Last week we were reminded about how accurate aggregate polling can be. The overall percentage of the popular vote in the Ontario provincial election was bang on to what polls were saying going into the final days of the campaign.

And so it’s no wonder, as we head into a most interesting week ahead, that pent-up, but yet to be released, election speculation is fueled further by recent National polls indicating the pendulum has, on this day and at this particular moment, once again swung in Mr. Harper’s favour.

Rest assured, however, as newsworthy as current National poll results are to a Hill media frustrated by summer-interruptus, that is not what’s in Mr. Harper’s polling strategist’s briefing notes.

National polls are just that, National in scope. A typical sample of a thousand is proportionate to the general population, and the ~3% +/- caveat refers to National data. Drill down to Provincial level data and the margin of error increases significantly, to say nothing of what happens to the margins at CMA or EA levels.

The polls we see published on a weekly basis are paid for by clients willing, or able, to afford National samples of a thousand or so only. Done properly, a thousand surveys is enough to accurately determine the National popular vote, and to feed the Media frenzy .

But a mere thousand surveys can be grossly insufficient in predicting an election outcome when you consider an electoral system does not determine a winner based on the total National vote. Heck no, that would be called proportional representation, and we wouldn’t want that now would we?

Indeed, given the riding-by-riding first-past-the-post electoral system, the only way to predict with a high degree of certainty how many seats each party will be elected in, is to conduct riding studies.

Riding studies are usually conducted on the basis of a few hundred surveys completed over two or three days. Rolling tracking studies are continuous in which moments in time can be captured and compared to any other moment in time. The sum of riding studies, and their rolled-up results, can be different from National level data. More to the point, a riding-by-riding analysis foretells who wins in each riding, and ultimately, which party will form a government.

Riding studies are hugely expensive. To conduct 308 riding studies would cost roughly $1.5 million, by comparison to a National survey which, depending on which pollster you hire, costs between $25-$50 thousand.

For this reason, no one conducts riding studies in all 308 ridings. For one thing few can afford it, but for another, there are at least 100 ridings where there is no need to conduct a poll. Based on regional and historical factors, the outcome in those ridings are a certainty. But for the other 200 or so ridings, the outcome is less certain, and for a hundred or so ridings there is no certainty whatsoever.

And certainly, there is no certainty as provided for by your average national poll on which all eyes seem so focused week-in week-out.

All eyes, save for these:    I am certain of this. That is one client who can easily afford to look at a hundred or more ridings, and I am just as certain he has, several times.

The self-assuredness I believe we will see from Mr. Harper this week won’t be because a riding-by-riding analysis has shown Conservatives are the slam-dunk favourites, but rather it is a near slam-dunk that opposition parties are cannibalizing each other by more than is necessary to allow Mr. Harper’s Party to sneak up the middle.

As NDP and Green’s polling numbers improve, and as Liberal and Bloc numbers deteriorate, the so-called Left is a mish-mash of votes-splitting in which the Conservatives are sure to be the ultimate benefactor.

Mr. Harper is in near certain Majority range not because the oft-Con-favourite Ipsos-Reid poll of 40% is accurate. Far from it. Mr. Harper is, today, gliding toward a majority because of vote-splitting. That, thanks to the power of money, and the polling that money can buy, is Mr. Harper’s carefully guarded little secret.

Nowhere is the well-kept-secret more a factor than in Quebec and Ontario. The Conservative gains in Quebec are already well accounted for. The game plan there is to merely ensure nothing changes.

However in Ontario the battlefield lines have only just been drawn in the provincial election aftermath. The Tory defeat and the reasons for it, tells us plenty about where Ontario is on the socio-political spectrum. The ‘Canadian crossing the road just to get to the middle’ has never been more apropos. Near perfect conditions for further vote splitting, I’d say. And while there is no one on the perceived Right with whom votes can be split, the perceived Left, consisting of the Liberals, NDP, and Greens, do themselves a huge disservice by giving a confused electorate too much to choose from.  

Sooner or later, the Left is going to have to decide which is more important, proving they each can survive and hobble along individually in a Conservative Majority, or the Environment.