When a party's support increases in a province, how does it translate in each riding? The answer to this question is crucial as to which forecasting model to choose. So let's look at some graphs.

Let's look at the Conservatives (a party who experienced quite a lof of changes at the provincial level during the last 5 year), in Quebec (where the model usually performs at its best, both for the federal and provincial elections) during the last election (2008). The next graphs displays the swing at the riding level, sorted by the level of support during the previous election:

Remember that province-wide, the Conservatives dropped from 24.6% to 21.7%, thus a decrease of 2.9 percentage points, or -11.8%. Therefore, if the swing was uniform, we should see all the dots on the same horizontal line, at 2.9 (red line). On the other hand, if it was proportional, we should see the dots as forming a straight line decreasing (representing -11.9% of the previous level of support) (green line). As we can see, both assumptions would clearly be wrong. There is considerable amount of variation between the riding-level swings. You have an easy-to-understand graphical representation of how those two simplistic models could be wrong.

By the way, you get similar plots if you use other parties and/or provinces. Just to prove that I didn't use the one example that proved my points, here is a similar graph for the Liberals in Ontario, in 2008.

Here, I didn't bother showing the proportional change line, since it is obvious it wouldn't fit the data. Rather, I added the projection of the model. While I agree it isn't very clear that the model is performing better, what I would like you to notice is that the projected values are not on a straight line. Using regional effect, incumbency or the level of support last election, (for each party) my model is able to better forecast the outcome, without having to rely on an arbitrary assumption such as uniform or proportional changes.

Let's go back to the CPC in Quebec. Another aspect that might be surprising is the fact that the Conservatives actually increased their share of votes in some riding, despite a drop at the province-level. How to explain this? Well, first of all, individual effects could play a role, such as a new candidate or other specific events (for instance, what happenned to Maxime Bernier). Second of all, we are looking at shares of votes here, meaning that even though a party lost some votes, it could still increase its share if the other parties lost even more votes. For the model, I project every party and then make sure the percentages sum up to 100%. However, on e downside of the fact that riding sometimes go against the provincial trend is that my model could lead to weird results. Indeed, since I'm estimating my coefficients, the model could give parameters such that the support for one party would decrease in one rding when this party increases at the provincial level. This can be a potential problem but since I'm using more than one election, hopefully those weird effects will not biase the estimates.

At the end, it is really difficult to find evidence that the swing is proportional, especially in Ontario and Quebec. This is the reason the proportional swing model performs significantly worse than any other models, incuding the linear one, as shown in my methodology pdf. And this is why I find it weird when I see some websites basing their model on this assumption.

I'll post soon about how proportional models can even be worse for small of third parties.

A new poll from Ipsos-Reid was released today (and thus the interviews were made before the court decision regarding the electoral spendings of the Tories). The Conservatives are standing at 43%! The Liberals are at 27% and the NDP at 13%. In Quebec, the Bloc would enjoy a level of support of 41%.

Translated into seats, that would give the Tories as many as 169 seats, including 151 safe! Only 54 for the Liberals and 29 for the NDP. The Bloc would get 53 MPs and fall one short to be the official opposition for the second time in history. But be very carefull with this poll, as some of the regional breakdowns seem very odd, to say the least. For instance, the Tories are at 52% (!!!) in Atlantic. This is far above the huge poll from Angus-Reid released yesterday. Also, in Quebec, the Liberals are a lot higher than usual, and the NDP a lot lower. Small sample size is probably the issue here.

In any case, here are the riding-by-riding projections. For regular readers, since the poll show the Tories and the NDP so far from previous levels of support in Atlantic, I used a simple linear model for this region. Again, my model is not meant to be used for extrapolations as big as that.

I'll update the general projections using an average of all recent polls later this week (I'm expecting a new Ekos poll). But just so that you know, even the projections using an average start showing a Conservative Majority. Do you smell elections?

I think for the first time since I'm doing projections on this site, we have a new Angus-Reid poll (with a massive sample size, online though) giving a majority to the Tories when used in my model. While the overall votes intentions are somewhat similar to other recent polls, the big difference is the big lead the CPC enjoys in Ontario (43% vs 30% for the Liberals). Moreover, the Tories could get an even bigger majority by winning a lotof close races.

This poll also has the merit to provide estimates for every province, even the four Atlantic ones. This is nice information but not really helpful for my model since this one works at the regional level for the Atlantic or the Prairies (because usually we never have polls for PEI for instance). I'm still a little bit skeptical of some of the numbers in Atlantic Canada. For instance, the NDP would be 10 points below its results last election. That's a huge drop. Also, please remember that I'm assuming that all the "weird" efffects of the 2008 campaing (the ABC, St-John East for NDP, Cumberland for the Tories, etc) are either gone or halfway gone (for ABC for instance). It does affect the number of seats, especially for the Conservatives.

