Monday, October 3, 2011

Final PEI Projection: Ghiz Liberals win majority government

Prince Edward Islanders head to the polls today in the first of five provincial elections slated for the fall. projects that the Liberals under Robert Ghiz will win a majority government, with Olive Crane's Progressive Conservatives forming the Official Opposition.
The projection has not changed since the last projection of September 30. The Liberals are projected to win 52.9% of the vote and 26 seats, with the Progressive Conservatives winning 34.2% of the vote and one seat.

The New Democrats, with 8.1% of the vote, and the Greens, with 4.1% of the vote, are not projected to win any seats.

The Progressive Conservatives are projected to win their one seat in the eastern part of the province. The Liberals take all of the rest.

Only two polls were released during the campaign, with another being released just before the campaign began.

The polls showed little movement, particularly when considering the large margin of error. The Liberals started the campaign at 59% support and dropped to 53%, while the Tories ranged between 31% and 36%. Where the New Democrats end up is disputed: the Corporate Research Associates pegged their support at 5% while MQO Research put them at 12%.
As the above chart shows, most of the campaign went un-polled, while no data was released for the last six days of the campaign.

The above chart boils down the final projection. Only one riding is projected to be close, putting the Liberal range at between 25 and 26 seats and the Progressive Conservative range at between one and two seats.

However, Prince Edward Island is a tiny province with even tinier ridings. There is a lot of potential for several surprises as only a few hundred votes need to swing one way or the other in virtually every riding in order to change hands. Being such a small province, the difficulty in projecting the result is compounded.

If we take into account the margin of error in the 2011 federal campaign's projection (with accurate vote results, that is), the potential exists for the Liberals to win as few as 23 seats and as many as 27. The Tories could win between zero and four. But it is unlikely that all of the errors would swing in one direction.

But what if the polls have it wrong? The 2007 provincial election was polled relatively well. Even the degree of polling error in the 2011 federal election isn't enough to change the race in Prince Edward Island dramatically.
The above chart adjusts the projection by the degree of polling error in the 2007 provincial and 2011 federal campaigns. For example, the Conservatives averaged 37.4% in the final polls of the 2011 federal campaign, and ended up getting 39.6%. This would mean an adjustment of PC support by a factor of 1.059. After making the adjustments, the vote projection is then smoothed out to add up to 100%.

If the polls are off in the same way as they were in 2007 (though it is, of course, unlikely that the same error would be reproduced exactly), the Liberals should take 55.6% of the vote to the Tories' 34.9%. This would not change the projection: 26 Liberal and one Progressive Conservative seat.

If the polls are off in the same way as they were in the 2011 federal campaign, the Liberals should take 52% of the vote and the Tories 36.2%. Again, that does not change the projection but does increase the likelihood of the PCs taking an extra seat.

But, in the end, the only question heading into tonight's vote in Prince Edward Island is how large the Progressive Conservative opposition will be. The potential for a Liberal sweep of all 27 ridings exists, but I consider it unlikely. The need to have some opposition in the legislature is an advantage to Olive Crane, and the final poll of the campaign indicated that support in Kings County was neck-and-neck between the Liberals and the Tories. Whether the PCs win one or two seats will likely be decided here, but a Liberal majority is inevitable.


  1. It's tough to generalise across elections, but looking at the spring federal election and this current slate of provincial elections, it seems that Liberal parties can keep in the running if they are incumbent (PEI, Ontario) but if they're fighting a campaign from the Opposition side, it seems 2011 has been a very unkind year. It's an interesting thing to consider as Liberals nationwide regroup, and perhaps good news for the incumbent, if atypical, parties in BC and Quebec.

  2. I think it would make more sense to generalize further - in that this year is shaping up to be quite incumbent friendly, regardless of party.

  3. BTW, with respect to the new poll out from EKOS ( - I tried weighting their numbers based on the reported 2011 federal vote, and got Conservatives 36.9, Liberals 36.5, NDP 20.7 and Greens 5.1. So it looks like EKOS has messed up their methodology again...

  4. Reverse weighting is never a good idea, you don't really know what was done with the numbers, but if I understand correctly EKOS did not mention the reported 2011 federal vote, but the current federal voting intentions.

  5. Oh, you are correct. Hrm, my bad. Darn, and that weighting brought the numbers in line so nicely.

    For the reverse weighting - IIRC quite a few British pollsters do this, to try and control for the "shy Tory" factor.

  6. Ryan you're right that a few British pollsters do that, or more accurately USED to do that - the 2010 pollsters actually did very little "shy Tory" weighting, since that effect was nonexistent in an election that was a wave against Labour. In fact there was talk of a shy Labour factor in play, given the results we actually ended up seeing.Ryan you're right that a few British pollsters do that, or more accurately USED to do that - the 2010 pollsters actually did very little "shy Tory" weighting, since that effect was nonexistent in an election that was a wave against Labour. In fact there was talk of a shy Labour factor in play, given the results we actually ended up seeing.

  7. PEI should be the first Province to test out a proportional representation model of elections, given its micro-population. There's no way that it's fair that the Tories can have 34.2% of the popular vote yet on track to only win one riding.

  8. The results coming in have the Tories taking way more than 1 riding. Right now they are leading in 7. Seems to be at least in part because the NDP vote collapsed, since the Liberal vote is running very close to Eric's projection.

  9. Results are in:

  10. The NDP vote is up significantly from the last election, but since they did not run a full slate their popular vote overall is not as high as it could have been. With a full slate the NDP would have been are more like 6%.

    Meanwhile, the Conservatives doing much better than the model suggests, about 6 percentage points ahead and four more seats. They are within a few votes of winning two more seats

  11. Looks like Corporate Research Associates pretty much called the final results with their last poll. They slightly understated the Tories and slightly overstated everyone else, but compared to the numbers coming out in the smaller-sample polls before that (12% for the PEI NDP?), they were close to dead-on.

  12. Yes, CRA did pretty well.

    The big problem was under-estimating Tory support, but a lot of ridings were decided by only a handful of votes. 425 votes separated the PC's 5 seats from my high seat range of 2, and in two ridings the PCs won by about 30 votes.

    Considering the size of PEI and the size of the ridings, there was only so much that the model could do, and of course it will only be as good as the polls. I'm disappointed I wasn't bang on, but PEI was always going to be difficult.

  13. How the CBC calls it a 8:30 PM EDT

    LIB: 22

    PC: 5

    GRN: 0

    NDP: 0

  14. in one riding, the liberal candidate beat the conservative candidate by 8 votes.

    PEI is indeed very small so it's hard to use models to predict results but in the last federal election, 308 blog also under estimated the number of seats the conservatives would take as well. Maybe in Ontario it will be shown that they were underestimated as well

  15. PEI is small enough that if you family has the flu on election day you could very well lose

  16. I see 3 ridings marked as PC held in your projection btw - but wasn't one of the Liberal seats won in a by-election?

  17. Ryan,

    No, two were held by the PCs at dissolution and one was vacant. It was a former PC seat.

  18. PEI already voted on PR and it was defeated as it should have been. PR replaces one bad system with another.


COMMENT MODERATION POLICY - Please be respectful when commenting. If choosing to remain anonymous, please sign your comment with some sort of pseudonym to avoid confusion. Please do not use any derogatory terms for fellow commenters, parties, or politicians. Inflammatory and overly partisan comments will not be posted. PLEASE KEEP DISCUSSION ON TOPIC.