory, before the poll was published, some NDP supporters were taking a win for granted, and told the survey firm they were voting Green. That boosted the Liberal numbers. It was only when the results were made public that they returned to the fold.

There is certainly a basis for that speculation. The Green tally collapsed from 20 per cent in 2017, to just seven per cent.

It’s possible some of those Green voters switched to the Liberal candidate. There was certainly a feeling that Finance Minister Carole James had blundered badly over the housing speculation tax.

Halfway through the campaign, James announced that 1.6 million households would have to prove, in writing, that they were not speculators. That misstep caused widespread anger.

Yet the fact remains the NDP won an election it looked almost certain to lose, according to the poll. Did the release of that poll play a part?

The question is worth asking, because in recent years, governments across the country have taken steps to shield elections from outside influence. Financial contributions from unions and corporations have either been banned or greatly scaled back.

The amount individuals can give to a campaign has also been reduced. And limits have been set on the sum parties can spend on advertising. The purpose is to prevent money from playing too large a role in what is supposed to be a contest between policies and vision.

Yet it does seem possible that polls can also influence an outcome, particularly if they are released within a few days of the vote.

Put another way, opinion surveys not only provide information, they might also push voters in a particular direction. Is that something we want?

There is, of course, a freedom-of-speech issue. Why shouldn’t polling companies have their say?

And media outlets are already banned from publishing poll results on election day. That probably suffices.

Yet care is needed in how we deal with material of this kind. If we don’t want money to sway election results unduly, perhaps we should be equally suspicious of placing undue weight on polls. The outcome in Nanaimo does suggest that voters might have been influenced in this manner.

The lesson is that more thought is needed about the role of opinion surveys in elections.

The question that remains, though, is why the Liberals lost by nine points when a poll shortly before the election had them 12 points ahead.