Final official June 28 2004 election results

Elections Canada has now published the final official results of the June 28 2004 federal election. This has allowed me to perform a little experiment. We all know the actual outcome on voting day differed significantly from the final pre-election polls in terms of the vote percentages. Suppose those last polls had accurately captured the final outcome. What might the distribution of seats have looked like? Using the final results as the base, I have calculated the seat outcome that would hypothetically have occurred.

It is first worth pointing out that almost all of the difference between the last set of polls and the final vote shares was in Ontario. The final poll in Quebec by Léger Marketing, for example, was almost dead on. I have used for the calculation the Léger poll in Quebec, and a weighted average of the final four national polls (Ekos, Ipsos-Reid, Compas & a national Léger Marketing survey, which differed from the final Léger poll in Quebec both in its numbers and the fact that it was completed a day earlier) . I can justify using the last Léger poll in Quebec because there was a trend towards the Liberals during the last week in Quebec, which the final Léger poll appeared to capture.

The result that we might have seen, if opinion had not moved in the last few days (to repeat, almost entirely in Ontario), was: Liberal – 114, Conservative – 109, BQ - 54 NDP – 30, and Other – 1.

The results by province are almost identical in the projection to the actual in Atlantic Canada, Quebec, Manitoba, Alberta and B.C. The calculation shows the Conservatives losing 16 seats in Ontario, offset by gaining 6 in Saskatchewan, while the NDP wound up losing 7 seats in Ontario and another 4 in Saskatchewan. I think the differences in Saskatchewan can best be explained by polling error rather than late movement, which I think clearly accounts for what happened in Ontario.