The BC Liberals keep increasing in the polls according to the latest from Angus-Reid. The gain of 2 points (as opposed to the last AR poll; The BC Libs have been polled higher by Ipsos for instance), coupled to the small decline of the BC NDP, show us a slightly tighter race in BC. However, the 15-points lead for the NDP is more than enough to secure a comfortable majority.

The rise of the Lib is again mostly done at the expense of the Conservatives, as these two have been exchanging votes for a long time now (just look at the graphs). If you input the AR numbers into the simulator, you see an easy win for the BC NDP. If you don't want to do this, here are the detailed projections, riding by riding. You also have the probability of winning in every riding. Remember, the probabilities come from a 1000 simulations where we take into account the uncertainty due to polls as well as the electoral system. As for winning the election, the BC NDP, with this poll only, would have a 100% chance of forming the government. However, in the worst case scenario (i.e: one where the NDP would perform significantly worse than the polls and be "unlucky" or inefficient with its vote), the BC NDP could get only 46 seats. What this means is that if the BC Liberals could climb back slightly more in the polls, a NDP majority isn't necessarily guaranteed.

Please notice that it doesn't mean that the BC NDP has already won. A lot of things can happen between now and next May. But as of right now, the odds are more than in favor of the NDP. The probabilities are for now, given the current information, not for the future.

I have one main issue with the poll though: the BC Conservatives would get their best results in metro Vancouver! Honestly, it doesn't make much sense. They should be higher in the interior, especially around Kelowna for instance. If these numbers are indeed true, I'd have to revise my current model as this one is based on the past elections and wouldn't assume that the BC Conservatives are taking their support mostly from metro Vancouver.

En effet, si le sondage Crop-La Presse de la semaine passée a raison, Philippe Couillard, à la tête du PLQ, réussirait non seulement à placer son parti en première position, mais il le ferait en prenant des votes au PQ!

Cela peut sembler étrange, mais depuis 2003, le PQ et PLQ ont toujours évolué dans la même direction. Lorqu'un baissait, l'autre en faisait de même (2007, 2012), et vice versa lors de progressions (2008). Je n'ai réalisé cela que récemment, en estimant le nouveau modèle (qui tient compte de qui grimpe et de quel parti tombe). Couplé aux données sur les 2e choix lors de la dernière campagne (qui montraient qu'en gros, il n'y avait pas d'électeurs péquistes ayant le PLQ comme second choix, et vice versa), je pense qu'il est plutôt raisonnable de partir du principe que sous l'ère Charest, le PLQ et le PQ ne s'échangeaient ainsi pas de votes. Les variations observées lors des différentes élections étaient davantage dues aux mouvements de l'ADQ/CAQ. 2012 est un bon exemple où il n'est pas déraisaonnable d'imaginer que le PLQ a perdu environ 10 points à la CAQ, alors que le PQ perdait 3-4 points à QS et ON.

Ainsi, de voir que Couillard irait chercher autant de votes à la CAQ qu'au PQ représente une petite surprise. Ou alors, cela démontre que ces sondages hypothétiques sont à prendre avec un grain de sel. Après tout, quelles raisons pourraient expliquer que Couillard réussirait à convaincre des électeurs PQ de changer de parti, sans passer par la case CAQ? Est-il perçu comme moins fédéraliste? Je suis intéressé de connaître votre opinion dans les commentaires.

Quoiqu'il en soit, vous avez sûrement remarqué que le nouveau simulateur est disponible (à la même adresse URL qu'avant) et vous pouvez donc vous amuser à faire vos projections. Ce que l'on remarque vraiment, c'est le désavantage que subit la CAQ dans la transformation des % en sièges. Un désavantage déjà évident le soir de l'élection. Un problème majeur pour François Legault. Il se retrouve à la tête du 3e parti, et ce dernier a un vote bien plus dispersé que la défunte ADQ. Or, être dispersé est un avantage lorsqu'on est premier ou deuxième, pas troisième. Aux cours des années, l'ADQ était rendue bien plus efficace dans son vote. Pour Legault, il faut soit augmenter l'efficacité, soit passer devant les autres partis.

