What makes a pollster blush?

Canadians knew the three-way race couldn’t last forever. And indeed, with two weeks until Election Day, a party has begun to pull into the lead.

Of course, in what has become a signature trait of the 2015 election, polls differ on exactly which party.

A Nanos Research survey shows the Liberals gaining a dramatic lead at the expense of the NDP. The Liberals are at 35 per cent support, the Conservatives at 31 per cent and the NDP are trailing at 24 per cent.

One from EKOS shows the Conservatives maintaining a commanding first place (33 per cent), which they’ve apparently held since mid-September.

“Some pollsters blush at claiming that their margin of error is within three per cent, 19 times out of 20, others don’t,” said Barry Kay, a polling expert at Wilfrid Laurier University. “But it’s all bulls---.” In the time of John Diefenbaker and Lester Pearson, clipboardwielding pollsters would go door to door, asking Canadians their voting intentions come election day.

But the modern era has not been kind to pollsters.

A Golden Age of telemarketing and telephone scams has conditioned Canadians to reject any call with even a whiff of an 800 number.

Despite popular belief, polling firms are indeed able to call cellphones, but the rejection rate is even higher due to owners fretting about using up minutes.

Roll it all together, and Canadian opinion polls are frequently decided by that bizarre subset of Canadians who pick up unfamiliar calls — and don’t immediately hang up when they hear a recording or a headset-wearing pollster on the other end.

Pollsters try to address the problem by employing a top-secret formula that weighs some responses more heavily than others — but there’s only so much that can be fixed with math when 90 per cent of cellphone users aren’t picking up.

“I’ve been doing this since 1980, and after 2000 the numbers started getting worse,” said Kay.

The margins of error are increasing, and with the first threeway race in living memory, slight changes can have noticeable effects on the published results.

Adding to that are the differences in polling methods, and it’s not surprising that two competing polls would show either the Liberals or the Conservatives in the lead.

EKOS uses robocalls. Nanos, meanwhile, uses live interviewers calling both cellphones and land lines.

But polls have always been an inexact science — akin to tailoring one’s opinion on eggs to the latest health study. Any number of studies will claim that eggs are healthy or deadly, but the consensus opinion is that they’re relatively fine in moderation.

Similarly, by looking at the aggregate data, analysts are able to pull some truths out of the sheer tonnage of 2015 opinion polling data.

“The aggregate trends over time tell you that the NDP is in sustained decline, and the Liberals are in a sustained rise and so are the Conservatives,” said David Moscrop, a political scientist at the University of British Columbia.

And when those numbers are plugged into seat projections, it’s looking like absolutely nobody will be coming close to a majority.

“When you’re a professor, everybody wants to know what you think will happen in the election,” said Zachary Spicer, an associate professor in the department of political science at Brock University.

And his well-rehearsed answer would line up with many people reading the tea leaves of Election 2015.

“It looks like the Liberals are going to have more seats than the NDP, and I would hazard a guess that the prime minister will probably be Harper,” he said.

In this regular feature until Election Day, writers capture a telling moment in time from the 2015 campaign trail.