Who is leading after the advance voting?

Election Canada published the number of people who cast a ballot during the advance voting: 4.7 mio. That's a 29% increase over the 3.7mio of 2015.

Given the fact that a bigger and bigger proportion of voters opt to vote during the advance voting period, I'm not sure the +29% guarantees an increased turnout. Just as a reminder, the increase between 2011 and 2015 was from 2.1 mio to 3.7mio, a 74% increase!

My guess is that the +29% will make it such that overall turnout will be similar or slightly lower.

Anyway, two polls have been published with data regarding people who declared they voted already. The first one, from Mainstreet, has a sample of 598. That's small and I'll therefore ignore the provincial numbers and focus on the country-wide ones.

According to Mainstreet, the Liberals received 35% of the ballots while the Tories were at 34%. The NDP was at only 13.9%, the Bloc at 6.8%, green at 6.4% and the People's Party at 2.9%.

Given that my projections put the Liberals around 30-31%, we can see a good turnout among Liberals. But this is to be expected, advance voting is for committed voters and the big parties have more of those.

David Akin for Global News had an interesting piece yesterday where he showed that the Liberals would have won way fewer seats based only on advance polls in 2015. Specifically, the Liberals would only have got 163 seats. While this can be interpreted as the Liberals didn't get their vote out during advance voting last time around, we also have to remember that the Liberals surged at the end of the campaign. Therefore, during the advance polls, this party wasn't as high as it was at the end. This is particularly true in Quebec. So having the Liberals at 163 seats during the advance polls is more of an indication that their turnout was actually good. My projections during that time had the Liberals actually in second position! And yes I know my projections in 2015 were fairly off, but with the correct vote percentages, my model wasn't that off. So even if we correct for the systematic mistake my model made, I'd have had the LPC at 140 seats around October 12th 2015, not 163. So to me, the 163 is a good indication the Liberals already got their vote out early last time around.

We thus shouldn't be surprised if we see the same pattern this year. At the very least, I wouldn't take the Mainstreet poll above as a clear indication that the polls are underestimating the Liberals. (Note: I asked David Akin if he could tell me what was the percentage of votes per party during the advance polls. This info is available but it's a ton of work to gather. I'll let you know and update this piece if he accepts to share). [Update: he shared, see at the end]

The other poll is from Angus-Reid. Among the 314 respondents who had already voted, they find that 34% voted CPC, 30% for the LPC, 19% for the NDP, 7% each for the Bloc and Green and 3% for the PPC.

We could average the two polls (keeping in mind that the Mainstreet one has a bigger sample size) and get:

CPC: 34.1%
LPC: 33.3%
NDP: 15.7%
Green: 6.6%
Bloc: 6.9%
PPC: 2.9%

Couple of remarks:

- It's close. The polls indicate a super close race and those two polls aren't changing that.

- Mainstreet (and IVR polls in general) have had the NDP lower than other polls (although it has mostly converged). It's interesting to see Mainstreet having the NDP lower among people who voted already too. My feeling is that there is a very strong house effect or mode effect.

- Both the CPC and LPC are doing better than their respective polling averages. That makes sense, as we discussed previously.

- Bloc is doing well. It seems the Bloc voters are committed and got out to vote. That's not always the case with the PQ in Quebec for instance.

- Green is not doing that great. We can clearly see that many Green voters are still unsure. With the NDP surging in the last week, my guess is many Green voters might want to keep their options open.

- I'm surprised by the PPC being so high. But maybe this is a different small parties. Its voters might well be few but super committed. Still, at 2.9% already, this is pretty good.

Mainstreet and Angus-Reid polls have not really agreed with each other during this election. The AR polls in particular have had a tendency to see the Tories more ahead. Alternatively, Mainstreet ahs been consistent in showing a large LPC lead in Ontario.

So instead of averaging the published numbers, let's compare the percentages among people who have voted to the polling numbers from each firm. For instance for Mainstreet, they had the Liberals around 30% last weekend. That means that the LPC would be at +5% (35%-30%) if we compare the polls and the advance voters. This is significant.

But using Angus-Reid, the Liberals are only beating their poll numbers by 1.6% (30%-28.4%), very similar to the 1.4% of the Conservatives.

So, who is right? Probably the average between the two, especially with such small samples. So here it is:

So the CPC and LPC are clearly getting their vote out. Is the LPC doing it better? In average it seems so, but it's entirely because Mainstreet found this to be true. Angus-Reid didn't. I really hope Akin will share the numbers so that we can compare the deltas here to the ones in 2015 (maybe the Liberals are just better at this game, always).

Smaller parties are doing worse. A clear indication that many of their voters are not fully decided yet (or that they don't have the machine to get the vote out).


So David Akin shared the numbers:

So let's see, the Liberals were around 35% in the polls back then, that means the Liberals performed at +2.6 during the advance poll. Conservative were at 31%, so that means +2 for advance polls, NDP at -2.7 and Green at -1.6%.

