Thursday, June 12, 2014

2014 Ontario Election - FINAL PROJECTION

Final projection map
It's finally election day in Ontario, and that means my final projection of the campaign. All the pollsters have released their final polls, which they got in before the midnight publishing deadline, last night. The trend that most of the polls show is that the Liberals are headed for a modest victory in the popular vote. If we average all the polls released yesterday (from Abacus, Ipsos, EKOS, Angus Reid, Forum Research and even Leger), the Liberals would win 36.1% of the vote, the Tories would win 32.7% of the vote, the NDP would win 24.7% and the Greens would win 5.2%. This would be a small loss (-1.5%) in the popular vote for the Liberals, who won 37.6% of the vote in 2011. But it would be an even bigger loss (-2.7%) for the Progressive Conservatives, who won 35.4% last election. The NDP and the Greens would see an increase in their vote share if the polls are correct, with the NDP up +2.0% and the Greens up +3.3%.

The Liberals' narrow win in the popular vote in 2011 was enough to come just one seat short of a majority. In stands to reason that if they increase their popular vote lead over the Tories, that they could theoretically win a majority this time. However, the NDP is the real wild card in this election. The NDP has been polling all over the place in the last few weeks of the campaign. Just yesterday, Ipsos had the NDP at 30% province wide, but EKOS on the other hand had them at 19%! A whole 11 point difference. How well the NDP does could mean the difference between a majority and a minority for the Liberals, because a lot of the seats the NDP would win with a high popular vote (as opposed to say, 19%) would come at the expense of the Liberals.

For my final projection, I inputted the regional polls from Angus Reid, Ipsos, Forum Reasearch, Abacus and EKOS (Leger didn't have any regional numbers). I also made a few last minute tweaks to my model, so that I could come up with a result that I would be comfortable with. The first change I made was to alter the poll numbers to reflect my own projected turnout model. I did this by looking at previous election results, and comparing them to the polls conducted prior to them. I found that in Ontario elections, there is a small “shy Tory” effect, which has resulted in the Tories often under-polling in elections. The opposite is true for the NDP, which often over-polls in elections. The Greens also under poll, while the Liberals usually poll consistently with actual election results. I took a look at the ratio of poll results vs. Real results from the last election, and applied it to my model to project the popular vote outcome for this election. I figured, the two elections are very similar, in that there is a somewhat unpopular Liberal government that appears to to be heading for re-election, but with the outcome of a majority or minority still in doubt. Therefore, turnout will likely be similar to the 48% in 2011, or even less - because this election is in June - when many people could be away (while the 2011 election was in October).

The second major change I made to my model was to alter the data in a few select seats, to even them out with recent riding polls conducted by Forum Research. I was noticing due to polls conducted by Forum Research in Trinity-Spadina and Davenport - that I had inputted into my model - that the NDP numbers in my model were higher there than in “safer” NDP seats in Toronto, like Parkdale-High Park and Beaches-East York. So, I boosted the NDP numbers (at the expense of the Liberals) in the latter two ridings, so that my model projected a higher NDP vote share in those ridings than in Davenport and Trinity-Spadina. I made a similar shift in the Waterloo region, by boosting Liberal numbers in Kitchener Centre and Kitchener-Conestoga, due to polls conducted in Kitchener-Waterloo and Cambridge showing higher than expected Liberal numbers. And I also boosted Liberal numbers in the main swing ridings of the 905 region, due to a higher than expected Liberal poll result in Burlington. The 905 ridings I boosted Liberal numbers were in Halton, Thornhill and Newmarket-Aurora. It stands to reason that the Liberals will do better than projected in other 905 ridings, but I felt it was most important to focus on those three ridings, since they are the most likely to flip to the Liberals in the region (closest in my model).

After all the changes I made, my model has shown the Liberals are projected to win 52 seats, which would be just two short of a majority. This would be a decline of one seat from 2011, but a gain from the 48 they held at dissolution. For the Tories, my model shows them winning 32 seats, which would be a disaster for the party, as they won 37 seats in 2011, and had 38 at dissolution. If this were to happen, one could logically expect their leader, Tim Hudak to resign. And finally, my model shows the NDP winning 23 seats, which would be a very good result for the New Democrats, who won 17 seats in 2011, and had 21 at dissolution.


For this projection, I am going to analyze each of Ontario's regions separately, so to also give a preview of what to look for on election day.

Ridings in seat charts are coloured by how they voted in 2011.

Toronto

Toronto's five lakefront ridings that are held by the NDP (Parkdale-High Park, Davenport, Trinity-Spadina, Beaches-East York and Toronto-Danforth) will be key seats to watch today. Recent poll numbers from the city show NDP numbers to have recovered from mid-campaign to the point where they could win all five seats once again. However, Forum Research polled Davenport and Trinity-Spadina almost 2 weeks ago, and showed the Liberals to be leading in both ridings. This was at the lowest point in the polls for the NDP in Toronto, which is why my model now shows the NDP ahead in both ridings. But with polls all over the place, the Liberals could win up to four of the five NDP held seats (Toronto-Danforth is probably safe). The only shift in seats my model shows in Toronto is Etobicoke-Lakeshore. That riding was won by the Liberals in 2011, but picked up last summer in a by-election by the Tories. My model shows the Liberals regaining the seat.



