Tuesday, May 5, 2015

Final Alberta election projection: NDP majority

Well, it is election day in Alberta, and if the polls are to be believed, the NDP is set to win an historic majority government in the province. Not only would it be the first time the NDP forms government in the province, but it would be the first time that any party other than the Tories have won since they first came to power in 1971. Is Alberta an NDP province? We're about to find out.

My projection comes late in the day, because I wanted to make sure there were no sudden last minute shifts in the voting intentions of the electorate. Last election, Forum Research released polls on both the Sunday and the Monday before the election, and together, they showed an eight-point swing from the Wildrose to the PCs in just one day of polling. Fast forward to election day, and the polls swung another 12 points, propelling the Tories to a surprise victory.

Once again, Forum Research polled the electorate on the day before the election (yesterday), and this time no major shift in the electorate was detected. Yesterday's poll showed that NDP support has continued to increase, topping at 45%. Both the Tories and Wildrose have polled about even for the last little while, and so the anti-NDP vote has not been able to coalesce behind either conservative party. Forum Research did detect a small swing from Wildrose to the PCs, which I have decided to extrapolate in this final projection.

For my final projection, I used yesterday's Forum's poll, EKOS's last poll from the weekend, and Mainstreet Technologies' massive poll released Friday. All three polls showed the NDP in the 44-45% range, Wildrose between 23% and 26% and the PCs between 21% and 23%. In my projection, I have made a rather bold assumption that the Tories will see a modest last minute swing in the polls, and will get over 27% of the vote. This will mostly come from would-be Wildrose supporters, switching their votes at the last minute. The NDP however will likely not see a last minute move from their voters, so they should still get somewhere in the mid 40s.

Of course, an NDP majority is a hugely unprecedented outcome, so other than knowing that the NDP will likely nearly sweep all of the Edmonton area, it is difficult to make an assumption about what other areas the NDP will do well. All the polls now show the NDP is ahead in nearly every region of the province, but how this will translate into seats is still a guessing game.

In Calgary, the NDP has built enough of a lead that it will be able to win a lot of ridings where they have been historically very weak. A lot of three way races will see the NDP coming up the middle with a relatively low share of the vote. Any last ditch effort from the Tories to stop the NDP will come in Calgary, where PC Premier Jim Prentice is from.

In the rest of Alberta, an area not known for its progressivism, the NDP still leads. It is unlikely that many of the truly rural seats in southern and central Alberta will go NDP, but it is becoming increasingly likely that the NDP will win many of the mid-sized cities across the province. Lethbridge at this point is a given for the NDP, but they could also win Red Deer, Medicine Hat and Grande Prairie. The NDP could also win a handful of seats in Northern Alberta, which are less rural and more remote.

As always with my final projections, I made a number of tweaks so that my model is as accurate as possible. I focused mostly on Calgary, where the most close races will be. I assumed any open Tory seat was probably toast for the PCs, the same with the Liberals. I also concentrated on the Edmonton suburbs, trying to match some of the ridings with a poll published a few weeks ago by Pantheon
Research showing the results by community.

The final result of my projection is an NDP majority of 50 seats. The Wildrose Party would form the opposition with 19 seats, despite finishing behind the Tories in the popular vote. The Tories would be close behind with 17 seats, but come up short of official opposition due to an inefficient vote distribution. The Liberals would win one seat, that of their leader David Swann, in Calgary-Mountain View.

In the Calgary area, my model shows the NDP winning 14 seats, the Tories nine, the Wildrose three and the Liberals one. The NDP would win seats across the city, as their rising tide will lift NDP boats in all corners of the city. The nine seats my model shows the PCs winning are located in the suburban part of the city, perhaps the area of the city least likely to go NDP. My model shows Calgary's two exurban seats wgoing Wildrose, while also winning a suburban seat in the south end of the city. The Liberals would retain their leader's seat, while a recent drop in Alberta Party fortunes has meant that their leader, Greg Clark is unlikely to win the riding of Calgary-Elbow, where he is running.

In Edmonton, my model shows the NDP winning all but two seats, but in reality could sweep all 26 ridings. The Tories would win the remaining two seats, Edmonton-McClung and Edmonton-South West. These two seats area is the wealthy southwest corner of thecity, so on paper it makes sense that they would eschew the NDP. However, most polls shows the NDP paradoxically leading among wealthy voters, so these two seats could just as easily go NDP. Plus, a poll in neighbouring Edmonton-Whitemud showed the NDP well ahead there, so don't be surprised if they do end up sweeping the city.

In remote Northern Alberta, there will be many three-way races that could easily go for the NDP, Wildrose or PCs. My model shows the NDP winning four seats, the Wildrose six and the PCs one. Wildrose was supposed to win the region back in 2012, but ended up with just one seat. This could happen again this time, but the party is buoyed by their leader running in Fort McMurray.