Si vous utilisez le simulateur avec les chiffres du sondage Crop, vous voyez que Couillard remporterait l'élection mais ne serait pas forcément majoritaire. Sans chef, la situation est quasi identique aux dernières élections, ce qui n'est pas forcément un bon signe pour le PQ!

As a gift for the New year, here is the simulator/calculator for the upcoming BC election.

You can now make your own projections for BC. Please notice that it doesn't include the probability of winning as I recently added to the model. It really only gives you the percentages in each riding as a function of the provincial voting intentions.

Let me know if you see any mistake.

I'm focusing a lot on BC recently. It's the next scheduled election and it allows me to try new things. If you have read my recent posts, you already know that one improvement to the model is to now account which parties are gaining and which ones are losing compared to the last election. For instance, are the Liberals going down because the NDP is up, or is it because the Conservatives are increasing? Depending on the situation, the geographical swing will not be the same. In BC, if the Liberals lose most of their votes to the CP, then they will drop massively in the interior. On the other hand, if they lose votes to the NDP, they will also be in danger in Vancouver and the lower mainland for instance.

The next improvement is to add uncertainty to the model. I still mostly see my job as transposing percentages into seats. But obviously, in order to do so, we need the correct percentages. I use the average of the polls to estimate the percentages, but as we've seen in recent elections, it can be quite wrong. Of course, for the more advanced readers of this blog, I also include the simulator (not yet for BC though, it's coming soon) which allows you to enter your own percentages.

But at the end, a lot of readers, especially during a campaign, simply come here and look at the projections. If I only display seats, it doesn't tell them whether this outcome is likely or not. So let's look at the sources of uncertainty and how I try to take them into account.

1. The polls.

The polls give us a snapshot of the current vote intentions. But these are given with ucnertainty. First of all, there is the natural statistical variation. Even if two polls are conducted using the same methodology during the same period, they will not necessarily give us the same results. And this is normal, especially with samples of only around 1000 respondents. The margins of errors are here for that (well, 19 times out of 20). So how can I include that in the projections? Well, I first average the polls (with some wights depending on whether I trust this pollster more, or if the sample size is bigger, if the poll is more recent, etc etc). Coupled with the sample size or these polls, this gives me a multinomial distribution. After that, I now simulate a large number of draws from this distribution. In other words, if I have the Conservatives at 35% in average, with a sample size of 1000, this party could be at 32% or 38%. Same for the other parties. So I randomly draw a 1000 times from this disbribution. In average, the Conservatives will still be at 35%, but they will sometimes be much higher or much lower. For every draw, I then run the projection model. In clear: I now run a 1000 projections instead of just doing it once as before.

Notice that there is a second source of uncertainty with polls: the methdology and "house effects". Not all polls are equal and some use internet-based survey, others use antomated phone calls, some reach landlines only, etc etc. The various methodologies have an impact on the results. In order to correct for this, I use weights or adjust the polls directly. But for now, let's ignore this fact.

2. The distribution or efficiency of the vote.

Even if we had the exact and correct vote intentions (or percentages), we still couldn't correctly predict all the ridings with 100% accuracy. Even by using the results of the election, I can't have perfect projections (I can be mostly right in average, but I'll sometimes overestimate a party by 2 points, and sometimes underestimate it by the same margin). Even after taking the region or the incumbency effect into account, there is still some variation. A party can be lucky and gets its vote out where it matters. During the last Quebec election, the Liberals not only got a result much higher than what the polls predicted, but their vote was very efficient. Indeed, they won a large number of the close races they were involved in.

