Published Oct. 28, 2015, in The Waterloo Region Record.
One observation to be drawn from last week's election results is that the pollsters' estimates of the national vote were actually very good.
What wasn't quite as accurate were some of the regional splits contained within them. Most notably, perhaps, in Quebec, where the late pre-election polls suggested a one-per-cent Liberal lead over the NDP, while the actual result was a 10-per-cent differential there.
Similarly in Atlantic Canada, the aggregated pre-election polls showed a substantial 30-per-cent Liberal lead over the Conservatives and New Democrats, when the actual results provided an even more dramatic 40-per-cent lead. Lesser, but significant disparities could also be found in the Western regions.
This is a major consideration in explaining the relatively poor performance of the seat projections, which almost uniformly pointed to a Liberal minority government. The Laurier Institute for the Study of Public Opinion and Policy (LISPOP) numbers suggested 140 Liberal seats, and was typical of most other projection websites. The discrepancies in Quebec and the Atlantic region alone account for two thirds of the projection's underestimate of the actual Liberal seat count.