I will no longer be situated at unabashedly-right.blogspot.com
The new URL is http://politically-un-correct.blogspot.com/
Wednesday, May 11, 2011
Tuesday, May 3, 2011
41st Canadian federal election - post mortem
As results trickled in last night, boy was everyone shocked. First of all, while squeaking out a majority was considered a possibility, nobody expected 167 seats. Toronto, thank you. Next, while the NDP surge was quite properly predicted, it was still a meteoric rise and a historic moment. Third, the Liberal decimation was even worse than what most expected - even Ignatieff lost his seat. Fourth, nobody thought the Bloc would be so inefficient in its vote distribution - 23% of the vote and only 4 seats, Duceppe not included. Fifth and finally, Elizabeth May finally won her seat. There was actually a riding poll showing her in the lead, but it was commissioned by the Greens and I was skeptical.
The polls...to put it bluntly, failed miserably. Especially on the Conservative share. They were good for Quebec, but wildly off (underestimating the Tories) in every other region. While I corrected for some of that by giving the Tories an extra percent of the vote, while penalizing the NDP by two, it was still not enough. At 152 seats in the projection, I was actually among the least bad. I had a total error of 38 seats, compared with the 40's and even 50's for many. This sort of sea-change election validates my experiment with projecting ranges. Even with ranges, however, the Bloc fell outside of the range in terms of seats and the Conservatives broke the ceiling of their vote share range.
This calls for the question - was it the projection model's fault? Plugging in the ACTUAL results into the model, I get precisely 167 Conservatives. However, I would have underestimated the Liberals, giving them 26 seats. The NDP would have received 104 seats, and the Bloc 11. I would have been right about no independent, but still wrong about Elizabeth May. That's a total error of 18 - much improved over the polling numbers, meaning that it was mostly the pollsters' fault.
All in all, this was a failed call because I missed the most important detail - majority or minority. Oh well...at least I got the order of the parties right.
The polls...to put it bluntly, failed miserably. Especially on the Conservative share. They were good for Quebec, but wildly off (underestimating the Tories) in every other region. While I corrected for some of that by giving the Tories an extra percent of the vote, while penalizing the NDP by two, it was still not enough. At 152 seats in the projection, I was actually among the least bad. I had a total error of 38 seats, compared with the 40's and even 50's for many. This sort of sea-change election validates my experiment with projecting ranges. Even with ranges, however, the Bloc fell outside of the range in terms of seats and the Conservatives broke the ceiling of their vote share range.
This calls for the question - was it the projection model's fault? Plugging in the ACTUAL results into the model, I get precisely 167 Conservatives. However, I would have underestimated the Liberals, giving them 26 seats. The NDP would have received 104 seats, and the Bloc 11. I would have been right about no independent, but still wrong about Elizabeth May. That's a total error of 18 - much improved over the polling numbers, meaning that it was mostly the pollsters' fault.
All in all, this was a failed call because I missed the most important detail - majority or minority. Oh well...at least I got the order of the parties right.
Sunday, May 1, 2011
FINAL ELECTION PROJECTIONS
Well...this is it. Tomorrow is Election Day, so it's time for my FINAL PROJECTION.
This election, due to the sudden rise of the NDP, there is a large amount of uncertainty in the projections, therefore I have a new approach, and that is to give best/worst case scenarios for each party, before giving an average projection.
All numbers will be given in this order - BC, Alberta, Prairies, Ontario, Quebec, Atlantic, Total.
Conservative best case:
22, 28, 24, 71, 8, 16 = 169 - L 28, N 97, B 13, O 1. Conservative MAJORITY 30
Conservative worst case:
19, 26, 20, 44, 3, 12 = 124 - L 57, N 111, B 16. Conservative SHORT 31
Liberal best case:
8, 0, 2, 38, 10, 15 = 73 - C 130, N 93, B 11, O 1. THIRD; Conservative SHORT 25
Liberal worst case:
0, 0, 0, 7, 1, 6 = 14 - C 162, N 121, B 10. THIRD; Conservative MAJORITY 16
NDP best case:
18, 1, 6, 27, 55, 13 = 120 - C 155, L 20, B 13. OPPOSITION; Conservative MAJORITY 2
NDP worst case:
11, 0, 5, 22, 44, 4 = 86 - C 158, L 43, B 20, O 1. OPPOSITION; Conservative MAJORITY 8
[Quebec/Canada]
Bloc best case: 21 - C 3/150, L 4/40, N 47/97. FOURTH; Conservative SHORT 5
Bloc worst case: 10 - C 6/153, L 1/37, N 58/108. FOURTH; Conservative SHORT 2
AVERAGE:
Conservative - 35-39% - 21, 27, 22, 63, 5, 14 = 152
Liberal - 19-21% - 3, 0, 2, 21, 5, 10 = 41
NDP - 27-33% - 15, 1, 4, 22, 49, 8 = 99
Bloc - 24-28% [Quebec]/6-7% [Canada] - 15
Green - 4-6%
Other - 1% - 1 [Quebec]
Conservative SHORT 3
Possible outcomes:
Conservative majority - Due to massive vote splitting in Ontario, this is still possible, as long as things go slightly more favourable than polls suggest - 3.0
Conservative plurality, Harper - If Harper wins more seats than last time (145+ seats) then the Liberals might choose to prop him up instead - 4.0
Conservative plurality, Layton - If the Conservatives lose seats and the NDP+Liberal total ends up near or over 155, Layton will become Prime Minister - 3.0
Analysis:
The Conservatives are closer to their upper end because they've traditionally been good at getting out the vote.