Therefore, in order to account for this source of uncertainty, I add a random term to each projection, for each party in each riding. This random term is drawn from a truncated Normal distribution, with a range of -3% to 3%. Over the 85 ridings, the added random terms cancel out each other (for instance, I could project that the Liberals would be slightly higher than expect in one riding for one simulation, but this party will then be projected lower than expected in another one). Because the added random terms have mean zero, it doesn't change the average, but it allows for additional outcomes. Again, think about the Liberals in Quebec earlier this year: they almost won the election thanks to the fact they received more votes than what was expected by the polls, but their vote was also efficient in the sense that it got out where it counted.

By running a 1000 simulations (I could run much more, but for now, 1000 seems right and enough), I actually have a 1000 projections in each riding. The projections vary a lot, thanks to the variations in vote intentions (the percentages) and the disribution of the vote (the added random term). At the end, the most likely scenario is still the one I use to display on my site. This is the one where we simply average the polls. But we now have a full distribution of outcomes and we can calculate probabilities.

Why is that useful? Well, let's look at the latest Ipsos poll for BC. They have the NDP at 48%, the Liberals at 35% and the Conservatives at 8%, with the Green at 7%. If I simply enter these percentages into the model, I get the following projections:

BC NDP: 61 seats

BC Liberals: 22

2 independants

So a comfortable majority for the NDP despite the Liberals coming back to the mid 30's. I could then show you the projections in each riding, like here. But it wouldn't give you any sense of uncertainty. For instance, is there a chance that the Liberals could win? Or what are the chances the NDP will win a particular riding?

In order to answer these type of questions, we need the 1000 simulations. Here you can find the probability of winning a riding for each party. Please note that right now, for the simulations, I exclude the potential 2 independants. It'll be fixed soon. As you can see, most ridings are currently projected to go to one party a 100% percent of the time. What this means is that even by fluctuating the parties within their margins of error, and even by adding the additional source of uncertainty with the efficiency of the vote, a lot of ridings just aren't currently competitive. This is because the NDP's lead is just too big at the moment (and the Liberals and Conservatives are splitting each other). But if the Liberals were to climb back a couple more points, we could have a lot more close races.

Maybe a little bit surprising is the fact that the Conservatives aren't projected to win a single riding. Not even out of a 1000 simulations. This is because at only 9%, they are way too far behind the other two parties. Even in Kelowna where they are projected at 24%, there isn't a single scenario where they'd win this riding. So even by being at the top of their margins of error (around 10-11% province-wide) and having a greater than expected swing there thanks to vote efficiency, they would still not win it. Please, notice that this is using this poll from Ipsos Reid only. And this is for now. A lot of things can happen between now and next May.

Nevertheless, despite the general lack of competitive ridings, out of a 1000 simulations, we got a couple of interesting scenarios. For instance, there is one scenario where the Liberals would get pretty close, with 38 seats. How is that possible? Well, in this simulation, they would beat the polls and receive 38.5% of the vote, while the NDP would do worse than the polls with 44.5% (i.e: both parties are at the very extreme of their margins of error, even slightly further). On top of that, the Liberals would be very efficient in their vote, allowing them to turn 38.5% into as many as 38 seats, while the NDP would only get 47. Still a majority, but a much closer election. But a very unlikely scenario, one that happened one out of a 1000 simulations. But one scenario that can be totally excluded given the current information.

With the percentages from the Ipsos Reid poll though, there isn't a scenario where the NDP doesn't win (and doesn't win a majority. One way to see this is to see that the NDP has enough ridings they are projected to win a 100% of the time). However, if the NDP was to drop at 45% for instance, then there would be outcomes, from the 1000 simulations, where the Liberals could win. It'd would be rare and thus unlikely, but not impossible.

It was a long post, but I needed it to show you how the added uncertainty works. During the election, from now on, instead of simply displaying the seats, I'll display the probability of winning the election (as well as possibly, the probability of a majority governement). It might not be very useful in BC if the NDP keeps its current lead, but it'd be very useful for the next Quebec, Ontario or federal election.

