Latest national poll median date: October 20
Projections reflect recent polling graciously made publicly available by pollsters and media organizations. I am not a pollster, and derive no income from this blog.

Monday, September 30, 2019

Projection Update: Mainstreet 9/27-29 and EKOS 9/26-29

The following polls have been added to the model:
Mainstreet's 9/27-29 national poll (national results only; current weight among national polls: 36%)
EKOS' 9/26-29 national poll (current weight among national polls: 20%)
For a full list of included polls, see previous Projection posts.

The main movement in the EKOS poll is a fall for the Tories and a rise for the Greens. The former helps the Liberals, especially because they expand their lead in ON to double digits. The Green increase, a few points across the board, does not lead to major seat implications in the projection. Before anyone gets too excited about the Greens' second place in QC, I'll point out that:
- the sample size there is 193;
- EKOS' two previous polls were outliers in having the Greens at 14% in QC; and
- Greta Thunberg was in Montreal for the march this weekend, when the poll was conducted.

The Bloc drops quite a bit in the EKOS poll. We've also seen a less dramatic pullback for them in the Nanos tracking, though they are still high in Mainstreet's numbers. For the NDP, it's a mixed bag: a very weak number in ON, but a very good one in BC.

There is no meaningful movement in the Mainstreet tracker (except maybe the NDP down slightly).

Projection as of the latest national poll (midpoint: September 28)
LIB - 155.4 (33.6%)
CON - 141.8 (35.4%)
NDP - 19.0 (12.5%)
BQ - 17.5 (4.9%)
GRN - 3.3 (10.4%)
IND - 0.6
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seats changing hands since the last projection:
- In QC, LIBs get Saint-Jean back from the BQ.
- In ON, LIBs get Kitchener South--Hespeler and Richmond Hill back from CONs.
- In ON, LIBs regain Windsor West from the NDP.
- In BC, the NDP gets Courtenay--Alberni back from CONs.

Projection Update: Nanos 9/27-29

The following poll has been added to the model:
Nanos' 9/27-29 national poll (national results only; current weight among national polls: 28%)
For a full list of included polls, see previous Projection posts.

There is no meaningful movement either in the Nanos tracker (except maybe the Bloc being down slightly) or the projection. It's a boring update!

Projection as of the latest national poll (midpoint: September 28)
LIB - 151.7 (33.4%)
CON - 143.9 (35.7%)
BQ - 19.1 (5.1%)
NDP - 19.0 (12.6%)
GRN - 3.3 (10.0%)
IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seats changing hands since the last projection:
- In QC, LIBs retake Longueuil--Saint-Hubert and Hochelaga from the BQ.
- In ON, CONs regain Kitchener South--Hespeler from LIBs.

Sunday, September 29, 2019

Trends 2019: Fourth Instalment

For new readers: these trends are different - and I would say more meaningful - than the ones you might see elsewhere, since they are retroactively recomputed* when old data becomes available. Thus, my "Trends" posts avoid showing "fake" variations caused by polls coming out with different lags. More details were provided in the first instalment of this series. My post on comparing models highlights other advantages of my approach.

What was the effect of "blackface," and what has happened to voting intentions in its aftermath? It's time for another trends post! As usual, all numbers in this post reflect my turnout adjustment of CON +1.5, LIB +0.5, NDP -1, GRN -1.


We finally see some movement!

- The Liberals lost 1.5-2 points in the immediate aftermath of "blackface." A modest recovery has occurred, but they are still about 1 point short of where they were before the revelations.

- Right after "blackface," the Tories saw a modest rise in popular support - so small that it might have been just noise. Nevertheless, the Liberal fall helped them briefly gain a tiny seat lead (though this would not have happened without the turnout adjustment). Conservative gains, whether real or due to statistical noise, have been entirely erased, and Scheer's party is now at its lowest popular vote support (and near its lowest seat projection) since my projections started.

- The Liberal seat lead in the past 3 days has hovered around 10. That is tiny: with these numbers, the Tories would have a 40% chance of winning the most seats in an election today. For the actual election, it's almost 50:50.

- The NDP has been steadily regaining the ground lost during the pre-writs period, and the recovery is now almost complete.

- Despite the NDP's gains, it has marginally fallen into fourth place in the seat projection due to the Bloc's abrupt rise post-"blackface."

- Both the NDP and the Bloc have extremely few safe seats, so despite them flirting with the 20-seat bar in the projection, they are both still in danger of not reaching the requisite 12 for official party status.

- The Greens gain when there are external events: NDP's pre-writs struggles, "blackface" and the climate strike. However, the intervening drops mean that they are still below their July levels.

Quite remarkably, comparing the latest numbers to the August 11 ones, we see that all four national parties are down marginally both in seats and in popular support - only the Bloc is up. This is a sign that Canadians are not drawn to any of the choices, and greater numbers of those that have the option of a palatable and meaningful protest vote (Quebecers) are choosing it.

Why is this happening? As I have commented on Twitter, I believe that all four parties' platforms are inferior to current policy; in other words, they're all proposing to make the country worse. More relevant for most voters, however, is the fact that all four leaders have shown serious shortcomings:

- Trudeau's ethical lapses are well-known, and the "blackface" revelations are yet another reminder of his lack of judgment. He appears to have no vision for the country, being simply on a mission to simultaneously outbid the Tories on financial goodies and pay at least lip service to all of the NDP's priorities.