All in all, we see the same pattern as this year with the two big parties locking in more votes than the polls would indicate at this point.

So I see no reason to see the table above as any indication that the polls are off or the Liberals underestimated. Quite the contrary, the numbers are in line with CPC and LPC being very close to each other.

Now, let's see if there is a late momentum for any party.

Projections update October 15th: the NDP getting more competitive in BC and the Prairies

While advance voting is over (and I discuss here, in French, why an increase in advance voting won't necessarily translates into a higher overall turnout; I also discuss how turnout can affect the projections), the polls continue to see a rise of the NDP. This morning Ipsos published a poll with the party of Jagmeet Singh up 5 points while Mainstreet's daily numbers -behind a paywall- show a smaller rise (it's logical, Ipsos last polled further back than Mainstreet).

While the conventional wisdom is that a rising NDP is bad for Trudeau, I find the answer to be more complicated. The NDP in particular is becoming more competitive in BC and the Prairies and it's the Conservatives that are hurt. Also the Green. This party already didn't have a great campaign, the NDP surging at the end is really not helping.

So anyway, here are the latest projections:


Riding by riding:

L'importance de la participation

L'importance de la participation
Le vote par anticipation est officiellement terminé. Nous n'avons pas encore les chiffres pour les 4 jours, mais Élection Canada a indiqué une hausse de 25% par rapport aux deux premiers jours en 2015. Est-ce que cela veut dire que l'on se dirige vers une participation encore plus élevée que le 68.3% de 2015? Cela serait remarquable vu que cette élection avait déjà vue une hausse importante par rapport au 61.1% de 2011.

Cependant, les choses sont probablement plus compliquées. Tout d'abord, il faut bien réaliser que le vote par anticipation est en hausse depuis des années. Les diverses agences électorales du pays ont rendu ce vote par anticipation de plus en plus accessible. Et il semble que les Canadiens soient de plus en plus intéressés par voter en avance.

Au niveau fédéral, la participation par anticipation ne représentait que 10.9% des électeurs en 2008. Cette proportion avait augmenté à 14.2% en 2011 avait de bondir à 20.8% en 2015! Une hausse majeure. Cependant la participation totale, bien qu'ayant augmenté, n'était passé que de 58.8% à 61.1% et finalement 68.3%. Ainsi on peut voir que la proportion de citoyens ayant décidé de voter par anticipation a doublé en 7 ans mais la participation générale n'a augmenté que de 16% (pas points de pourcentages).

On a vu la même chose au Québec. La participation par anticipation n'était que de 10.23% en 2007 mais elle était à 17.9% en 2018. Or, la participation générale a elle baissé de 71.28% à 66.45% si l'on compare les mêmes années. En proportion, cela veut dire que 14% des électeurs en 2007 l'avaient été par anticipation alors qu'il s'agissait de 26.9% en 2018.

On avait vu une massive hausse en CB en 2017 du nombre de votes par anticipation mais la participation gĂ©nĂ©rale n'avait que peu bougĂ© par rapport Ă  2014.

On va attendre de voir les chiffres pour les 4 jours cette année mais il faut se souvenir de cette tendance à la hausse de la proportion d'électeurs qui choississent de voter par anticipation. Rien ne garantit une hausse de la participation générale.

Participation et projections

Pourquoi est-ce que je m'intéresse à cela? Car la forte hausse en 2015 avait en partie expliqué la grande victoire de Trudeau. Non seulement son parti avait fait mieux que les sondages (surtout au Québec), mais ses votes s'étaient traduits en davantage de sièges que prévus. En effet, même en utilisant les vrais pourcentages de vote de l'élection, mon modèle de l'époque n'aurait jamais prédit le PLC à 184. Le modèle l'avait plutôt à 170. Je ne suis pas le seul dans ce cas. Canadian Election Watch a aussi montré que même avec les vrais pourcentages, le modèle sous-estimait le PLC assez significativement en Ontario et CB. Eric Grenier, avant d'être 100% à la CBC, aurait prédit 154 sièges PLC avec les vrais pourcentages! Une sous-estimation massive.

Cela montre bien que quelque chose s'est passé en 2015. Le vote Libéral a changé, est devenu plus efficace. Et mes recherches montrent que la hausse de la participation peut en partie expliquer cela (il y a définitivement une corrélation entre la hausse de la participation et la hausse du PLC). Cette hausse avait été particulièrement importante chez les jeunes, une population qui soutenait le PLC massivement. Cette hausse avait ainsi permis au PLC de remporter des sièges inatendus.

Cela veut dire que si la participation de 2015 devait ne pas se répéter cette année, en particulier chez les jeunes, il se pourrait que les Libéraux perdent davantage de sièges que prévus. J'ai un modèle à l'essaie qui va tenir compte de cela, mais j'attends de voir les chiffres par anticipation avant de décider si je vais l'utiliser ou non.

