As the election officially started in Ontario, some potentially important variations have occurred. The NDP in particular seems to be on the rise. Not enough to make this a competitive race as the PC of Doug Ford is still firmly ahead and, if the election were tomorrow, would almost be guaranteed to win the most seats. But it might be the first signs that this election could, at the very least, become more competitive. I live in BC and I'm fully neutral here, but I have to admit that elections are more fun to cover (and do projections for) when they are at least slightly competitive. After all, there are only so many ways of saying "this party is almost sure to win".
Before going on, I'd like to mention that my lack of updates recently was due to me being on vacation in Mexico. I know, I know, it's totally unacceptable for me to miss the first few days of the campaign. But I'm now back and will write more regularly.
Alright, back to the projections. For you, busy people, here there are. Remember that the probabilities of winning are estimated using simulations. Beyond the technicalities, just see the percentages as representing the uncertainty that currently exists based on all the available information (past results, polls, etc). As you can see, there isn't currently much uncertainty regarding who would win.
 |
| Voting intentions; seat projections with confidence intervals; Chances of winning the most seats |
Some technical notes. I made several updates to the model. I slightly tweaked the regional coefficients (estimated with past elections) as I now account for having lost a long term incumbent (more than 2 terms; It shows a decrease of between 3-8 points when that happens). I also finally accounted for by-elections results (it makes a difference in Sault Ste. Marie). Finally, I got access to roughly a dozen of non published riding polls. In every one of them, my predicted winner was the same as what the poll was showing but some numbers were quite different. Riding polls aren't always very accurate (they literally caused me to make 2 extra mistakes in last year's BC election) but I can't completely ignore polls showing a very different situation.
Beside the changes to the model itself, recent polls have shown the NDP rising. After weeks of a stable situation (PC ahead around 40-42%, OLP second around 28% and NDP third around 20-22%), the situation finally changed. The Liberals are collapsing (this is what having record breaking satisfactions rates can do to you) while the Tories of Ford appear to be losing some steam.
How can we explain these changes? Most likely because people have started paying attention. We usually say that there are three moments during a campaign where people listen: the beginning, the debate and the very end. For months it has been incredibly clear that Ontarians do not want the Liberals to be reelected. Kathleen Wynne is incredibly unpopular. On the other hand, not everyone is comfortable with the Conservatives, especially not with Doug Ford as leader. This naturally should help the NDP whose leader, Andrea Horwath, isn't particularly popular but also isn't disliked or divisive.
If Kathleen Wynne was hoping for a comeback, she had to win the first debate. Unfortunately for her, that wasn't the case at all based on a Mainstreet poll.
Does it mean this campaign will ultimately become a PC vs NDP race? It's too early to tell but I'd bet good money on this. I'd be utterly shocked if the Liberals were able to climb back. Maybe if there wasn't a third option like the NDP but there is indeed one!
The PC's chances are at 99.8%. This is very close to absolute certainty. What this means really is that even if we account for the possible mistakes by the polls (the simulations use actual margins of error close to 5-6%, so there is a lot of uncertainty added to the model. Do NOT think that I'm assuming the polls will be perfect next month) or by the distribution of the votes/electoral system, almost every possible scenarios end with the PC winning the most seats.
Some of you will be quick to point out that in the Canadian system, winning the most seats doesn't mean you "win" or you end up Premier (just ask Christy Clark!). This is correct. The chances of a majority are at 89.8%. Make no mistake, this is a level of certainty way higher than for most elections I've covered. You can try to pretend that "anything can happen" but the fact is that right now, if the election was tomorrow, it'd be a monumental surprise if the PC didn't finish first. And it'd be a huge surprise if Ford wasn't getting a majority. This would be a bigger surprise than Trump winning the election. Or when the BC Liberals won in 2013. It doesn't mean things won't change bwteen now and election day though.
How dire is the situation for the Liberals? They have fallen to third in terms of votes and seats. The GTA, usually welcoming to this party and a source of many seats during the last 10 years, is moving away from Wynne. As a matter of fact, between Ford being quite popular in the 416 (Toronto proper if you prefer) as well as in the 905 and the rise of the NDP, it's not impossible for the Liberals to end up with a ridiculously low number of seats. That include potentially winning zero! Odds are very low (9 simulations out of a 10,000) but it's not impossible. On the other hand, the model does give the Liberals a 0.1% chance of finishing first (it would require the polls to be more wring than ever and an incredibly efficient vote). So again, the model does include a lot of uncertainty (as it should since polls have been shown to be quite wrong at times).
The NDP isn't currently a bigger challenger because its vote is too concentrated. However this could change if this party manages to pass the 30% mark. And if we start observing a massive swing from OLP to NDP (a movement quite possible given the second chances of Liberals voters that have been shown to heavily prefer the NDP as 2nd choice; Think of a "anybody but Ford" movement from progressive voters), then we could start seeing an actual contest. The PC won't be in real danger as long as they stay around 40% but Doug Ford can't afford to drop much below this threshold if the NDP keeps rising. Especially if Ford wants a majority. The latest Ipsos poll was showing that people were quite divided as to which outcome they preferred. If the PC were to fail to secure a majority, it's not crazy to think the NDP could try to get the support from the Liberals. I don't really want to enter this discussion as it's all conjectures, but keep in mind that this is a very possible possibility (again, ask Christy Clark!).
I don't have much more to add for now. I'll write specific analysis later this week (such as a look at how the NDP could become a real challenger). For now, I'll leave you with the histogram of the possible outcomes.
You can see the more concentrated NDP vote is causing the distribution of seats for this party to be more concentrated. Having your vote concentrated is good when you are a small party but less ideal when you are a big one that aspire to take power.
And finally, the riding by riding projections. As usual, you can always make your own projections using the simulator.
Ontario Projection May 14th 2018 Details by bryanbreguet on Scribd