- Scheer is entirely uninspiring (bringing back ineffective Harper-era policies), tone-deaf on climate (building highways helps reduce emissions?!), and lies about his own work experience (as well as basic arithmetic). His response to the "blackface" episode - immediately going to "racist" and "unfit to be Prime Minister" - reinforced his sophomoric image.

- Singh has shown himself time and time again before the campaign to not know the issues and not have the organizational skills to lead a party. However, he has performed better during the campaign (notwithstanding that some of his announcements were squarely within provincial jurisdiction), which may explain the NDP's recovery.

- May continues to show poor judgment (recall her ill-fated attempt to distinguish between "separatist" and "sovereigntist"), and the Green Party's platform costing turned out to be completely amateurish, exposing the party as wholly unprepared to govern - and perhaps even to be a serious partner for a minority government.

So, um, yeah. Normally, as an economist, I want people to think more about the issues. But this time, the choice is so depressing that we might as well all sit back and go fully into "political junkie" mode, the future of the country be damned.


*Changes made on or after Sept. 25 are only reflected back to the start of the campaign (Sept. 11). These include the re-weighting of standalone regional polls (i.e. regional numbers not part of a national poll) and the Guilbeault baseline adjustment in Laurier--Sainte-Marie. These only make a barely visible (if at all) difference on the graphs.

Mapped Projection Update: Mainstreet 9/26-28

The following poll has been added to the model:
Mainstreet's 9/26-28 national poll (national results only; current weight among national polls: 42%)
For a full list of included polls, see previous Projection posts.

This poll is good news for the Liberals, who draw into a rough tie. Unlike Nanos, Mainstreet only sees a very slight and insignificant rise in Green support.

The NDP and the Bloc both rise to their highest level in Mainstreet polling since at least the start of the summer. As a result, the Bloc is at a new high in the seat projection, and the NDP is at a one-month high.

Projection as of the latest national poll (midpoint: September 27)
LIB - 151.7 (33.5%)
CON - 143.3 (35.7%)
BQ - 19.8 (5.2%)
NDP - 19.0 (12.6%)
GRN - 3.2 (9.7%)
IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seats changing hands since the last projection:
- In QC, the PPC takes Beauce from CONs.
- In QC, the BQ takes Longueuil--Saint-Hubert and Hochelaga from LIBs.
- In ON, LIBs get Kitchener South--Hespeler back from CONs.
- In BC, the NDP gets Cowichan--Malahat--Langford back from GRNs.

Here are a map of the projection and a potential (pre-Speaker election) House of Commons sitting chart:


Projection Update: Nanos 9/26-28

The following poll has been added to the model:
Nanos' 9/26-28 national poll (national results only; current weight among national polls: 32%)
For a full list of included polls, see previous Projection posts.

Today's Nanos tracker shows the Tories and Liberals stable, with the NDP down, the Bloc up, and the Greens slightly down but still at a high level.

The projection moves only slightly. Some of the movement is due to shifts: Bloc up, NDP down; some of it is due to the most recent Nanos numbers gaining more weight as all other polls are 2 or more days older (the Mainstreet tracker was not publicly released yesterday): Tories and Greens up, Liberals down.

Projection as of the latest national poll (midpoint: September 27)
LIB - 151.2 (33.5%)
CON - 146.8 (36.2%)
BQ - 18.7 (5.1%)
NDP - 17.1 (12.0%)
GRN - 3.3 (10.0%)
IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seats changing hands since the last projection:
- In QC, the BQ retakes Jonquière from CONs.
- In QC, the BQ regains Saint-Jean from LIBs.
- In BC, GRNs regain Cowichan--Malahat--Langford from the NDP. (The model has three parties - these two plus the Tories - within 0.5%.)

Saturday, September 28, 2019

Comparing 2019 Seat Projections

In August, I wrote a post about the methodology of various poll-based projections you'll find around the web. This post takes a look at what these models have been showing, and how to explain the differences between them. I will only review projections that provide both seat counts and vote shares; those showing only the former are excluded from the analysis.

I have been keeping track of this graph for over a month:
Quito Maggi's (President and CEO of Mainstreet Research) projection was posted on Twitter on Sept. 27; the national vote shares were stated as CON 36, LIB 34, but the population-weighted regional breakdown is consistent with a tied national vote. Forum's projection is based on a Sept. 19-21 poll. All others are aggregator projections posted on Sept. 27 or 28. TCTC = Too Close to Call; CEW no adj. = Canadian Election Watch without turnout adjustment.

The first thing to note is that all these projections agree: the Tories are tied or better in terms of the popular vote, but the Liberals lead in terms of seats.

Beyond that, there are differences in both: (i) the seat projections, even for models with nearly identical vote share projections, and (ii) the vote share projections themselves. I will address these separately.

Seat Projection Given Vote Share
To compare the seat projections for the two main parties, link the two red dots on the graph (corresponding to my projections with and without turnout adjustment), and extend that line out. Models on this line give similar results as mine conditional on the popular vote, while models above are friendlier to the Liberals, and those below are friendlier to the Tories.

