Pre-election Outlook

Overview
The federal election is about to be called for June 28. In this note I will briefly outline my overview of how it might go and provide some polling statistics you won’t see elsewhere. My view is that the media has, since the sponsorship scandal broke, been endeavouring to construct a simple narrative that sees the election as a gladiatorial contest between Martin and Harper. The problem with this is that the Conservatives are non-existent in Quebec and very weak in Atlantic Canada, and running far behind their performance in the 2000 election (as the Alliance and the PC’s combined) elsewhere. To win a majority I estimate they would need to win about 140-145 seats, about 70% of the total west of the Ottawa River. The media have been saying over and over that Stephen Harper is underestimated (the latest example can be found in Graham Fraser's column in the Sunday May 23 Toronto Star), which means, because the media keep saying it, that the exact opposite is true – he is by no means being underestimated.

I view the appropriate question about the campaign as: Will the Liberals be able to win a majority? My own intuition is that it will be exceedingly difficult and, in fact, unlikely. To do so they must recover in Quebec. This appears improbable at the moment, but I have always found Quebec politics difficult to read and have been surprised many times in the past. One thing that might be a consideration in Quebec voting is the de-merger referendums scheduled for June 21st. There are going to be quite a few. However, it is unclear today what the impact might be, and which party might benefit.

If the Liberals flame out like John Turner in 1984 the Conservatives could win the most seats but a majority is almost certainly out of reach. I have thought for a long time that the national media has been underestimating the NDP. Yes, they see them returning to a traditional level of support, 18% or so in the polls, about 20 seats in the House of Commons but they sort of dismiss them with a yawn.

The NDP's support in the polls has averaged 17% over the past couple of months but when I convert the polls of the past couple of weeks to seat numbers I calculate they would win on average roughly 29 seats. In addition, the numbers on leadership when available tell me that large numbers of voters still don’t know much about Jack Layton. I think the NDP could quite possibly increase their support to greater than 20% (their previous high) during the campaign, particularly if Martin performs poorly. And a strong showing by the NDP will have a critical bearing on whether or not the Liberals can form a majority government, and whether any two parties can form a majority if the Conservatives are able to win more seats than the Liberals. The media ought to have been paying more attention to the NDP. However the media rules for balanced campaign coverage mean the NDP will henceforth be more or less on an equal footing with the other parties. Plus the party appears to be competitive in terms of money for advertising.

Polls
Much of the fluctuation in the polls in recent months in my view is just so much statistical noise. I actually think changes since February have been marginal.

The National Post and the Globe have made much of the McGuinty budget affecting support in their polls for the Liberals. They both have them at 42% but one (Compas) has the Conservatives at 39% (which is inconsistent with all previous polling) while the Reid has the Tories at 28%. Clearly, someone is wrong.

I have created a weighted average of the recent polls (since April 1) in the table below including the Saturday May 22 Reid poll and the new CROP poll in Quebec (which is far more likely to be accurate than Reid about the Quebec numbers.)

April-May Average National and Regional Polls Province...Lib....Cons...NDP...BQ...Green...Oth...Total...Sample...MoE
Atl........46.9...30.0...18.6..........1.3.....3.1....100......4,048...1.5
Que........35.1...8.8....7.5...44.7..1.7....2.4.....100.....17,691...0.7
Ont........46.6...29.1...18.8..........2.7.....2.8...100.....11,297...0.9
Mb/Sk.....34.0...33.4...26.4..........2.6.....3.6...100.......3,779...1.6
Alta........27.5....53.2...13.2.........3.5.....2.6....100......4,512...1.5
BC.........32.1....30.0...27.2..........6.2.....4.6....100......5,800...1.3

Cda........38.9....26.7...16.9..11.4...3.1.....3.1....100......48,512...0.4

When I apply these numbers to my seat generator I get the following outcome:
Liberal - 158, Conservatives - 69, NDP - 28, BQ- 53. A bare majority is 155 so it is clear that the Liberals were (during this period)just over the line. Contrary to the findings in the Ipsos-Reid poll published in Saturday's Globe, I find the following seat totals Liberal-141, Conservatives-76, NDP-30, BQ-61.

Pre-election Polls Compared to Final ResultsOne phenomenon of previous campaigns has been that the Liberals lose support during the campaign. This happened, for example, in the 2000 election. This time the Liberals start from a weaker position than four years ago, so it is more likely they are closer to their base vote and any drop-off will be smaller. It is worth noting that the drop-off for the Liberals appeared to be greater in the West than the East (except in Atlantic Canada where pre-election polling had the Liberals very high. Could it happen again? Yes, and if the pattern repeats itself, a Liberal minority would result.

Comparison Pre-campaign Polls Compared to Final 2000 Results
..............Lib......All.....BQ.....NDP.....PC....Oth.
Atl........-16.0......-0.7...........+4.5...+11.4...-0.8
Que.........+1.2......-2.1...-0.3....-2.2....+2.7...+0.6
Ont.........-5.4.......+3.7..........-1.6....+3.1....0.0
Mb/Sk.......-11.2......+8.0..........+4.9....+0.3....-1.8
Alta........-7.6.......+4.8..........+0.5....+3.2....-1.1
BC..........-10.1......+7.7..........-0.2....+1.3.....0.0

This table compares polls from late September to the election call in the third week in October to the final official results.

My conclusion: fasten your seat belts; it’s going to be a wild ride.