Last update to the CPC leadership projections: Bernier still heavy favourite

Mainstreet just released its latest poll for the CPC leadership that is ending this Saturday. With that said, I'm not sure a last minute poll is that useful given that most members mailed their ballots already. But based on Mainstreet's numbers, even the last minute trend is favourable to Bernier (and not to Andrew Scheer). On the other hand, we finally have a regional breakdown provided for free (i.e: without paying $100). I decided to keep using my model with the fundraising data but it convinced me to update the projections. In particular, I realized I might have overestimated Bernier in Quebec. And since this province is so important, any adjustment made there can have dramatic consequences on the overall picture.

I thus updated my projections of earlier this week (read for the detailed methodology). The main result remains: Maxime Bernier will be the next leader of the Conservative Party.

Look, we can do the simulations, keep in mind how tricky a leadership race is and how the electoral system (and the incorrectly filled -and therefore invalid- ballots), but Bernier is by far the favourite here. It'd be shocking to me if he were to lose this weekend.

The number of points is really similar but the chances have changed. Since Scheer is higher in Quebec (and Bernier lower), it makes the race slightly more competitive. Still, make no mistake, for Andrew Scheer to win, he needs the poll and fundraising data to be wrong, very wrong.

As for the distribution of possible results in the first round, here it is:

I said it before and I'll say it again: if you don't believe my projections, simply think about it: Bernier will almost surely get around 50% in Quebec. Unless his vote is incredibly inefficient (remember, each riding is worth 100 points), it means Quebec alone will provide him with around 3500-4000 points in the first round. This alone pretty much guarantees Bernier to be ahead after the first round. This is, after all, more points than any other candidate is projected to get across the entire country, except for Scheer (and maybe O'Toole and Lemieux who are both right around 4000 in average).

Bernier will likely not win in the first round, which means that we'll go through a number of "useless" rounds where we redistribute the votes of the candidate who finished last. But there will be so few votes (or points) to redistribute that it won't change anything to the big picture. It will only get interesting around the 7 or 8th round (so when we'll likely be eliminating Trost or Raitt).

Again, if you think I'm too confident for Bernier, remember that I introduced a ton of uncertainty into the simulations. I also made some adjustments to the fundraising data based on the polls that are helping Scheer and hurting Bernier. I even didn't mostly used the next to last Mainstreet poll (instead of this one here) because it was better for Scheer (well, that's not really the why. I used the next to last poll because I don't think a last minute poll is useful when members have already voted).

The Mainstreet regional breakdown also shows that Bernier is ahead pretty much everywhere except in Saskatchewan (province of Scheer). Another indication of the fact Bernier is likely to win.

Some will argue that the first round isn't what really matters. Based on the ranked ballots used, second (and third, fourth, etc) choices are really important. This is only partially true. For a candidate to rally from behind, he needs to benefit from a systematic and massive report of the votes of other candidates. It likely won't happen here because members had to indicate all their choices at once. And sure, you can start arguing that Chong supporters might be more likely to support O'Toole than Bernier or Leitch based on policies, but remember that real life second choices are seldom that clear and "reasoned". You know, that's the same logic that a NDP voter can't have the Conservative as second choice. Except it happens.

I'll say this, I'm absolutely shocked that no candidate officially made a deal with another. Like telling his or her supporters to put another candidate as second choice. But hey, between this race and what Quebec Solidaire chose this weekend, it seems Canadian politicians don't like to make deals.

At the end of the day, I have no reason to believe Bernier isn't a popular second choice. And since he'll finish comfortably ahead after the first round, he will very likely not be caught and ultimately win. In my simulations, the only way Bernier loses is if the polls/fundraising data were wrong and wrong pretty much in every province.

Anyway, we'll see this weekend. Last time I tried to predict a leadership race (the PQ last year), I got the right winner (despite every single poll showing otherwise) but I was quite off in terms of percentages. At the same time, my biggest achievement in electoral prediction was when I had correctly called Stéphane Dion to win the LPC leadership. So far, leadership races have been good to me.