Clearly, 338Canada and the CBC Poll Tracker give very similar results to my model, controlling for the vote share. This has quite consistently been the case since August, with the CBC Poll Tracker bouncing around just a bit because it gives the number of seats projected ahead while 338Canada and I give the sum of seat win probabilities. Given that these are by far the post popular seat models, my personal opinion is that Canadians are being well served. This line suggests that the Tories need to win by about 3% to even the seat count.

At this moment, Too Close to Call is below the line, though it had been above by roughly the same amount just before "blackface," and roughly on the line before that. Therefore, I would include it in the same group of models as 338Canada, CBC Poll Tracker and mine.

Lean Tossup has consistently been well above the line. This is due to at least two, and possibly three factors:
- Restrictive poll inclusion policy
- Use of riding polls
- (Possibly) Use of demographics
The use of riding polls likely causes Lean Tossup to put a very high weight on Mainstreet results, since almost all riding polls are conducted by Mainstreet (and there are a lot of them: around 40 so far); this is compounded by a restrictive poll inclusion policy. It is therefore unsurprising that Lean Tossup's projection is near Quito Maggi's. Because Mainstreet is especially favourable to the Liberals in crucial Ontario, this increases Liberal vote efficiency and pushes those projections up the graph.

Calculated Politics has shown no consistency: it has been alternately slightly below the CBC-338-CEW line a couple of weeks ago to almost as far above the line as Lean Tossup just a few days ago. I have not investigated why this is the case.

Finally, Forum Research is well below the line 338-CBC-CEW line. This is a head-scratcher as the regional distribution of support in the poll on which this projection is based is very good for the Liberals. Something "interesting" might be going on with Forum's seat model.

Vote Share Projection
Note: For this section, ignore pollsters' projections (since they use only their own data).

Because poll aggregators mostly use the same data, the vote share projections (before any turnout adjustment) should be very similar. And that's what we broadly see, with the Tory lead ranging from 0.2% to 1.4%.

Close followers of this blog will have noticed, however, that my polling average - even without the net C+1 turnout adjustment - often leans more Conservative than other websites', as it does now. This is due to the interaction of the two following factors:
1. My vote share model assumes that each pollster has an unknown house effect. As a result, it tries to stabilize the weights placed on different pollsters: those with recent polls will still have higher weight, but not by nearly as much as without the stabilization.
2. Nanos and Mainstreet, the two pollsters with daily tracking, tend to exhibit a Liberal lean.

As a result, when polls from firms other than Nanos than Mainstreet are a bit older, they carry higher weight in my average than in others' averages; when all pollsters have come out with recent data, this gap should in theory disappear. And that's exactly what we saw right after blackface: other projections converged to mine after DART, Ipsos and Angus Reid released new numbers (along with almost every pollster active on the national vote intention scene), as evidenced by the thread below:

This is where I will shamelessly advertise my projection trends for the popular vote, which I believe have two meaningful advantages over others you might see:
(Note: I'm specifically referring to the up/down trends. I'm NOT saying that the levels of my popular vote estimates are the most accurate - the turnout adjustment is merely an educated guess that could be wrong this year. And I'm NOT talking about seat projections at all.)

1. The trends are "real" because they attenuate variations due to the mix of pollsters that happen to have recently conducted a poll. Other aggregators do not stabilize the mix of pollsters in their averages (or do so to a much lesser extent than I do), while pollsters' own trends use much smaller samples and are more affected by statistical noise.

2. Stabilizing pollster mix allows me to more aggressively weight recent polls without generating undue volatility (this is done by assigning negative weight on the recent pollsters' older polls). The "blackface" episode illustrated this starkly: my polling average moved swiftly on Sept. 21 and then stayed stable while the bulk of the post-"blackface" polling data came out, while others required more time (2-3 days) to eventually move as much as mine. And yet, the trends my model produces are quite stable when there is no major news while other projections show, for example, Liberal drops every time Angus Reid releases a poll.

Now that I've whet your appetite, keep an eye out for updated projection trends, probably coming tomorrow!

Projection Update: Nanos 9/25-27 and ON Regional Polls

The following polls have been added to the model:
Nanos' 9/25-27 national poll (national results only; current weight among national polls: 22%)
Mainstreet's 9/23-26 GTA breakdown
Forum's 9/25-26 Toronto poll
For a full list of included polls, see previous Projection posts.

The big news from the Nanos poll is a notable jump in Green support, at the expense of the Liberals. One has to wonder if this is just a transitory effect of the global climate "strike." Nevertheless, ironically, this shift is helping the Tories in the projection for now (as well as the Tories' own marginal increase).

Despite their near-2% drop in support, the Liberals don't lose too much ground in the projection because:
- Mechanically, the date moving one day ahead reduces the weight on yesterday's Mainstreet tracker, which was rather poor for the Liberals. Of course, this will be reversed this afternoon unless Mainstreet has them up significantly.
- The Bloc is also down quite a bit - and in fact their drop in QC is bigger than the Liberals' national drop.

The Greens are at their highest popular support level in the projection since mid-August. The Mainstreet and Forum polls are discussed in an update to the polling-related adjustments post.

Projection as of the latest national poll (midpoint: September 26)
LIB - 153.0 (33.6%)
CON - 145.1 (36.0%)
NDP - 18.2 (12.4%)
BQ - 17.5 (5.0%)
GRN - 3.2 (9.8%)
IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seats changing hands since the last projection:
- In QC, CONs retake Jonquière from the BQ.
- In ON, CONs regain Kitchener South--Hespeler from LIBs.

