The Popular Support Numbers:

Popular support numbers from both SES Research and Ipsos-Reid (the two pollsters who were in the field most often during the campaign) show that the Tories peaked in Ontario in the days immediately following the debate (Ipsos had them as high as 38% and SES had them over 40% in the province). From there the Tories dropped quickly.

Three days later the Tories had fallen to 30% in Ontario, made a modest recovery to 34% (Ipsos), then fell back to between 30 and 32% (SES and Ipsos) by the end of last week. The trend was clear, the Tories looked destined to finish in the very low 30s in Ontario. That is what happened.

Overlay on that, the fact that Environics, the one firm we know was in the field polling yesterday (in selected ridings), was within the margin of error in each of the five ridings they surveyed.

So, the popular support numbers were right.

The Seat Projections:

This election featured more seat projections than any election in Canadian history. This art (and it is an art) used to be the domain of obscure political scientist and their research assistants, but this election, almost every major polling firm came out with a seat projection in the last week of the election. Without exception, they were wrong.

It was, however, these seat projections that garnered the most attention in the media. They were viewed by pundits, reporters and voters as scientific and accurate. After all, the model used by CanWest Global had been accurate within 5 seats for the past 15 years and the Ipsos-Reid model had accurately predicted the last Ontario election.

What was forgotten is a basic tenet of research, extreme situations put any model to the test. This election was an extreme situation with a volatile electorate, close riding races across the country, and a close race nationally.

So, to the pollsters:

Don't beat yourselves up, you got it right. Next time, though, stick to polling and leave the prognosticating to the Tarot Card readers.