I now have improved on the first version of the BC model. In particular, I now have some regional coefficients for the BC Conservatives as well. To achieve so, I looked at the 2009 results and observed where the BC Conservatives were higher. Still assuming that they took their votes from the BC Liberals, I managed to come up with some regional variations for the share of votes that the CP will take from the Liberals. In particular, the model now assumes that the Conservatives will take more votes in the interior that they would in Vancouver for instance. It make sense and I'm happy not to have to assume a uniform effect for the Conservative throughout BC.

There was one new poll last week, from Angus-Reid. The NPD is still largely ahead and would win a comfortable majority. However, the BC Liberals gained 3 points. At 29%, they are still far from being competitive, but I'm guessing they will any good news they can get. For the race to really become competitive, the Liberals have to go back above 35%. In order to do so, the first task would likely be to bring the Conservatives down. Even though this party is now polled at only 12%, far below where they were a couple of months ago, it is still a big increase since the 2009 election (where they got 2% of the votes with 24 candidates). An increase made essentially at the expense of the incumbent party. As with my previous projections, I still have no seat for the Conservatives. They are however highly competitive in some part of the interior, in particular in Kelowna.

The detailed projections can be found here.

Fell free to comment and let me know if you see anything that doesn't make sense. The model is still not completed. But it's getting there.

Canadians from many provinces will likely have an election sooner than later. Among them, we can think of Québécois (who just elected a minority governement) as well as Ontarians (minority governement with a Premier who just resigned). However, the next election actually scheduled is the upcoming BC one, in May 2013. This is why I'm very pleased to publish the first projections using the new (and first) BC model.

Keep in mind this a version 1.0. In other words: very preliminary. Making accurate projections requires a lot of work. So far, I've used past election data to estimte regional coefficients and I found a way to introduced the (almost new) BC Conservative party. Nevertheless, a lot of fine tuning will be required before the election (and before I actually release the model to the public as a simulator).

If you want to know the details of the model (in particular, how I introduced the BC Conservatives despite a lack of data from the past elections), read further down. For now, I'll simply talk about the projections themselves.

I'm using the latest poll from Angus-Reid. Polling in BC has been relatively stable over the last year, especially for the NDP who has consistently enjoyed a healthy lead. The BC Liberals have experienced some ups and downs with the Conservtives, who even took second place for a couple of months. The Liberals are currently second though. But being second with a 25 points deficit isn't much to brag about.

Combine the very large lead of the NDP, the fact that Liberals and Conservtives are essentially splitting each other, and the electoral system, and you get the following projections:

NDP: 73 seats

BC Liberals: 10

BC Conservatives: 0

Green: 0

Independents: 2

The NDP would win by a landslide. At this point, and except if all polls are completely wrong, the NDP would not only retake power after 3 liberals governements, but they would hold a comfortable majority. The BC Liberals would fall hard and most of the 10 seats they are currently projected to win aren't safe. The Conservatives would come close a couple of times, but ultimtely would likely fall short. They need to go back over the 20% mark in order to win seats. Please remember that these are seats projections without much uncertainty added. By the time of the election, I'll also provide an idea of the seats range for each party. So don't take the 0 seat for the Conservatives as clear evidence that this party wouldn't elect a single MLA. It simply means that this party isn't currently in a position of becoming even the official opposition. As for the two independents (in Detal South and Prince River North), I'm simply assuming they stay stable from the past election; I'll most likely revise this assumption later).

The detailed projections, riding by riding, are here.