Friday, September 27, 2019

Projection Update: Research Co. 9/24-26

The following poll has been added to the model:
Research Co.'s 9/24-26 national poll (current weight among national polls: 12%)
For a full list of included polls, see previous Projection posts.

We weren't done for the day! A Research Co. poll is out, and shows the Liberals with a 3-point lead. The regional breakdown for this poll, though, isn't terrible for the Tories, as they are lower than usual in the Prairies. As a result, they don't lose that much ground relative to the Liberals in the seat projection.

This poll has another piece of bad news for the Greens: 11% in BC. We are nearing the point where the Greens' number in Atlantic Canada surpasses their number in BC (both are now between 14% and 15%, or one point more without the turnout adjustment). The problem is that there is less vote division in Atlantic Canada due to the NDP being much weaker there than in BC, so BC support is more useful than Atlantic support for the Greens.

Projection as of the latest national poll (midpoint: September 25)
LIB - 154.3 (34.0%)
CON - 143.6 (36.0%)
NDP - 18.2 (12.5%)
BQ - 18.1 (5.1%)
GRN - 2.9 (9.1%)
IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seat changing hands since the last projection:
- In QC, LIBs get Saint-Jean back from the BQ.

Projection Update: Mainstreet 9/24-26

The following poll has been added to the model:
Mainstreet's 9/24-26 national poll (national results only; current weight among national polls: 35%)
For a full list of included polls, see previous Projection posts.

For the second consecutive day, Mainstreet shows the Liberals losing ground, now being marginally farther behind the Tories than they have been in public Mainstreet polling since late March. This is excellent news for the Tories in the projection, even though they are actually down marginally from yesterday's Mainstreet poll. This poll is also good news for the Bloc.

Despite both the Mainstreet and Nanos tracking polls moving against the Liberals today, the Abacus poll from this morning means that they are still up on the day.

Projection as of the latest national poll (midpoint: September 25)
LIB - 152.3 (33.7%)
CON - 145.0 (36.2%)
BQ - 18.4 (5.1%)
NDP - 18.3 (12.5%)
GRN - 3.0 (9.3%)
IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seats changing hands since the last projection:
- In QC, the BQ retakes Jonquière from CONs.
- In QC, the BQ regains Saint-Jean from LIBs.
- In ON, CONs get King--Vaughan, Northumberland--Peterborough South and Richmond Hill back from LIBs.

Projection Update: Abacus 9/23-26, Nanos 9/24-26, Mainstreet's Laurier--Sainte-Marie Poll

The following polls have been added to the model:
Abacus' 9/23-26 national poll (current weight among national polls: 25%)
Nanos' 9/24-26 national poll (national results only; current weight among national polls: 21%)
Mainstreet's 9/24 Laurier--Sainte-Marie poll
For a full list of included polls, see previous Projection posts.

This Abacus poll is the first one from Abacus showing a Liberal lead since late May. Furthermore, the regional breakdown is very good for the Liberals, with a lead in BC and widening ones in QC and ON. This moves the projection significantly towards the Liberals.

The Nanos poll shifted slightly against the Liberals, but the silver lining for them is that the Bloc is also down.

The Laurier--Sainte-Marie poll shows what I've been suspecting since making the model in late July: that Guilbeault would enjoy a sizable personal effect. I have made a baseline adjustment now that there is a quantitative basis for it. Obviously, this moves the riding away from the Bloc.

You'll note that the Greens move up in the vote share projection thanks to a good number from Nanos, but down below 3 for the first time in the seat projection. That's due to their weak 12% in BC in the Abacus poll. The NDP is up to 7 seats projected ahead in BC, their best in a long time.

Projection as of the latest national poll (midpoint: September 25)
LIB - 156.2 (34.1%)
CON - 142.5 (35.9%)
NDP - 18.1 (12.5%)
BQ - 17.3 (5.0%)
GRN - 2.9 (9.3%)
IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seats changing hands since the last projection:
- In QC, CONs retake Jonquière from the BQ.
- In QC, LIBs retake Laurier--Sainte-Marie and Saint-Jean from the BQ.
- In ON, LIBs get Kitchener South--Hespeler, Richmond Hill, Northumberland--Peterborough South and King--Vaughan back from CONs.
- In BC, the NDP gets Cowichan--Malahat--Langford back from CONs.

Thursday, September 26, 2019

Projection Update: Mainstreet 9/23-25 and Riding Poll, Nanos 9/11-24 ON Breakdown

The following polls have been added to the model:
Mainstreet's 9/23-25 national poll (national results only; current weight among national polls: 36%)
Mainstreet's 9/23 Rimouski-Neigette--Témiscouata--Les Basques poll
Nanos' 9/11-24 ON breakdown
For a full list of included polls, see previous Projection posts.

Today's Mainstreet poll shows the Tories expand their lead by 1.1%, which offsets the 2% movement toward the Liberals in the Nanos tracker. Moreover, the day that was dropped, 9/22, does not appear to be a good Liberal day as adding it three days ago had moved the Mainstreet tracker to its lowest Liberal number. Therefore, this puts a dent in the Liberal recovery trend that seemed to be developing.

