No matter how you look at the numbers – be they seats, votes or polling results before the vote on Oct. 21 – everyone claims some sort of victory and points out someone else’s loss.
In analyzing the federal election in which everyone seems to have lost something, the only winners may be the pollsters. With the exception of a few outliers, most of the major polling houses did very well in the Canadian 43rd General Election.
Here is a quick look at the score of the final polls of the top five finishers, all taken between the 16th and 20th of October:
All pollsters did a good job, Leger best of all. The one party that tripped up the most polls was the NDP, predictions for whom varied between 16% and 21%. Apart from this, however, none of the top pollsters were further away than one or two points from the actual election result for each party.
While this lends credence to the assumption that pollsters observe each other's' work and "herd" towards the end of an election period, this is not the case. Each pollster has his or her own weighting formulae, which take into account turnout modelling, allocating undecided votes and estimating late shifts. While they are all different, these formulae have the effect of harmonizing towards Election Day, as voters become more sure of their intentions and late trends become clearer.
In Election 43, the online pollsters were the most accurate, but not by much. Random selection methods are also in the top five. It is obvious, however, that online, self-selection polling methodologies can no longer be dismissed out of hand.
So, while the Liberals lost 20 seats and their majority, and the Conservatives lost their chance at forming government, and the NDP lost half their caucus, the Greens lost the battle of the millennium, and the PPC lost their leader, Canada's pollsters – the best of them at least – won the day with uncannily accurate prescience about the outcome of our quadrennial bun fight.
© Marketing Research and Intelligence Association