PQ leader Pauline Marois campaigns on September 3, 2012.
Credits: MAXIME DELAND/AGENCE QMI
MONTREAL - Claire Durand said pollsters are right to be afraid of predicting the winner of Tuesday's Quebec provincial election.
She should know: She's the secretary-treasurer of the World Association of Public Opinion Research (WAPOR), an international society of experts on polling.
The University of Montreal sociology professor used stats from the exclusive Leger Marketing poll for QMI Agency and her model spat out a Quebec Liberal Party minority government. The same Leger poll, published Sept.2, showed a separatist Parti Quebecois just shy of a majority and the Liberals in third place.
Durand's model ran under the assumption that what she claimed is a worldwide phenomenon also occurred in Quebec: right-leaning voters lied or hid their intentions from pollsters.
The Liberal party is arguably the most right-leaning - particularly in terms of economic issues - of the three major Quebec parties.
"Until 2002, pollsters in Quebec used to give about 60% of the so-called undecided vote to the Liberals," she said.
Quebec pollsters have stopped that practice, she said, adding that her formula of giving the Liberals a high percentage of undecided voters produced accurate results in the provincial elections of 2003 and 2007. Durand said the 2008 Quebec election didn't produce enough polling data to make accurate predictions.
Using the latest Leger results, Durand disproportionately gave the Liberals 50% of the undecided vote and 70% of the undecided non-francophone vote. The result is Liberals with 33.1% compared to the PQ at 29.5%.
She said her model is a better reflection of the traditional hidden Liberal support.
While Durand said the most likely election scenario is a PQ minority government, she too isn't making predictions.
Durand said the goal of her model was to show the Quebec electorate that they need to vote with their conscience and not be too influenced by polling data.
"If polls are going to influence people then I want them to know that statistically, with my models, all three parties are almost at equality (in voting Quebecer voting intentions)."