The Rimouski-Neigette--Témiscouata--Les Basques poll resulted in a small pro-Bloc/anti-NDP adjustment that is nevertheless meaningful as the model had been showing an almost exact three-way NDP/LIB/BQ tie around 24-25%.

The Nanos ON numbers are interesting. Although the headline mentions the Liberal slide in the 905, the model was expecting them where they ended up after the slide. This means that in Nanos' earlier sample, the Liberals were doing slightly better than expected in the 905. Given that small sample sizes should make us question how much of the slide is real, I think that these numbers are actually encouraging for the Liberals in the 905. Minor changes to the ON adjustments were made.

Incorporating those ON numbers into the provincial average is tricky since they cannot be treated as separate polls (they're part of already-included national numbers), and the 7-day release periods do not line up with the national 3-day release periods. The latter is important since regional numbers from each national poll are included as relative values to the national numbers in that poll. I ended up estimating national support for those 7-day periods. All this didn't make much difference: 0.1 seats were shifted from the Conservatives to the Liberals.

Projection as of the latest national poll (midpoint: September 24)
LIB - 149.8 (33.6%)
CON - 146.4 (36.3%)
BQ - 19.3 (5.1%)
NDP - 18.4 (12.8%)
GRN - 3.1 (9.0%)
IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

We see that the Liberal seat lead is essentially unchanged from last night, but the Bloc and NDP have both increased.

Seats changing hands since the last projection:
- In QC, the BQ takes Rimouski-Neigette--Témiscouata--Les Basques and Saint-Jean from LIBs.
- In ON, CONs regain Kitchener South--Hespeler from LIBs.

Projection Update: Nanos 9/23-25

The following poll has been added to the model:
Nanos' 9/23-25 national poll (national results only; current weight among national polls: 22%)
For a full list of included polls, see previous Projection posts.

Today's Nanos update has the Tories and Bloc down, the NDP up, and the Liberals and Greens marginally up. The Liberals increase their lead in the projection, while the NDP regains third spot.

Projection as of the latest national poll (midpoint: September 24)
LIB - 153.2 (33.9%)
CON - 144.4 (36.0%)
NDP - 18.5 (12.7%)
BQ - 17.8 (5.0%)
GRN - 3.1 (9.1%)
IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seat changing hands since the last projection:
- In BC, the NDP gets North Island--Powell River back from CONs.

Wednesday, September 25, 2019

Projection and Methodology Update: Angus Reid's 9/16-18 ON Poll

The following poll has been added to the model:
Angus Reid's 9/16-18 ON poll (current weight among ON polls: 2%)
For a full list of included polls, see previous Projection posts.

Poor Angus Reid! After conducting a huge sample national poll (almost 4,000 respondents), "blackface" happened, and the horserace numbers became instantly outdated. It resorted to publishing the non-horserace results of this poll. Fortunately for us, Angus Reid did publish the voting intention numbers for ON, along with a detailed geographical breakdown, which I have used to update within-ON regional adjustments. Obviously, due to its age, this poll has a very low weight and does not affect ON provincial numbers much, but as you'll see below, many seat leads changed hands due to the changes in 905 adjustments (even though the overall effect on the expected value of the number of seats was quite neutral).

I have also made a methodological change for weighting standalone regional polls like this one. The old method - just use the regional weights designed for the differences between regional and national numbers - caused such polls to depreciate too slowly: their results are not only affected by the region's shifts relative to the country, but also by national shifts themselves. I have therefore made an adjustment to depreciate these polls more quickly (more details here), which marginally helped the Liberals as weight was removed from last week's Campaign Research ON poll: the Liberal lead there changed from 3.8% to 4.0%. This moved about 0.5 seat from the Tories to the Grits.

Projection as of the latest national poll (midpoint: September 23)
LIB - 150.7 (33.7%)
CON - 146.9 (36.4%)
BQ - 18.5 (5.1%)
NDP - 17.7 (12.5%)
GRN - 3.1 (9.1%)
IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seat changing hands since the last projection:
- In ON, LIBs get Mississauga--Streetsville, Oakville, Mississauga--Lakeshore and Vaughan--Woodbridge back from CONs.

Projection Update: Innovative 9/20-25

The following poll has been added to the model:
Innovative's 9/20-25 national poll (current weight among national polls: 17%)
For a full list of included polls, see previous Projection posts.

This IRG poll is much like the previous Mainstreet poll: good Liberal results (relative to Tories) by a pollster with usually good Liberal results. However, unlike the Mainstreet poll, it did move the needle slightly because IRG hadn't come out with a poll for a while; thus, this increases the weight on IRG. And unlike the Mainstreet poll, this one is just as good for the Liberals as IRG's summer polls - and in fact better than its early September one.

The NDP is buoyed by a good BC number (and a poor one for the Tories), which erases its losses from earlier today. The Greens, however, are down again, seemingly resuming a two-month trend that paused only for the NDP's struggles in late August and the "blackface" incident.

Projection as of the latest national poll (midpoint: September 23)
LIB - 150.2 (33.7%)
CON - 147.5 (36.4%)
BQ - 18.4 (5.1%)
NDP - 17.7 (12.5%)
GRN - 3.1 (9.1%)
IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seat changing hands since the last projection:
- In QC, LIBs get Saint-Jean back from the BQ.
- In ON, LIBs get Burlington back from CONs.

