Taking election polls with a grain of salt

Queen's Park in Toronto

In advance of the leader’s debate tonight, two competing polls have come out in recent days that indicate both PC leader Tim Hudak and Liberal Premier Dalton McGuinty are leading in voter intentions. This means two things: that this race is likely closer than anyone anticipated two months ago, and that polls are worth what you are willing to invest in them, and little more.

That EKOS has the Liberals ahead of the Progressive Conservatives by 34.9 to 31.4 per cent and Abacus Data has the PC’s ahead of the Liberals 37 to 33 per cent means nothing, and here’s why. Polls are like weather forecasts in many ways: at best, they are a plausible prediction of what may or mat not happen (with a margin of error plus or minus 2.3 percentage points, 19 times out of 20) based on reasonable estimations. At worst, they are lies.

Polls and weather forecasts, interestingly enough, are both predictions that contain an acceptance of their own fallibility built right in. We accept that there is a phenomenal likelihood that they will be wrong, and that that’s okay. It’s like an old joke about what other profession but weather forecaster would we accept such levels of inaccuracy.

Yet somehow, polls remain the best measurement we have for voter intention in a lead-up to an election. And so important decisions are based on the results of opinion polls. There has long been a debate as to the extent that polls themselves actually help form public opinion. In other words, if the public sees enough polls indicating that one candidate will beat another, or if they see enough lawn signs in their communities bending one way or another, they too may blow with the wind.

Our society has come to invest quite heavily in polling as a means of understanding the political climate, yet we also inherently understand them to be seriously flawed. Not methodologically, necessarily, but in their very nature. They represent snippets of the population who may or not be educated on the issues, or had any intention of voting. Maybe they were simply at home when their phone number was randomly dialed.

And as Jane Taber points out in the Globe and Mail,

“there has been much controversy about polling during the Ontario election campaign. Darrell Bricker and John Wright of Ipsos Reid recently criticized pollsters for their wildly different polling methods and interpretation and then they took the media to task for publishing them. They argued their colleagues had become “hucksters selling methodological snake oil” and that reporters were promiscuous in their reporting of polls.

The idea behind polling, in its platonic ideal, is to capture an accurate representation of what the public is thinking on any given issue. But the immense possibility for exerting political pressure through the power that society invests in polls has allowed some pollsters to become news-makers and help form public opinion, rather than simply reporters of stated opinions. Some have over-stepped their bounds and, as such, have become little more than “hucksters selling methodological snake oil” to many.

Despite all this, I like polls. And I think the public likes knowing how others are feeling as much as politicians and their handlers like knowing how their candidate is doing, and what more can be done to improve upon the numbers. But at the end of the day, they are simply numbers – they are not votes. And the danger arises when pollsters, and their power to sway the public, themselves become the news.

We forget that, like the weather forecast, polls mean as much as we want them to. And the reality will likely be different.