Overall this poll is just more bad news for the Liberals. And I now think that Harper must be thinking of the best way to not pass his budget. It's not that easy, because he needs an issue that would force all three opposition parties to vote against and take him down, but this issue has to be defendable during a campaing. If you ask me, coming back with abolishing public funding to political parties is probably the best way for Harper. Not only would all parties have to take him down, but this policy is not impopular if you look at the polls.

Here are the riding by riding projections.

Until now, my model was merging Alberta, Manitoba and Saskatchewan together. I was alllowing the coeffecient to vary depending on the province though. I decided to change that and to estimate the model for Alberta and the Praires separately. I used to merge the two because I was favouring the polls from Nanos and this pollster combines Alberta and the Prairies. But since I'm now using an average of all the recent polls, I don't need to do that anymore. Moreover, while Alberta seems pretty stable in recent years (with a complete domination of the Tories), Man/Sask experienced bigger variations from the other parties, in particular the Liberals. The new model is projecting more seats for the Liberals in the Prairies. SOmething that makes sense since the LPC increased from 17% to 24% in this region, while both the CPC and NDP fell

The new model is thus used for those projections and you can also use the new model yourself. Let me know if you find a problem.

Some comments about those latest projections. It's almost sad: more than 2 years after the last election, a lot of talk, a lot of ads, and where are we now? At almost the very same situations! Of course, the difference being that the Conservatives would start a campaign projected at 10 seats shy from a majority, something that hasn't been the case in the past. But still, things are pretty stable. However, this overall stability hides some real variations at the province-level. For instance, the CPC is now stronger than ever in Ontario but has declined a little bit in the West.

I'm still predicting no elections this spring. There will be a lot of noise around the budget, but at the end, one party will pass it along the Tories. Or this party will simply not have enough MPs for the vote. I'm predicting this party to be the NDP who will jump on the first measure they could back.

We finally got a new Nanos poll and the numbers are good (again) for the Conservatives. Overall, if added to the average with the other recent polls, the results remain similar: CPC around 146 seats, with a potential majority if a lot of close races won. The Liberals and NDP would basically not change much as opposed to the current House of Common.

But what if I only use the Nanos poll? The results are displayed as the picture of this post. The CPC would be very, very close to a majority. But the NDP would end up with 36 seats. How is that possible? after all, with the same poll, 308 would predict only 28 seats NDP seats. Well, again, two models, two methodologies. But in this case, I have to wonder: how on earth could the NDP lose seats with this new poll? After all, if we compare this poll to the 2008 results, we see:

- The NDP is up nationally.

- More importantly, in Ontario: the Tories are down (slightly), the Liberals are down and the NDP is up to 23.4%! Yet, according to 308, that would translate into only one gain for the NDP?? This seems pretty low to me.

- The Liberals are down almost everywhere. In particular they are down in Atlantic Canada, a region where they won a lot of seats. They are down in Ontario, and we know that you win elections by being high where the people are. Then they are up in the Prairies and BC, but still way too far from the Tories to expect to make bug gains. So there too, I have a hard time believing the Liberals would actually gain 7-8 seats if this poll was true.

So even when accounting for the drop of this party in the Prairies and in BC, I really fail to understand how the NDP could be projected so low (the NDP is down in the Atlantic as well, but all their seats won there were with a big enough margin). But of course, the problem for both our models is that we never saw the NDP as high as 23.4% in Ontario, so we don't really know what would happen. Still, I have a hard time explaining a 5-points gain in the great Ontario and only one more seat won. Especially when we know that with our current electoral system, the "winning-zone" is when a aprty approaches the 25% mark.

What do you think? Would the NDP gain more than 1 seat in this province with 23.5% of the vote? By the way, here is the pdf for every riding. Let me know if you find mistakes (again, especially for Ontario, we are in extrapolations territory and the model is thus more likely to make mistakes).

More good news for the Tories (or more job for them to dismiss the polls in the medias). After the Ekos poll showing a big lead for the Conservatives, we now have some kind of confirmation with the latest Ipsos poll showing a very similar lead. Those projections use those two polls. Here are some comments.

[update for the Harris-Decima poll]: 10 points lead for the Conservatives, but the NDP lower in this poll. If I use this poll along with the other two, it gives me: CPC 145, LPC 75, NDP 32, 0 Green (but two races now, including one in BC where HD puts this party really high) and 52 Bloc. The same story remains true though, with the CPC now in reach of a potential majority. Btw, funny to see that Ipsos has the NDP at 30% in BC, while HD has this party at only 19%. I know this is the margin of error, but it shows the importance of averaging polls, especially at the province-level.