Projection Update: Mainstreet 9/22-24

The following poll has been added to the model:
Mainstreet's 9/22-24 national poll (current weight among national polls: 32%)
Update Sept. 25: The tweet above with the full regional breakdown was deleted. Here's a link to the national results.
For a full list of included polls, see previous Projection posts.

This Mainstreet poll is good news for the Liberals: the Conservative national lead is essentially wiped out. Moreover, this is the first public release of the regional breakdown by Mainstreet in over 2 weeks, and it shows massive Liberal leads in QC and ON, with the Tory vote even more concentrated in their strongholds than in other polls.

However, all this apparent good Liberal news has failed to impress the model. Why?
- In terms of the national Conservative-Liberal gap, Mainstreet is still showing a better situation for the Tories than over the summer, while the model is where it was in late July. So rejigging the weights for Mainstreet polls toward this one (due to its recency) actually hurts the Liberals.
- The strong Liberal position in ON relative to national numbers is merely consistent with past Mainstreet releases, so it doesn't move the model much.
- The model is impressed by the strong Liberal number in QC. However, it is also impressed by the very strong Conservative result in Atlantic Canada.

The upshot is that the projection moves only marginally toward the Liberals. If you take today's Nanos and Mainstreet polls together, they had no aggregate effect on the (entirely insignificant) Conservative seat lead of 0.3, which makes sense as Nanos showed a 1.7% improvement in the Tory relative position, Mainstreet showed a 1.2% improvement in the Liberal one, and Mainstreet's sample is larger.

Instead, the action today (if you can call it that) was a slight drop for the NDP and the Greens.

Projection as of the latest national poll (midpoint: September 23)
CON - 149.3 (36.7%)
LIB - 149.0 (33.6%)
BQ - 18.8 (5.1%)
NDP - 16.8 (12.3%)
GRN - 3.2 (9.1%)
IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seat changing hands since the last projection:
- In PE, CONs regain Egmont and Charlottetown from LIBs.
- In QC, the BQ retakes Jonquière from CONs.
- In QC, LIBs retake Longueuil--Saint-Hubert, Hochelaga and Rimouski-Neigette--Témiscouata--Les Basques from the BQ.
- In ON, LIBs get Markham--Stouffville and Kitchener South--Hespeler back from CONs.
- In BC, the NDP gets Skeena--Bulkley Valley back from CONs.

Projection Update: Nanos 9/22-24

The following poll has been added to the model:
Nanos' 9/22-24 national poll (national results only; current weight among national polls: 20%)
For a full list of included polls, see previous Projection posts.

First, in case you missed the Léger update late last night, you may want to read the previous post. That update saw the Liberals essentially pull back into a tie with the Tories, and 11 seats changed hands.

Today's Nanos update shows a jump in Conservative support, who overtake the Liberals despite a marginal increase in the latter's support. The model views this as fairly neutral news, though, as the previous sample markedly improved for the Liberals (moving from 9/18-20 to 9/19-21) - it could be that 9/21 was simply a bad Tory day in the Nanos poll. Other parties, especially the Greens, are down.

Projection as of the latest national poll (midpoint: September 23)
CON - 149.0 (36.6%)
LIB - 148.5 (33.5%)
BQ - 19.2 (5.1%)
NDP - 17.0 (12.3%)
GRN - 3.3 (9.3%)
IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seats changing hands since the last projection:
- In ON, CONs regain Essex from the NDP.
- In BC, CONs gain Skeena--Bulkley Valley from the NDP.

Tuesday, September 24, 2019

Projection Update: Léger 9/20-24 and Insights West BC Polls

The following polls have been added to the model:
Léger's 9/20-24 national poll (national and QC results only; current weight among national polls: 15%)
Insights West's 9/6-10 and 9/19-23 BC polls
For a full list of included polls, see previous Projection posts.

Update Sept. 24: I had missed the PDF link at the top of the Le Devoir article! Hat tip to Evan Scrimshaw from Lean Tossup for pointing that out. Post now updated to include full regional breakdown; further update to come for changes to within-QC adjustments.

Update Sept. 25: The within-QC adjustments have been updated. This cost the Liberals about 0.5 seats to the Conservatives' profit, essentially reversing the effect of adding the national breakdown. However, in terms of seats ahead, it's the Bloc that gains one net seat each from the Grits and Tories.

The Insight West polls do not factor into the BC provincial average because decided+leaning numbers are not available, and the distribution of undecided voters isn't close to uniform geographically or demographically. However, I did use the regional breakdown to define some regional adjustments. These cost the Greens 0.3 seats due to their underperformance on Vancouver Island, and were mildly positive for the Liberals and NDP due to Conservative underperformance in Metro Vancouver. Three seats were flipped: the NDP and Tories swapped Burnaby South (CON to NDP) and Skeena--Bulkley Valley (NDP to CON - but shifted back after the Léger poll), while the Liberals marginally got Cloverdale--Langley City back.

The Léger poll shows no effect of "blackface" nationally. This helps the Liberals in the projection, as other polls had shown an effect. This is another data point consistent with the previously discussed hypothesis that the "blackface" effect is already fading. The QC results, however, are concerning for the Liberals, especially given that Léger oversampled there. Crucially, the Bloc is ahead in the "rest of QC" area, contrary to the late August and early September Léger polls. The Tories also do well in QC, contrary to other recent polls. On the other hand, the poor Tory numbers in Atlantic Canada and ON really help the Liberals.