The Liberals are in the middle of their range, as they usually are, but even their upper limit is a historic worst.
The NDP is slightly at the lower end [of a very wide range, high uncertainty due to their shocking rise] due to the fact that they are less organized on the ground; still they will become the official opposition.
The Bloc is in the middle of their range, since their usual ballot-box penalty should be canceled out by incumbents hanging on by name recognition.
I have Andre Arthur [independent] keeping his seat in Quebec but there are differing opinions out there.
The Greens are not expected to win May's riding, due to much lower exposure this campaign.
This election, due to the sudden rise of the NDP, there is a large amount of uncertainty in the projections, therefore I have a new approach, and that is to give best/worst case scenarios for each party, before giving an average projection.
All numbers will be given in this order - BC, Alberta, Prairies, Ontario, Quebec, Atlantic, Total.
Conservative best case:
22, 28, 24, 71, 8, 16 = 169 - L 28, N 97, B 13, O 1. Conservative MAJORITY 30
Conservative worst case:
19, 26, 20, 44, 3, 12 = 124 - L 57, N 111, B 16. Conservative SHORT 31
Liberal best case:
8, 0, 2, 38, 10, 15 = 73 - C 130, N 93, B 11, O 1. THIRD; Conservative SHORT 25
Liberal worst case:
0, 0, 0, 7, 1, 6 = 14 - C 162, N 121, B 10. THIRD; Conservative MAJORITY 16
NDP best case:
18, 1, 6, 27, 55, 13 = 120 - C 155, L 20, B 13. OPPOSITION; Conservative MAJORITY 2
NDP worst case:
11, 0, 5, 22, 44, 4 = 86 - C 158, L 43, B 20, O 1. OPPOSITION; Conservative MAJORITY 8
[Quebec/Canada]
Bloc best case: 21 - C 3/150, L 4/40, N 47/97. FOURTH; Conservative SHORT 5
Bloc worst case: 10 - C 6/153, L 1/37, N 58/108. FOURTH; Conservative SHORT 2
AVERAGE:
Conservative - 35-39% - 21, 27, 22, 63, 5, 14 = 152
Liberal - 19-21% - 3, 0, 2, 21, 5, 10 = 41
NDP - 27-33% - 15, 1, 4, 22, 49, 8 = 99
Bloc - 24-28% [Quebec]/6-7% [Canada] - 15
Green - 4-6%
Other - 1% - 1 [Quebec]
Conservative SHORT 3
Possible outcomes:
Conservative majority - Due to massive vote splitting in Ontario, this is still possible, as long as things go slightly more favourable than polls suggest - 3.0
Conservative plurality, Harper - If Harper wins more seats than last time (145+ seats) then the Liberals might choose to prop him up instead - 4.0
Conservative plurality, Layton - If the Conservatives lose seats and the NDP+Liberal total ends up near or over 155, Layton will become Prime Minister - 3.0
Analysis:
The Conservatives are closer to their upper end because they've traditionally been good at getting out the vote.
The Liberals are in the middle of their range, as they usually are, but even their upper limit is a historic worst.
The NDP is slightly at the lower end [of a very wide range, high uncertainty due to their shocking rise] due to the fact that they are less organized on the ground; still they will become the official opposition.
The Bloc is in the middle of their range, since their usual ballot-box penalty should be canceled out by incumbents hanging on by name recognition.
I have Andre Arthur [independent] keeping his seat in Quebec but there are differing opinions out there.
The Greens are not expected to win May's riding, due to much lower exposure this campaign.
Subscribe to:
Posts (Atom)