The latest title in the For Dummies reference books series was released this week: Public Opinion And Polling For Dummies.
Yep. Now you, too, can learn about the ins and outs of market research in 56 easy-to-read pages. No PhD required. Call it DIY polling.
The timing couldn’t be better, or worse, depending on your perspective, after the debacle in Alberta last week that left polling firms across the land with egg on their face. All of them predicted a Wildrose majority. Didn’t happen. Not even close. The Tories ran away with it.
I’ll get to the Forum poll of a few weeks back that showed Mayor Rob Ford’s approval ratings are on the rise.
But first, the Alberta results no one saw coming. Twitter played host to an interesting mano-a-mano featuring Ford’s crack pollster for hire (and former chief of staff), Nick Kouvalis, versus Ipsos Reid CEO Darrell Bricker.
Bricker expressed disgust with the Alberta election polling. A “catastrophe,” he called it.
Kouvalis offered that Bricker shouldn’t throw stones, since Ipsos hasn’t exactly been hitting the nail on the head in its recent elections polls. The truth on that count is a little more complicated.
But speaking of glass houses, Kouvalis shouldn’t talk. The Marketing Research and Intelligence Association (MRIA), the governing body of marketing professionals, is investigating his company, Campaign Research, over that messy business of misleading phone calls in Liberal MP Irwin Cotler’s Montreal riding during the last federal election.
Ipsos didn’t see a Wildrose collapse in the offing either, but the firm has been more honest than most in the biz about its discomfort with the proliferation of one-person polling operations with questionable political connections – and a knack for bending the rules when it comes to public opinion research.
Ipsos’s infamous open letter to journalists during last fall’s Ontario election didn’t name names but warned of “hucksters selling methodological snake oil,” and polls “distorting our democracy” and “confusing voters.”
That entertaining Bricker-versus-Kouvalis pissing match aside, it’s arguable that our political system is being held hostage by polls. Maybe we need to consider some kind of moratorium extending the ban on reporting on polls beyond the current 24 hours before an election. Some industry types say it’s high time.
But the Supreme Court of Canada has already pronounced a fat no on that matter.
In 1998, it struck down a section of the Canada Elections Act banning the publication of public opinion polls in the 72 hours before a federal election.
The court was divided on the question, though. Dissenting judges noted the pervasiveness of polls and their reduction of politics to a “horse race… [reducing] the level of discourse to the lowest common denominator.”
Dissenting justice Charles Gonthier wrote that polls “tend to pre-empt the discussion of issues and short-circuit the democratic process,” and that our reliance on them has become “a substitute for public policy.”
Gonthier wrote that the media’s attention to polling results “tends to distract voters’ attention from substantive issues. The problem becomes more acute when some voters consciously use survey results to make decisions.”
These problems have only gotten worse since 1998 with the growth of robocalling and now-dominant online polls.
Nowadays, anybody can purchase automated polling services (usually from U.S. providers) for a fraction of the cost of hiring traditional polling firms with in-house staff who conduct phone surveys.
Ipsos senior VP John Wright says there are lots of “questionable” polling companies out there, many of which make no secret of the fact that they have a political agenda.
The MRIA has identified the problem as well.
Says executive director Brendan Wycks: “They are not market research companies they are political telemarketing firms. The political advocacy and voter intention work they do couldn’t be further removed from statistical-science-based research.”
Industry pros have two words to describe the work these firms do: “mugging” (marketing under the guise of interviewing) and “sugging” (soliciting under the guise of interviewing).
But polls are only part of the problem. The media’s reporting of them, often absent information contextualizing them, clouds the picture in the public’s mind.
Now to that poll done by Forum on the mayor’s approval rating.
The Globe, Star and Sun all reported what the poll results claimed – that the mayor’s numbers had risen to 47 per cent, compared to 41 per cent in March.
The papers, following standard practice, reported the margin of error in the Forum poll, in this case plus or minus 3.4 per cent based on the sample size of 812 respondents. Which is to say that the poll’s numbers are considered accurate within 3.4 per cent 19 out of 20 times.
But the significance of the margin of error wasn’t explained in the news stories. It rarely is.
The MRIA calls the margin of error of “waning relevance” in assessing polls. Its own guide to reading polls says a margin of error of plus or minus 3.4 per cent really represents a spread of 6.8 percentage points. So the mayor’s approval rating could be as high as the stated 47 per cent or as low as 40 per cent, which would mean his support has flatlined.
Polls are weighted to take demographic characteristics into account, but their random nature means their samples aren’t always representative of the population at large, which can also skew results. The Canadian Press, for example, will not report on national polls with samples of fewer than 1,000 respondents.
Polls helped stop a civil war in Ireland when surveys revealed support for peace talks. They’ve also been described as the lungs of democracy, keeping our leaders on their toes. But increasingly, it seems, they’re more tools to push hidden agendas than to give the masses a say in public policy.
enzom@nowtoronto.com | twitter.com/enzodimatteo