Voilà, je mettrai à jour les projections ce matin, une fois que nous aurons les nouveaux Mainstreet et Nanos. Je m'attends à ce que cela reste une course très serrée.

Projections update, October 12th 2019

With Campaign Research publishing today, we pretty much had every single pollster providing new numbers in the last few days, with the exception of Abacus. In Quebec the Bloc is now ahead (in seats, not -yet?- in votes) while the NDP is rising in BC and Ontario.

This blog will be short, it's a simple projections update. No blah blah or analysis. So here are the numbers:


Riding by riding:

   Proj 12.10.2019 by bryanbreguet on Scribd

Projections du 11 octobre: Le Bloc premier au Québec

Techniquement le Bloc est premier à égalité, mais j'espère que vous allez tolérer la petite clickbait en bon français. Surtout que la tendance est définitivement positive pour le Bloc.

Après 2011, je ne suis pas sûr que je pensais jamais écrire sur une telle projection. Lors de cette élection, beaucoup pensait que c'était la mort du Bloc. Mais 2019 est une toute autre élection. Et avec Blanchet qui a gagné le premier débat et possiblement le 3e (donc 2e en français; Je pense perso que Bernier a gagné et Blanchet 2e, mais le PPC ne va pas faire mal au Bloc), je pense que les proejctions auront le Bloc seul en tête d'ici peu.

Voici les projections. Le texte and analyse continuent après.

Les projections:

La carte:

En détails:

   Proj 11.10.2019 by bryanbreguet on Scribd

La hausse du Bloc est impressionnante, tel que vous pouvez le voir. En termes de sièges, mon modèle a aussi commencé à faire de petits ajustements par régions au Québec. Les quelques sondages fournissant cette information ont montré un PLC très concentré dans la région de Montréal et un Bloc plus élevé que prévu en région. Cela fait changer de bord quelques comtés.

Si l'on regarde par région, on voit le Bloc dominant dans le 450 maintenant:

Alors que le PLC ne garderait qu'un siège dans la région de Québec.

Et oui Maxime Bernier est projeté gagnant.

Reste du Canada

Si le PLC a commencé sa chute au Québec depuis un petit bout de temps, il semble que ce soit maintenant aussi le cas dans le reste du Canada. Plusieurs sondages (Forum, Mainstreet, Dart aujourd'hui mais pas inclus dans ces projections car ils n'ont pas encore publié les chiffres régionaux) ont tous le PLC sous les 30%. Et c'est le NPD qui semble en hausse, probablement en partie en raison de la victoire de Singh lors du débat en anglais.

Cela nous donne une situation qui serait très, très instable. PLC+NPD+Verts ne donnerait pas une majorité! PCC+Bloc donnerait 169, soit une majorité mais sans speaker pour la Chambre. Absolument aucune idée qui serait Premier Minister et qui pourrait passer un vote de confiance. Ce serait très probablement un parlement très court avec une nouvelle élection dans l'année.

[Sondage après débat] Qui a remporté le débat des chefs?

[Sondage après débat] Qui a remporté le débat des chefs?
Vous savez comment ça fonctionne. Deux questions.

Qui a remporté le débat des chefs?
Justin Trudeau
Andrew Scheer
Jagmeet Singh
Yves-Fançois Blanchet
Maxime Bernier
Elizabeth May
Make a Poll with Poll Maker

Si l'Ă©lection Ă©tait demain, pour qui voteriez-vous?
Le Parti Libéral
Le Parti Conservateur
le NPD
Le Bloc Québécois
Le Parti Vert
Le Parti Populaire
Un autre parti

Projections update, October 10th: It's starting to move!

Are people finally paying attention? Did the French (TVA) and English debates finally change things? With Blanchet winning the former and Singh the latter, it does seem to have shaken things up. It's especially obvious in Quebec where we now have polls putting the Bloc above 30% and ahead of the Liberals, but it'll likely affect the ROC soon.

The projections have, maybe surprisingly, moved very little. Today's polls are actually not bad for the Liberals because both Nanos and Mainstreet give this party a good lead in Ontario and Nanos still sees the Grits well ahead in La Belle Province. I believe Nanos to be dead wrong in Quebec but I can't let my subjective opinion influence which poll is included.

Still for Quebec, while Forum, Leger and Mainstreet all see a massive rise of the Bloc and decline of the LPC, Nanos, Ipsos and Innovative released numbers fairly recently with the Liberals still ahead. Given that I don't want to only include the polls published in the last two days, my polling average in Quebec hasn't moved as much as you may think. I truly believe the projections will change dramatically in the next few days. In the meantime, don't hesitate to use the simulator to make your own projections.

Before moving onto the numbers, here's how different the numbers in Quebec are between Nanos and Mainstreet (the numbers of this morning). If I only use Nanos, I get 53 seats LPC, 11 Bloc. With only Mainstreet, this is 24 LPC, 40 Bloc!



Riding by riding:

   Proj 10.10.2019 by bryanbreguet on Scribd