Election surveys: Don’t always believe what you read and hear

There’s a backyard chicken activist. A homeless guy. A fistful of current and former politicians. A university professor. A millionaire businessman. And a news anchor.

In all more than a dozen people are jockeying to be Calgary’s 36th mayor. With such an eclectic mix of arguably qualified candidates to choose from it should be Calgary’s most interesting mayoralty race since Dave Bronconnier’s upset-win nearly a decade ago.

Yet less than half of those candidates are considered “serious contenders” and according to recently commissioned polls there are a few candidates who truly can win the city’s top seat including a former oilman a three-term alderman and a self-proclaimed “fiscal conservative” with a TV-ready smile.

But poll-believers beware: None of the recent surveys included more than 500 people — arguably about half the number considered credibly accurate — in recent “polls.”

One survey has long-perceived heir apparent Ald. Ric McIver coasting to an easy victory; another has him hanging by a thread. An online survey has Wayne Stewart a relative unknown businessman with a huge lead.

Or it could be a photo finish between McIver and recently retired TV anchor Barb Higgins who according to yet another poll leap-frogged over everyone to take the lead the day after she announced her mayoralty bid.

But with two months until the election it’s far-fetched declaring any candidate a frontrunner much less a winner says Mark Pickup assistant political science professor at B.C.’s Simon Fraser University.

“It’s difficult making such pronouncements so far out from an election” says Pickup. “We’ve seen this far too often on a national level where two months out from an election media will pronounce a winner… and often that doesn’t happen.”

Complicating matters is that the methodology and credibility of the surveys varies almost as much as the results. Timing of the polls also plays a huge role.

NRG Research’s poll which contacted 500 Calgarians by phone within 24 after Higgins threw her hat into the ring put the former CTV announcer in front of the pack — at 16 per cent and McIver close behind at 15 per cent while 51 per cent of voters were undecided.

Problem with that survey says Pickup is that it has a “very large measurement error” because “if one is to get a representative sample the usual procedure is to be interviewing in advance over multiple nights” he adds.

Alberta Courage Network’s online poll which gave mayoralty candidate Wayne Stewart 34-per-cent support had problems of its own: an extremely low sample size of 206 respondents and an ideologically biased database.

“I would admit that there was some skewing in this poll” says Kelly Ernst president of ACN. “You’re probably going to see one side of the spectrum in that poll — we fully admit to that.”

Ivrnet Inc.’s automated phone survey during three days in mid-July suggested McIver had a 31-point lead over his closest opponent. But again the methodology is not without flaws says a former Ivrnet researcher who designed the firm’s survey technique.

“Ivrnet polls randomly” says Neil Mackie Alberta Liberals’ communications director. “Everyone else polls randomly but they build proper demographic models…. It’s what’s called a ‘representative sample’” he says. “If you’re polling during the day theoretically you could get all 92-year-old grandmas at home that day.”

With such low sample sizes the scientific credibility of the surveys could be justifiably questioned. But Andrews Enns vice-president of NRG Research says size isn’t everything when it comes to polling.

“The scientific aspect comes into it in terms of your methodological rigours that go into it how you polled your sample the randomization” he says. “I’d be surprised to see anyone in the city do 1000. In urban centres anywhere from 400 to 600 is kind of your standard sample size.”

And as a mid-summer poll it really just provides an “interesting snapshot” quite early into the campaign says Enns. “Quite frankly a lot of that campaign won’t happen until after Labour Day” he says.

Large or small scientific or “snapshot” published poll results can affect voters’ decisions campaign volunteers’ morale and hamper financial contributions.

“Voters use it as a check for their instincts and to see if someone is credible” says Mackie. “In a lot of ways you could say that Barb Higgins is not credible at all because she has no administrative experience no governing experience nothing.

“But she walks in and ties with McIver in the polls and that’s totally based on name recognition" he says. "What that does is that it creates that impression in the public that she is credible. And it then creates the impression that other people are not credible because they are scoring within the margin of error.”

Mackie and Pickup say this can be problematic for lesser-known candidates who struggle to stay in the spotlight attract volunteers and vital financial contributions. “Polls drive money and donations and money drives all in politics” says Mackie. “People want to be with the winner.”

Email: thowell@ffwd.greatwest.ca