While not in the "majority territory", the Tories are getting incredibly close. In particular, if you sum the safe wins (more than 5% margin) and the potentials (within 5% of the leader or leading by less than 5%), you have this party ata max of 162! Of course, for a majority, they would need to win approx. 70% of the close races. This would be a really, really high conversion rate. But let's not miss the big picture here: the polls are really positive for the CPC and really bad for the Liberals. And this is BEFORE being in an actual election. I think it might be important. Let's look at what happenned at the beginning of the last election. The source is the SFU study group on elections.

Prior to the elction, believe or not, Stéphane Dion was actually neck-to-neck with Harper. But then the election started, people started paying attention and the Liberals dropped. And they never recovered. Now let's look at what happened last year when Michael Iggnatieff saif he wanted elections as soons as posible. It happened in the fall:

Again, a surge of the Tories and a drop of the Liberals. Why am I showing you that? Simply because I wouldn't be incredibly surprised to see the same thing happening again. So imagine the potential lead for the Tories if they already are 15 points ahead now? Of course, the opposite is also possible: people will start paying attention to Iggy and will like him. But if the former event happens, remember this post lol

Ok more comments. The Liberals are low everywhere in those two polls. Ipsos is hard on them even in the Atlantic! If I were to use only this poll, the Liberals would hit a low of 67 seats! With those numbers, I really wonder if this party will vote against the budget or not.

Ok let's talk a little bit about the NDP. My competitor (308) always projects this party to be really, really low. On the other hand, DemocraticSpace (not active at the moment though) or myself consistently project this party at around 30 seats or more. Why this difference? Well many things. First of all, even by using simple models such as a uniform-swing model (which, if you read my methodology, is less efficient than my model but still gives a good idea of the overall results), you also get the NDP at more than 30 seats. All that to say that the NDP at 17% today is not the same as the NDP at 17% in 2004. Things have changed, the NDP votes is more concentrated. This is something that 308, without a proper model at the riding-level, cannot take into account.
Secondly, yes the NDP is lower in the votes intentions, but who benefits from this? Not the Liberals, and this is the key here. The party stealing most of the votes is the Green party. And because, as I showed here, the NDP became much more efficient in his votes, this party can afford to lose some votes to the Green. It isn't costing them actual seats, yet. It would be another story if those votes were going to the Liberals. Especially in BC and Ontario. At the end, I believe 308 underestimate the NDP and I overestime it. So just do an average of both.

The Greens are still projected at one seat, in Ontario. But we all know they are overestimated in the polls. So we'll have to wait and see. Finally, not a lot to say about the Bloc. They are still below 40%, but because the federalist parties are almost evenly split, they manage to gain a lot of seats.

The riding-by-riding projections are available here. Don't forget to let me know about the changes for the Atlantic provinces.

I've got a lot of comments for one of my previous post. Those comments were mostly about the Atlantic provinces and whether or not I was right in my forecasting. So I went back to the drawing board and made some (actually needed) adjustments. At the end, I believe my projections are more realistic. But at the end, because of various reasons, forecasting those 4 provinces is still less reliable or accurate than Ontario or Quebec for instance. Here is why:

- One of the problem is that the polls are conducted at the regional level. But the variation is usually different in every province (for instance, it is totally possible that one party increased in one province but decreased in the other one). So right from the start, we are facing a problem hard to overcome. On top of that, even at the regional level, the sample size is usually quite small.

- The big, big problem is the 2008 election where so much things were going on. In NF-L, you had the ABC (Anything But Conservatives) campaign of the former Premier. A campaing that was indeed very successfull. Thus, while the Conservatives did pretty good overall in the Atlantic, they dropped by a lot in this province; Other riding-specific things happened. For instance, the fact that Elisabeth May was running without a Liberal opposition. Or still in NF-L, there was the return of a NDP candidate in the riding of St-John-East which boost this party.

My model (as opposed to the other ones) is able to take into account and estimate those riding-effects. But the real question is more: are those effects still present? For instance, is the ABC campaign gone or not? Will the NDP still enjoy such a high level of support in St-John-East? Or are those effects gone?

- The answer is not clear and almost impossible to really assessed. So at the end, after many tries, I decided to go with a compromise: the effects are halfway gone. Therefore, if the Conservatives were (say) down 30 points in NF-L, they are now down only 15 points. The only effect that is completely accounted for is the Elisabeth May effect, for obvious reasons (for the Liberals, I use the level of support that the model would have predicted if they had run a candidate). Those are arbitrarily choices, I know it. But it seems to work better than assuming that all the effects are still there or that they are completely gone. If you think I'm mistaken, please feel free to let me know by commenting.

- While I use my estimated coefficients for those effects, I decided to go with a uniform-swing model for the remaining of the projections. After many tries, it appeared to me to actually work better. On top of that, for past elections, the Atlantic is one of the region where the uniform model actually performs quite well. I might switch back to a fully-estimated model later on, but for now on, I'll stick with this hybrid model.

So check the lastest projections (there will be new ones later today, using the latest Ipsos poll) and let me know if you think the percentages are more realistic.