This update only uses the partial results contained in the Le Devoir article. The breakdown for other provinces and within-QC areas will be added in a future update when they become fully available. For now, the model infers that the Léger results in the rest of Canada must be quite good for the Liberals given that their QC number is poor, but their national number is decent.

Projection as of the latest national poll (midpoint: September 22)
CON - 148.5 147.9 148.5 (36.4%)
LIB - 148.2 148.7 148.2 (33.3%)
BQ - 19.3 19.4 19.2 (5.1%)
NDP - 17.7 17.8 (12.4 12.5%)
GRN - 3.3 (9.5%)
IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seat changing hands since the last projection:
- In PE, LIBs get Charlottetown and Egmont back from CONs.
- In QC, the BQ takes Saint-Jean and Rimouski-Neigette--Témiscouata--Les Basques from LIBs.
- In QC, CONs take Jonquière and Saint-Hyacinthe--Acton from the BQ.
- In QC, CONs take Beauce from the PPC.
- In QC, CONs take Beauport--Limoilou from the LIBs.
- In QC, the BQ gains Montarville from the LIBs.
- In ON, LIBs get Kanata--Carleton back from CONs.
- In ON, the NDP gets Essex back from CONs.
- In BC, LIBs get Cloverdale--Langley City back from CONs.
- In BC, the NDP gets Burnaby South back from CONs.

Projection Update: Mainstreet 9/21-23

The following poll has been added to the model:
Mainstreet's 9/21-23 national poll (national numbers only; current weight among national polls: 26%)
For a full list of included polls, see previous Projection posts.

Correction 9/24: Forgot to enter a weight for Mainstreet's BQ number on the spreadsheet.

This Mainstreet poll basically repeats the results from yesterday, with a mild uptick for the Liberals (and downtick for the Greens). The projection moves marginally against the Liberals, as a result of the model applying more aggressive negative weights on old Liberal-favouring Mainstreet polls. But the basic story is: no change.

With this update, 97% of the national weight is on post-"blackface" data.

Projection as of the latest national poll (midpoint: September 22)
CON - 151.4 (36.6%)
LIB - 145.7 146.1 (33.1 33.2%)
BQ - 19.2 18.7 (5.1 5.0%)
NDP - 17.1 17.2 (12.3%)
GRN - 3.6 (9.6 9.7%)
IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

These results are virtually identical to last night's projection.

Seat changing hands since the last projection:
- In BC, CONs retake Cowichan--Malahat--Langford from GRNs.

Projection Update: Nanos 9/21-23, Ipsos 9/20-23

The following polls have been added to the model:
Nanos' 9/21-23 national poll (national results only; current weight among national polls: 16%)
Ipsos/Global News' 9/20-23 national poll (national results only; current weight among national polls: 14%)
Mainstreet's 9/18 Compton--Stanstead poll
For a full list of included polls, see previous Projection posts.

The Nanos poll puts the Liberals back in the lead, consistent with the fading "blackface" effect narrative. The BQ is still high, despite the dropping of the 9/20 sample, when it first jumped. Despite the latter, this update shifts the projection toward the Liberals.

The Ipsos poll shifts the projection toward the Tories, but only marginally due to Ipsos' slight Tory lean. Overall, the Nanos effect dominates, and the Conservative lead in the projection shrinks slightly.

Finally, the Mainstreet poll caused a riding adjustment for Compton--Stanstead, which is of little consequence since the Liberals are comfortably ahead there anyway.

Projection as of the latest national poll (midpoint: September 22)
CON - 150.7 (36.5%)
LIB - 146.3 (33.1%)
BQ - 19.1 (5.1%)
NDP - 17.2 (12.3%)
GRN - 3.7 (9.9%)
PPC - 0.5
IND - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seats changing hands since the last projection: none.

Monday, September 23, 2019

Projection Update: EKOS 9/18-22

The following poll has been added to the model:
EKOS' 9/18-22 national poll (current weight among national polls: 6%)
For a full list of included polls, see previous Projection posts.

This EKOS poll is further bad news for the Liberals in QC, which moves the projection slightly. In ON, it's mixed: good numbers from an absolute standpoint, but EKOS had been producing very Liberal-friendly numbers in ON.

EKOS says that the "blackface" effect appears to be receding after being initially severe. Suppose the effect was bad 9/18-20, but went away on 9/21. This is consistent with the Nanos tracking (9/18-20 being the Liberals' worst period). It also reconciles the DART (9/20) and Campaign Research (9/18-19) polls being very poor for the Liberals with the Abacus (9/18-22), EKOS (9/18-22) and Angus Reid (9/21-23) polls producing milder results. This theory does not explain, however, the trend observed by Mainstreet (still worsening as of 9/20-22) or the good 9/19-21 Liberal numbers from Forum. Nevertheless, given the noisy nature of polling, the hypothesis seems plausible to me - it's normal that a couple of polls wouldn't quite tell the same story.

I should mention that the model simply takes the midpoint date of polls, and does not attempt to infer the numbers from individual nights. So for now, the effect of "blackface" isn't receding in the projection. We shall see what the next few days hold...

