Skip to main content

John Wright, shown in 2011, is senior vice-president of Ipsos Global Public Affairs.LIAM RICHARDS/The Globe and Mail

A spate of spectacularly inaccurate polls in several provincial elections and Monday's federal by-elections has one of Canada's leading pollsters worried that the day is fast approaching when public opinion surveys will be banned during campaigns.

John Wright, senior vice-president of Ipsos Global Public Affairs, said in his view shoddy polling methods by some companies and the media's undiscerning appetite for horse-race numbers – no matter how dubious – are dragging the industry's reputation through the mud.

He wants the market research industry to crack down on pollsters who don't meet minimum standards. And he wants media outlets to be far more selective about the polls they publish, rejecting surveys from companies that refuse to fully disclose all their weighted and unweighted data.

If that doesn't happen soon, he predicted there'll be a move to ban the publication of polls during campaigns.

"I think at some point, if this continues, there'll be a real question raised whether these sort of things should even be contemplated being done during campaigns," Wright said in an interview Tuesday.

Most polls were significantly off the mark in recent provincial elections in Quebec, Alberta and British Columbia.

Then on the eve of Monday's four by-elections, a Forum Research automated phone survey was published giving the Liberals in Manitoba's Brandon-Souris a 29-point lead. In the end, the Conservatives eked out a win with 44.1 per cent of the vote to 42.7 for the Liberals.

The Winnipeg Free Press questioned the poll's reliability, reporting that a number of constituents had been called as many as six times by Forum. It also carried the results, as did other media outlets.

If voters were in fact called repeatedly for the Forum poll, Wright said the survey sample would not be representative of the riding's population.

In close contests like Brandon-Souris, where fewer than 400 votes separated the Conservatives and Liberals, Wright said it's especially important to ensure polls are accurate and properly conducted.

"If one vote was influenced by a bogus piece of work, that's one vote too many," he said.

Forum denies anyone was called more than once for the same survey.

In 1998, Wright said the Supreme Court ruled there's no need to ban polls during campaigns because the industry and the media would police themselves sufficiently to ensure voters were not being misled by bad surveys.

"I think that's a system that's completely broken now," he said, adding that he's no longer sure the top court would rule the same way, given recent experience.

"Will there be banning of polls? Who knows," Wright said.

"But certainly when you get enough of these things where you can't tell whether there's any merit or science or believability to it all and you get players who seem to disregard the very fundamentals of doing it the right way, then you wonder how far off it can really be."

Forum's Lorne Bozinoff defended his company's record, noting that it accurately pegged the results in two other by-elections Monday: Toronto Centre, Bourassa in Montreal.

"We know [automated phone polls] can work, there's just no question about that. The question is why were they off in Brandon?" he said in an interview.

Bozinoff said "there's no way" anyone was called more than once per survey, although some people may have been called for each of the three separate polls the company conducted on the Friday, Saturday and Sunday before the vote.

He speculated that the difference between the final Brandon poll and the actual by-election outcome may have been that the Conservatives had a better "get out the vote" ground game than the Liberals. As well, he said some constituents who were angry about the perception of a fixed Tory nomination may have found they just couldn't bring themselves to vote Liberal once they got into the ballot box.

Whatever the reason, the Liberals' supposed 29-point lead didn't materialize – a point used by Conservatives to deflect attention from their party's disappointing by-election results.

"The pundits were saying that we were some 30 points behind in Brandon-Souris … but our candidate did very well and we came out with a victory," said Paul Calandra, parliamentary secretary to Prime Minister Stephen Harper.

The Tories hung on to Brandon and another Manitoba riding, Provencher – both long-time Conservative bastions. But their share of the vote plunged as much as 20 points in Manitoba and almost disappeared entirely in the Toronto and Montreal ridings.

The NDP gained strength slightly only in Toronto Centre, lost a bit of ground in Bourassa and sank into a distant third in the Manitoba ridings, after coming a respectable second in 2011.

The Liberal Party alone increased its vote share in all four ridings, hanging on to the traditional Grit seats of Toronto Centre and Bourassa, despite an all-out challenge from the NDP, and making huge gains in the two Manitoba ridings.

Interact with The Globe