Southwest

Recent polling suggests there could be an NDP surge in the southwest, which has resulted in my model showing the NDP holding all of their current seats in the region, plus Windsor West (which a Forum Research poll showed they were ahead in), which would be a gain from the Liberals. One key seat to watch in the region to measure NDP fortunes is Kitchener-Waterloo. Forum Research did a poll in the riding, showing the Liberals narrowly ahead of the NDP in the riding. If there is an NDP surge however, it means the NDP will probably win it. This would be a hold for the New Democrats who won it in a by-election from the PCs in 2012. Another race to watch to measure NDP fortunes would be London West. Forum showed the party ahead there (but it was close), so losing it would be bad news for New Democrats.

There are a number of Liberal vs. PC races in the region to watch for as well. My model currently shows the Liberals gaining one seat from the Tories (Cambridge), and the Tories gaining one seat from the Liberals (Brant). Forum Research had a poll showing the Liberals way ahead of the Tories in Cambridge, which has never voted Liberal in its history. In Brant, the election was very close there in 2011, and the Tories are running a former MPP. Brant is actually the only riding in the province that I have the conservatives picking up from their opponents. Other Liberal vs. PC races to watch in the region include Kitchener Centre (an open seat, close last time) and Kitchener-Conestoga (if Liberal gains in Cambridge come true, there could be Liberal gains in this neighbouring riding as well).

If you're a Green supporter, then you'll want to watch Guelph, to see how their leader Mike Schreiner does.



Eastern

Toward the end of the campaign, it looked like the Progressive Conservatives were going to make big gains in the east, perhaps picking up as many as three or four seats from the Liberals, which would spoil any chance of a Liberal majority government. However, recent polling suggests the Liberals have recovered in Eastern Ontario, which could mean a status quo result, like my projection shows. Forum Research polled two key ridings in the region: Ottawa West-Nepean and Glengarry-Prescott-Russell. Both showed a narrow Tory lead. However, that was when the conservatives were polling very strong in Eastern Ontario. With the Liberals now coming on strong, it stands to reason that they could hold on to those two seats. Another seat in the region, Ottawa-Orleans will also be key, as it is an open seat and was very close in 2011. In my opinion, the Liberals need to hold onto all of their seats in the region to win a majority.



North & Central

There is probably just one seat to watch in this region, and that is Sudbury. Sudbury was a very close race between the Liberals and the NDP in 2011, and with the Liberal incumbent retiring, it means the seat is the NDP's to lose. And there have been two riding polls that have shown the NDP to be ahead with a decent margin. My model reflects these poll numbers, which is why I have the NDP gaining Sudbury. Forum Research also did a riding poll in Thunder Bay-Atikokan which saw the Tories coming on strong. In 2011, it was a close Liberal-NDP race, so this would be a huge shock if it came true.



905 belt

The 905 region, which surrounds Toronto is almost always the most important political region in the province. The region is the province's bellwether, swinging to and fro depending on the political winds of the time. The region is also the most populous in the province, being home to 33 ridings. There are a number of key ridings in the 905, which will be pivotal for the Liberals to win if they hope to form a majority government.

For the Liberals, they need to win all of the seats they hold right now in the region, and make some pick ups. My model currently shows them gaining Burlington, Halton, Newmarket-Aurora and Thornhill from the Tories. This may be enough to win a majority if the Liberals can gain seats in Toronto, and hold onto their seats in Ottawa, and elsewhere. Another riding that could flip Liberal is Northumberland-Quinte West, where they are running a former MPP who lost the seat in a close race in 2011. They also have an outside shot at Bramalea-Gore-Malton, which the NDP won in 2011, but is not an historically NDP riding.

For the Tories to save face in the election, they need to make gains in the 905. But my projection doesn't show them doing so. Losing seats would be a disaster for the PCs. The only gains they have an outside shot at making would be Brampton-Springdale, St. Catharines and Oakville (from the Liberals) and Niagara Falls (from the NDP). But all riding polls have shown the Tories are trailing in key 905 ridings.

For the NDP, there is an outside shot of winning Oshawa, but I wouldn't hold my breath on that one.



Conclusion

If polls are to be believed, than this race is the Liberals' to lose. The main story line to watch for tonight is whether or not they can win a majority government of at least 54 seats. For this to happen, the Liberals need to hold almost all of their current seats (acknowledging a couple of losses here and there), and gain enough back, and then some. To do this, they need to win seats in the 905 from the Tories, and in the 416 (Toronto) from the NDP. My gut feeling is that the election will come down to whether or not the Liberals can make these gains in Toronto. This election might just come down to the downtown voters who have jumped the NDP ship due to being uncomfortable with the party's (new found?) populist rhetoric.

We'll know for sure after 9pm tonight, when the polls close. 


 Methodology

The basis of my model extrapolates the results of the 2011 election using regional polling averages from recent polls. If a party is polling at double the level they won in 2011, then that party would see its support double in each riding in the region my model (that is, if they won 5000 votes in a riding, they would be projected to win 10000). I have also taken into account recent by-elections (only in the case where a non-incumbent part won), but projecting the results of the by-elections backward to the last election based on poll numbers from around the time of each by-election. That is, if a party did 5 points better than polls indicated they would've won in a by-election, then I have made their 2011 result 5 points more than they actually received in 2011.


I've also made numerous tweaks in the model where I've felt appropriate, to boost particular candidates, or to use estimates based on recent federal election results. Also, I have inputted riding specific polling for ridings that have been polled during the campaign (and altered nearby similar ridings to match the same shifts). Please refer to previous blog posts for more details.

2 comments:

  1. I'm loving it!
    mike( Timmy) Hudac needs a real job.
    Guess that gets his Fire 100,000 down to 99,999.
    Enjoy the night folks.
    DSH

    ReplyDelete