If you didn’t know September was just around the corner then pollsters coming out of the woodwork should have been a clue. A second clue is Costco has enough Christmas stuff on display to choke a reindeer.
Yesterday morning the Financial Post headline had a Darrel Bricker Ipsos poll with the Conservatives nearing 40% and the Liberals lagging behind under 30%. Better yet, if you are of that persuasion, Ipsos claims that in “vote-rich” Ontario the Conservatives lead the Liberals by a whopping 43% to 31% margin. Really?
In recent weeks my gut told me the ever-present Stephen Harper was pulling ahead of the absent professor, part Deux. But 40%? Hang tight, I said, and lets see what Decima, Nanos, or Strategic might have to say.
Sure enough not more than a few hours later the Globe an Mail headlined a more tempered Harris-Decima poll showing the two front runners still in a relative dead heat in the 30% range. If that isn’t contradictory enough, take a look at what the competing pollsters have to say, on the same day, about Ontario. Ipsos has Ontario at 43% to 31% for the Conservatives, whereas Decima surveys nearly the opposite with 40% for the Liberals and 34% for the Conservatives.
So what gives? Who’s right?
Thankfully it doesn’t happen often that polling numbers are so far apart. But occasionally it does. While the “margin of error” caveat is always a possible explanation, in this case the contradictions far exceed the allowable margin. So too, I would hope, are the Ipsos results beyond whatever methodological bias may be imposed by the right-leaning penchant at the top.
This is not the first time Ipsos has been giddy and/or alone to report the Conservatives outflanking the Liberals by such a wide and unsustained margin. The wording and positioning of questions in a survey are a critical form of art in the polling business which even with the best of unbiased and unpartisan intention can sometimes inadvertently creep its way back into the survey design. An occupational hazard some might argue.
Don’t get me wrong, the Ipsos poll may not be an outlier. It may indeed be ahead of the curve picking up on a trend others will confirm in the days and weeks to come. Especially if the “me too” syndrome plays out upon undecided voters who, like sheep, sometimes like to align themselves with who they perceive – or who others have strategically positioned – as the winner.
But until then there are two other significant methodological differences, other than questionnaire design, that must be taken into account before reaching any conclusions about the contradiction in these two polls. One is sample size, but the other, and perhaps more importantly, is timing.
The Ipsos poll was based on a sample of 1000 respondents whereas Decima polled 2,000 Canadians. The margin of error between a thousand (3.3%) or two (2.2%) is only 1.1% which hardly explains the National difference in the numbers let alone the complete opposite findings in Ontario.
Which brings us to timing and the most likely explanation for why two credible and quality organizations are such polar opposites. Pun intended.
The Ipsos poll was, as we say in the business, crammed into field over 3 nights from the 18th to the 20th of August, fresh off the heals and seals of Mr. Harper’s excursion to the north. By contrast, the Decima poll was paced over an eleven day period from the 13th to 23rd. So who’s right?
Likely they both are. At least in relative and directional terms, but not in absolute terms. It’s possible that over a protracted Decima period there can be wild Ipsos 2 or 3 day swings. This is what Harris-Decima’s rolling tracking methodology is so good at averaging out but could, if one only looks at bi-weekly or monthly averages, miss out on so called turning and/or tipping points which the Ipsos poll could very well have perfectly timed. But a 10 point difference in opposite directions in Ontario has to be more than just about good or bad timing. Something isn’t right in the land of polling and where, in the next election, a majority or a minority will be determined. Ontarians are fickle but not that fickle. Time will tell.
Until this latest plethora of polling results I was firmly on the side that an election this Fall was inevitable. Mr. Ignatieff couldn’t repeat the month of June and again hope to get away with it. By the same token forcing another election it isn’t clear Canadians want has its own set of risks. If the Ipsos numbers are true and hold, Mr. Ignatieff may end up looking more-and-more like Mr. Dion than he’d prefer by having to wait until 2010, by which time the economy may further recover to only further benefit Mr. Harper.
Unfortunately for Mr. Ignatieff it appears he is caught in a game of Russian roulette where he has to take the first shot. However, ordinarily in the ill-fated game not all six chambers are loaded.