Wednesday, April 23, 2014

An exceptional poll

TC does not ordinarily take note of just one poll unless there is something unusual or special about it.  Well there is a highly unusual result in the latest Angus Reid poll released on April 22.

Ordinarily, the numbers go up and down routinely, but the data within one poll is typically within shouting distance of the previous survey and is not worthy of note. Unless an election campaign is underway it is unusual for public opinion to depart sharply from previous patterns, particularly with regard to a single poll.

However, the regional data for Ontario in the Reid poll have produced a result that is quite different from the patterns in other regions in the same poll, and strikingly different from the Ontario data in other national polls from 2014. In particular, the Liberals have averaged a five point lead on the Conservatives in 2014 but in this poll they have a seventeen point deficit.

Another oddity: the poll comes close to replicating actual vote shares in Ontario in 2011.

When one turns the numbers from this survey into a national seat projection one gets another unusual result: the NDP is in second place in seats, a status they haven't enjoyed since the rise of Justin Trudeau (one reason for this is a relatively high poll share for the NDP in Quebec). The boost to Conservative fortunes in the survey from a very good result in Ontario is still not enough, however, to get them close to a majority.

For me to take this survey seriously, it would be necessary to see something like this result replicated in other polls. For the moment I would have to say this result doesn't look right. One should not forget like many other Canadian polls, this is an online survey, a methodology I continue to regard as experimental.