Friday, October 3, 2008

First projection

Ok friends, Romans and countrymen, the long awaited projection is here.

I explained the methodology for obtaining base numbers yesterday.

Before getting into the projection, I'll explain the methodology for including polls:
  • Only one poll per polling company will be used at any time;
  • A poll must break down into at least one of following regions to be used: Ontario, Quebec, Atlantic (NB, NS, PE, NL), SK/MB, AB, BC - these are the regions I project off of;
  • Polls will only be used for those regions from the bullet above into which they break, national numbers are used to project the three northern seats;
  • For our purposes, the date of the poll is the average of all days in a sample (i.e. a poll conducted from Sep 29-Oct 1 has a date of Sep 30, a poll conducted from Sep 28-Oct 1 has a date of Sep 29.5)
  • The newest poll(s) is/are given full weight, all other polls are rated relatively with only polls taken within 6.5 days of the newest poll being considered (i.e. if the newest poll has an average date of Sep 30, it is given full weight; a poll with a date of Sep 23 would be given no weight and a poll with a date of Sep 23.5 would be given a weight of 0.5/7)
So for today's projection, we are using the following polls:

Angus Reid - Sep 24-25 (0.21 weight)
EKOS - Sep 29-Oct 1 (1.00 weight)
Ipsos-Reid - Sep 23-25 (0.14 weight)
Nanos Research - Sep 29-Oct 1 [ON, QC, Atl only] (1.00 weight)
Strategic Counsel - Sep 28-29 [ON, QC only] (0.79 weight)
Harris-Decima -Sep 28-Oct1 [ON, QC, Atl, BC only] (0.93 weight)

National results (change vs. 2006)
CPC 142 (+18), Lib 72 (-31), BQ 49 (-2), NDP 42 (+13), Ind 3 (+2)

Atlantic (change vs. 2006)
Lib 22 (+2), NDP 6 (+3), CPC 3 (-6), Ind 1 (+1)

Quebec (change vs. 2006)
BQ 49 (-2), Lib 12 (-1), CPC 11 (+1), Ind 2 (+1), NDP 1 (+1)

Ontario (change vs. 2006)
CPC 58 (+18), Lib 31 (-23), NDP 17 (+5)

Saskatchewan and Manitoba (change vs. 2006)
CPC 22 (+2), Lib 3 (-2), NDP 3 (n/c)

Alberta (change vs. 2006)
CPC 27 (-1), NDP 1 (+1)

British Columbia (change vs. 2006)
CPC 20 (+3), NDP 13 (+3), Lib 3 (-6)

North (change vs. 2006)
CPC 1 (+1), Lib 1 (-1), NDP 1 (n/c)

A few remarks...

The Alberta results stand out, however, the polling average shows the Conservatives down 10% in Alberta with the Greens and NDP each up about 5. If that were to actually happen on election day, it would be conceivable for at least one urban seat to flip. In this projection, it shows the NDP picking up Edmonton--Strathcona by a margin of 38 to 32.

There are two indpendents showing up in Quebec, one is obviously Andre Arthur. The other is incumbent MP Louise Thibault, formerly of the Bloc, in Rimouski-Neigette--Témiscouata--Les Basques. This riding shows the best (or worst) vote split ever... Ind 23%, CPC 22%, BQ 18%, NDP 18%, Lib 16%, Grn 4%.

A very interesting race in Central Nova where the projection shows the NDP winning and Peter Mackay placing third. Is this realistic? I'm not sure. This may be a good argument for tweaking the formula. What do you folks think? Current results project NDP 36%, Grn 35%, CPC 29%.

I'll continue to update daily until October 14. I welcome your suggestions if you think any aspect of the formula needs to be adjusted.

NOTE: These projections do not reflect the wishes, nor predictions, of the author. They are based on simple math using the criteria outlined here and here.

4 comments:

nuna d. above said...

Thibault won't likely have as much of a get-out-the vote team on election day as the BQ or Tories, so she may lose. I think your Atlantic Canadian numbers are low for the Tories. Do you have Lunn keeping his seat in BC?

nbpolitico said...

I also think that the results are low for the Tories - very low. The poll average currently shows the Tories down 6.5% in Altantic Canada from 2006, which doesn't seem right. I'm looking forward to a poll from Atlantic Canada with a larger sample size.

I am reluctant to pull out too many riding-by-ridings as there are all kinds of chances for errors at the riding level which, I hope, would cancel each other out to give a better regional and national picture. But, in any event, I have Lunn winning by a healthy margin in this projection.

nbt said...

A very interesting race in Central Nova where the projection shows the NDP winning and Peter Mackay placing third. Is this realistic? I'm not sure. This may be a good argument for tweaking the formula. What do you folks think? Current results project NDP 36%, Grn 35%, CPC 29%.

Louise Lorefice in the lead? Probably not realistic. But what the hell do I know?

To be honest, your projections may not be all that far fetched in Central Nova since the most concerning issue right now in is the closure of Trenton Works which saw close to 1,500 locals out of work after it closed in the spring and shipped the remainder of the positions to Mexico.

Now there are men who have never worked anywhere else on the unemployment line and food banks and social programs are overwhelmed.

Plus, it didn't hurt the dippers in Central Nova when their provincial counterparts held their convention in Antigonish tow summers ago. Darrell Dexter has really connected with a lot of those steel workers who have lost their jobs.

It'll be interesting to see how it plays out.

ww said...

This should be very interesting to watch.