Ok if you're still reading, it means you're interested in the details of the methodology. As usual, I used past elections in order to estimate regional coefficients in order to transpose the provincial swing into riding-level ones. BC is a little bit annoying because they changed their electoral map between 2005 and 2009. Not only that, but a complete transposition of the results (like the 2005 results with the 2009 map) isn't even possible because the boundaries of the voting reas have changed as well. For this reason and the fact that the provincial swings between 2005 and 2009 were very small, I only used the 2001 and 2005 elections to estimate the coefficients. I could use the 2009 as well by working at the regional level, but as I've said, the swings were so small that it wouldn't influence the coefficients very much. Unfortunately, while models from this site usually include an incumbency effect, it isn't currently the case in the BC model. The reason is because after the 2001 election, the Liberals had all but two seats. So it's almost impossible to disentangle the regional effects from the incumbency one (in some regions, the Liberals had all the seats, making the estimation impossible). An incumbency effect could be estimated using 2005-2009 but again, different maps, so no actual riding-level swing to use.

One tricky part for the upcoming election is to include the BC Conservatives. While this party technically existed back in 2005 and 2009, they were at that time a very marginal party. It's only recently that they have enjoyed polling well into the high 10's or low 20's. Therefore, past elections won't be very useful in order to help us project this party. However, the one evidence we have from polls is that the Conservatives are taking their support almost entirely to the Liberals (just look at the graph in the Angus-Reid polls to see that). Therefore, I proceed this way: first, I use the coefficients to project the NDP, Green and Liberals. Then, the Conservatives take a specific percentage of the Liberal vote. For instance, if the Conservatives are at 20%, they would take 20/45 of the Liberals votes from the last election (45% being the result for the liberals in 2009). There is thus no regional specific coefficient at the moment.

I spent the last couple of weeks trying to improve the methodology for every model. In particular, on top of regional and incumbency effects, I want to be able to account for where the votes is coming from. Specifically, if the Liberals lose votes to the NDP, it is different than when they lose votes to the Conservatives. The regional variation won't be the same. It's the same as what happened in Quebec at the federal level. When the Bloc was losing votes to the Conservatives, the Bloc was decreasing a lot in Quebec city. On the other hand, when the Bloc was losing votes to the NDP, they were pretty stable in Quebec city (where the NDP wasn't increasing as well as elesewhere in Quebec). It's the same logic applied here. The regional coefficients estimted from 2001 and 2005 show us the regional variations that would occur if the Liberals were decreasing and the NDP increasing. But the Liberals are mostly losing votes to the Conservatives this time around. So it's reasonable to expect the swings to look differents than if the BC Liberals were losing 20 points to the NDP only. Therefore, I account for this fact in the BC model. Namely, the liberal swing of -20 points is split between a loss of 6-7 points to the NDP and a loss of the remaining points to the Conservatives. The 6-7 Lib/NDP swing is thus transposed using the coefficients, while the remaining 13-14 points Lib/Conservatives swing is transposed using the assumption that the Conservatives are taking the same share of liberal votes everywhere. Of course, it'd be nice if the share of lib votes taken by the Conservatives was regionally adjusted, but for now, it's hard to really estimate this without pure guessing.

I'll keep working on the model. You cn also expect the full details os the new federal model soon. Then, it's gonna be time to start working on models for Quebec and Ontrio.

Lors de la dernières élections, j'ai souvent fait part de la difficulté d'incorporer ON, un nouveau parti, aux projections. Le problème étant bien sûr l'absence de données antérieures. Mais la tâche était encore plus compliqué pour son chef, Jean-Martin Aussant, étant donné sa plus grande visibilité.

Au final j'avais utilisé pas mal d'hypothèses pour accorder un bonus au chef d'ON. Et au final cela avait plutôt bien fonctionné dans le sens que nous avions Aussant comme étant compétitif dans son comté, et ce fut le cas.

L'hypothèse principale était qu'Aussant allait représenter environ 8-10% des votes de son parti si ce dernier restait aux alentours des 2-3%. Regardons alors les résultats officiels. Aussant a récolté 7869 sièges alors que son parti en a eu 82,855. Ce qui veut dire qu'Aussant a représenté à lui seul 9.5%.

Ainsi, mon système n'a pas trop mal fonctionné. Un peu de chance, mais pas seulement. Cela pourrait être utile à l'avenir.