To the Editor:
If current polls mean anything, then we are headed toward a change of government in the next federal election. It seems Justin Trudeau can do little wrong as the leader of the Liberal Party, despite a series of embarrassing snafus and a less-than-stellar speaking record in the House of Commons. He has, it seems, the Midas touch.
Or does he?
It all depends on how fully you are prepared to place your faith in the research that puts the Liberals in the lead by a full 10 points over the federal Conservatives. Increasingly, that trust in the pollsters seems misplaced. That is not to deny the Liberals' steady surge in popularity, but rather to encourage you not to read too much into the trend.
Political polling in Canada is in the middle of a crisis after forecasts of election outcomes have consistently missed the mark, well beyond their theoretical "margins of error." In Alberta's last provincial election, for example, the opposition Wildrose Party seemed poised to seize government by a landslide - until election day, when the incumbent PCs staged an astonishing comeback. Same thing happened in B.C., where the incumbent Liberals defied the polls and won a majority, even as their obituary was being written.
Pundits explained away the apparent failure of polls in both cases by claiming there was a huge last-minute shift in voter intentions, too late to be captured in the polls. The explanation is plausible, but then so is another one: that the pollsters just got it wrong.
Some pollsters themselves are calling into question our overreliance on the tool, even as the reliability of the polling techniques used is being called into question. Calgary public opinion researcher Brian Singh has spoken openly about this trend, noting there is a rise on "software-based polling methods." By that, he means online polling using Internet panels of self-selected respondents (the net effect is that recruiting is not truly random). He also points to so-called "robocalls", i.e. the computer-based calling system that keeps dialing until it finds someone who will accept the call. That also introduces a potential bias into the results.
"Gone," Singh says, "are the days of excellent response rates to telephone polls."
Many of you already know this. You will have gotten "the phone call" - a call from an auto-dialer that maddeningly makes you wait for several seconds before you hear a human voice. Think about it: If the basis of reliable sampling is random selection, the autodialer is immediately biased - only the most patient, or people with a lot of time to spare, will wait to take that call and actually answer the question.
Of course, if you do talk to a human being, it's notable that many of the employees of these firms seem not to be very well trained at their jobs.
Serious pollsters will tell you that polling is not about watching the horse race, even though that is how it ends up playing out in the news media. In any election, the only "news" on some days is how the numbers are trending. The problem is that a trend is not the same as a fact - one week's surge by any party does not mean it will be in the lead on election day. It's fair to ask, as well, whether voter behaviour whipsaws back and forth, as engaged citizens react to whatever party appears to be in the lead. This fuels the shift to "strategic voting" - for example, voting for a candidate you don't actually like in an attempt to influence the provincial or national outcome.
Equally troubling for pollsters is how the gap between what people say they will do and what they actually end up doing is becoming increasingly disconnected. They may tell a faceless caller they intend to vote NDP, for example, but end up making a different decision at the ballot box.
There's no easy solution for pollsters, but for the public the answer is relatively straightforward. Take any poll with a massive grain of salt, and, when the result seems out of step, ask yourself whether this is a fluke or just bad methodology.
And, finally, there's nothing like having your ear to the ground to get the real story. That means at the doorstep and in the coffee shops, and also what's being said in the blogosphere. Every astute politician today has an advanced social media presence, not just to influence followers, but also to monitor what is being said.
Remember the true purpose of some of those headline-grabbing polls - sheer entertainment. How much their findings resemble reality is an open question.
Doug Firby, Troy Media Corp.
