No polling error of Albertan proportions occurred last night. The polls called the Parti Québécois as the eventual winner and the party did emerge on top.
But the errors that did occur made the resilience of the Liberals a surprise. Instead of the predicted strong minority or weak majority for the PQ we have an almost even split between the PQ and the Liberals in the National Assembly.
Léger Marketing ended up with the least amount of error in their final poll. For every party except the Liberals, the firm's final poll for the Journal de Montréal was off by no more than 1.1 percentage points. But support for the Liberals was underestimated by more than four points, and as a result the Liberals finished a strong second instead of a close third.
The polls from Léger and CROP (for La Presse) that pegged PQ support at 32 or 33 per cent from the mid-point of the campaign turned out to be on the money, with the party only marginally under-performing expectations at 31.9 per cent .
The polls from Forum Research for the National Post and EKOS Research (reported on Twitter by the president of the firm) that estimated PQ support at 36 per cent on the eve of the vote turned out to be over-estimating the party's pull by a fair amount. If the PQ had taken 36 per cent of the vote they would have easily won a majority government.
Support for François Legault's Coalition Avenir Québec was forecast more accurately. The party took 27.1 per cent of the vote, just under the 28 per cent predicted by Léger and CROP and over the 25 per cent of Forum and EKOS. Legault was banking on a last minute swing in his favour. That did not occur, and without a strong organization on the ground he was at a severe disadvantage compared with his two main opponents.
Québec Solidaire took home six per cent of the vote, which was what Forum and Léger expected. CROP, and especially EKOS, over-estimated the small party's chances by a fair bit.
But Liberal support (31.2 per cent) were missed across the board. Forum, which saw the Liberals sitting at 29 per cent on Monday night, was the closest but still came up short by more than 2.2 points. Coupled with the overestimation of the PQ, the final Forum poll did not reflect the close race that the election turned out to be. Léger had the Liberals at 27 per cent and behind the CAQ by a point while CROP had them at 26 per cent and trailing Legault by two points. EKOS, with the Liberals at 23 per cent, severely underestimated the party's ability to get their supporters to the voting booths.
And that was what appears to have made the difference. The PQ, CAQ and QS generally performed as expected, with their vote haul within the margin of error of the final polls from the Quebec-based pollsters with long histories in the province. The Liberals, however, performed well beyond expectations.
Does that mean the polls were wrong? Voting intentions do not always equal voting behaviour. People who said they would vote Liberal might have turned out in greater numbers than those who said they would vote for other parties. Perhaps the social stigma of supporting a party perceived to have a bad reputation kept people from revealing their true intentions. We do not know, but in his final act Jean Charest proved that he can still pull a rabbit out of his hat.