In Alberta's conservative southern region, my model shows the NDP winning the three urban seats (two in Lethbridge and one in Medicine Hat). It shows Wildrose winning the rest of the region, except for Banff-Cochrane, which could end up being a three way race. My model shows the PCs ahead there, but any of the three parties could win it.

Finally, in central Alberta, my model shows the NDP winning five seats; both Red Deer ridings and three more rural ridings. My model shows Wildrose retaining four of the five ridings they won in 2012, while it shows the Tories with four seats as well.

No matter what happens, it will be an exciting election night tonight. We'll find out for sure what the results will be when the polls close at 8pm Mountain Time (10pm Eastern).

Here are the projected results for each riding. Ridings are shaded by how they voted in 2012:

Monday, May 4, 2015

Prince Edward Island election today: Final projection

Today is election day in Prince Edward Island, which means my second and final projection of the campaign. Unfortunately since my last projection, there has only been one public poll released, and it is now over one week old. The poll, conducted by Corporate Research Associates (CRA) was published on April 24. It showed the governing Liberals in the lead with 44%, nine points ahead of the opposition Progressive Conservatives who were at 35%. The poll showed a narrowing of the race from numbers published a few days earlier by Abingdon Research. The CRA poll also showed the NDP at 15% and the Greens at 6%.

The CRA poll was commissioned by Charlottetown's The Guardian newspaper, which did not publish any tables. However, the corresponding article in The Guardian mentioned that the Liberals were well out in front in Kings County (58% to 31%), the traditionally more Tory-friendly part of the province. In Queens County, which contains half of the province's population, the race was much closer. The Liberals led there 39% to 37%. Numbers from Prince County were not published, but we can extrapolate from the other two county results that the Liberals lead there 43% to 35%.

While there has been no province-wide polling conducted in the last week of the campaign, we did get to see a riding poll from Kellys Cross-Cumberland, where Green Party leader Peter Bevan-Baker is running. The Greens commissioned Oraclepoll Research to poll the riding, and the numbers showed that Bevan-Baker was in the lead, with 38% of the vote, to Liberal incumbent Valerie Docherty's 34%. If the Greens do with the riding, it would be only the second time in PEI history where a party other than the Liberals or Tories won a seat.

Both the CRA poll and the Oralacle Research poll in Kellys Cross-Cumberland have been used to form my final projection. But due to the lack of polling and PEI's quirky nature, I had to adjust the numbers in a few ridings, specifically three ridings that are open seats where the Tories are running strong candidates, and a fourth where a floor-crosser is running for re-election. The three open seats with strong Tory candidacies are Charlottetown-Brighton where PC leader Rob Lantz is running, Borden-Kinkora and Rustico-Emerald. In the latter two ridings, the Tory candidates also ran in the 2011 election where they increased the PC share of the vote considerably over the provincial average, despite both losing to Liberal incumbents. Now, those Liberal incumbents are not running again, meaning the PC candidates will probably get even more of a boost in the polls. I also gave a boost to Hal Perry, the Liberal incumbent in Tignish-Palmer Road. Perry was elected as a Tory in 2011, but crossed the floor to the Liberals. I assume he will retain the 6% swing he got for the Tories from his personal popularity.

There are so many different local factors that will effect the races in each of Prince Edward Island's 27 electoral districts. These factors will make any prediction or projection of today's election very difficult, if any accuracy is to be had. It is hard to speculate which ridings will swing which way; we can only make assumptions based on known candidacy strengths. Last election, some ridings swung strongly Liberal, while others strongly PC. While many of these swings can be explained in hindsight, it would have been near impossible to accurately speculate on those swings before the election. Who would have predicted the Liberals to get a 24% two-party swing in their favour in Evangeline-Miscouhe, while the Tories got a 12% swing in Stratford-Kinlock and an 11% swing in Rustico-Emerald? Overall, between 2007 and 2011 there was only an overall average swing of 0.2% across the entire province, yet the Tories gained four seats from the Liberals, and the Liberals gained one back.

The 2007 to 2011 trend map shows how local factors are a more important indicator of how a riding will vote.

If the huge Liberal numbers in Kings County turn out to be true, th
en they should be able to win the two Tory seats in the county: Souris-Elmira and Georgetown-St. Peters. The only other Liberal gain I am projecting is in the riding of Tignish-Palmer Road, where the Liberal incumbent was elected as a Tory in 2011. With the PCs doing well in Queens County, my model shows them gaining four seats there from the Liberals: Tracadie-Hillsborough Park, Charlottetown-Brighton, West Royalty-Springvale and Rustico-Emerald. Additionally, my model shows the Tories picking up Borden-Kinkora (in Prince County) from the Liberals. Finally, my model shows the Greens winning their first ever seat in Kellys Cross-Cumberland. In total, this gives the Liberals 19 seats, the Tories 7 and the Greens 1. Despite polling in third, the NDP would be shutout, although they have an outside shot at a couple of Charlottetown seats.

Here are the projected results for each riding. Ridings are shaded by how they voted in 2012:

Polls close at 7pm Atlantic Time (6pm Eastern).