Tuesday, November 8, 2011

The stakes of the Bonaventure by-election

The last by-election in Quebec saved Pauline Marois’ leadership of the Parti Québécois. Will the next one end it?

The September resignation of Nathalie Normandeau, cabinet minister in Jean Charest’s Liberal government, has triggered a by-election in the riding of Bonaventure for December 5. 

The riding on the Gaspé Peninsula in eastern Quebec had been held by Ms. Normandeau since 1998, and aside from a brief PQ period between 1994 and 1998, has been a Liberal fortress since the 1950s.

In the 2008 election, Ms. Normandeau won the riding with 64 per cent of the vote. The PQ finished second with 29 per cent.

By any measure, this is a riding that the Liberals should be able to win easily. They have enlisted a local mayor, Damien Arsenault, to run as their candidate against the PQ’s Sylvain Roy.

But the last by-election in Quebec was also supposed to be a Liberal lock.

You can read the rest of the article on The Huffington Post website here.

The election in Saskatchewan last night was not much of a surprise. In the end, the inclusion of the last Forum Research poll suggested a closer race than was the case. Without it, the 49-9 result would have been closer to the 44-14 projection of Sunday and well within the seat ranges that were determined at that stage.

Overall, only six ridings were called incorrectly giving the projection an 89.7% accuracy rating, and the vote projection was only off by 1.2% per party, making it the second best of the five provincial elections called this year. A full post-mortem will be completed soon.

What was most interesting, however, was that the projection I made using Forum's regional results would have been very close to the mark. I had it at 47 seats for the Saskatchewan Party and 11 for the NDP.

This has confirmed to me the need for a regional model going forward in every province. I have a system in mind for incorporating both regional and province-wide data into such a model, and it should even be able to take into account riding polls and sub-regional data when available. I'll construct these models as the provincial elections that are scheduled for the next few years come along, and hopefully by 2015 I will have a very precise model in place for the federal election. I'd like to have that model calibrated to the sub-regional level as well.

Alberta and Quebec are the next ones on the docket, so I will be focusing on them going forward. I will also be taking a look at the leadership conventions for the Bloc Québécois and New Democrats, with something special in mind. Look for that in the coming weeks.

This fall provincial election season had its ups and downs. At the very basic level, three calls need to be made for any election: who will form the government, who will form the opposition, and whether it will be a majority or minority legislature. That means 15 calls needed to be made across the five provincial elections this fall, and ThreeHundredEight made 13 correct calls out of those 15 (the minority in Ontario and the Liberal opposition in Newfoundland and Labrador eluded me).

Of the 297 ridings at stake in the five elections, 259 were called correctly for an accuracy rating of 87.2%. That is a great improvement over the 76% in the federal election.

This site has always been an on-going project, so improvements will be made and hopefully in the next series of elections the accuracy will be upped yet another notch. Each election is a new learning experience!