Projection as of the latest national poll (midpoint: September 22)
CON - 151.3 (36.5%)
LIB - 145.4 (33.0%)
BQ - 18.9 (5.1%)
NDP - 17.8 (12.5%)
GRN - 3.6 (9.7%)
IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seats changing hands since the last projection:
- In BC, CONs regain Burnaby South from the NDP.
- In BC, LIBs retake Victoria from GRNs.

Projection Update: Angus Reid 9/21-23, DART 9/20 and Mainstreet 9/20-22

The following polls have been added to the model:
Angus Reid's 9/21-23 national poll (current weight among national polls: 27%)
DART's 9/20 national poll (current weight among national polls: 8%)
Mainstreet's 9/20-22 national poll (national results only; current weight among national polls: 23%)
For a full list of included polls, see previous Projection posts.

This Angus Reid poll has high weight, as its field dates are a day more recent than the Nanos and Mainstreet tracking polls released today, while the DART poll has low weight due to being a day older. The big news is in QC, where Angus Reid and DART both corroborate the Bloc's rise observed by Nanos, Forum and, to a lesser extent, Mainstreet. This has a significant impact on the Bloc's seat projection, and that comes mostly at the expense of the Liberals. Also, despite the pro-Conservative turnout adjustment, the Bloc is now ahead of the Tories in QC in the vote projection.

On the other hand, despite the national headline numbers, the Angus Reid and DART polls did not move the projection against the Liberals much in the rest of Canada. The reason is that the model is smart enough to expect a Conservative-leaning result from DART and a poor Liberal result from Angus Reid, given these pollsters' past surveys. Also, the Angus Reid poll is decent for the Liberals in the rest of Canada: they did not lose ground in ON, and in fact gained marginally (not statistically significant) in the Atlantic and BC.

In fact, more than half of the Tories' 5-seat advantage came from their 1-point increase in the Mainstreet tracker (remember that the previous projection was almost a perfect tie). The NDP benefits from a decent Angus Reid number in BC and a good DART number in ON.

Update Sept. 23: Note that the support levels for the NDP, Greens and BQ were not precisely given in the iPolitics article. They were estimated from the text, and will be updated if they become publicly available.

Projection as of the latest national poll (midpoint: September 22)
CON - 151.2 (36.5%)
LIB - 145.9 (33.1%)
BQ - 18.6 (5.1%)
NDP - 17.7 (12.4%)
GRN - 3.6 (9.7%)
IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seats changing hands since the last projection:
- In PE, CONs regain Charlottetown from LIBs.
- In QC, the BQ takes Saint-Hyacinthe--Acton and Jonquière from CONs.
- In QC, the BQ takes Beloeil--Chambly, Hochelaga and Longueuil--Saint-Hubert from LIBs.
- In ON, CONs regain Kanata--Carleton from LIBs.
- In AB, the NDP gets Edmonton Strathcona back from CONs. (Note: I am not using a paywalled Mainstreet survey for which the article title suggests a healthy Tory lead in this riding.)
- In BC, the NDP gets Burnaby South back from CONs.

Projection Update: Abacus 9/18-22 and Nanos 9/20-22

The following polls have been added to the model:
Abacus' 9/18-22 national poll (current weight among national polls: 19%)
Nanos' 9/20-22 national poll (national results only; current weight among national polls: 20%)
Mainstreet's 9/15 Ottawa Centre poll
Mainstreet's 9/? Shefford poll
For a full list of included polls, see previous Projection posts.

Abacus shows a stable race nationally from its previous poll on the eve of the campaign. The regional breakdown, however, is somewhat encouraging for the Liberals, with a small lead in ON (compared to a near-tie previously) and a tie in BC. Moreover, while the Liberals are down slightly from the previous Abacus poll, the Bloc is stable rather than up.

The Nanos poll shows the Tories dropping into a near-tie with the Liberals, with the NDP also down, and the Greens up. The Bloc is still high; it will be interesting to see if the Bloc drops tomorrow, three days after its big jump. Like yesterday, this had a neutral effect on the projection, as the negative shift for the Tories is offset by the replacement of older Nanos samples by those one day more recent.

The Ottawa Centre and Shefford polls resulted in adjustments for those ridings.

All this means a tiny seat shift toward the Liberals and vote shift away from them. The NDP falls marginally behind the Bloc in the national seat count, and behind the Greens in the BC vote share; this is mostly due to Bloc and Green strength rather than NDP weakness.

Update Sept. 23: I've just noticed that Mainstreet uses weights based on 2016 census counts rather than the most recent population estimates. This makes a small difference for the BQ numbers in QC inferred from the national numbers. But things are so close that this actually flips the seat lead!

Projection as of the latest national poll (midpoint: September 21)
LIB - 150.7 151.0 (33.7%)
CON - 150.8 150.9 (36.8%)
BQ - 16.0 15.7 (4.9%)
NDP - 15.6 (11.7%)
GRN - 3.8 (9.7%)
IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seats changing hands since the last projection:
- In PE, LIBs get Charlottetown back from CONs.
- In QC, CONs retake Jonquière and Trois-Rivières from LIBs.
- In ON, LIBs get Kanata--Carleton back from CONs.

Polling-Related Adjustments 2019, Round 3

This post reviews polls of smaller areas than the six standard polling regions (BC, AB, SK/MB, ON, QC, Atlantic), and polls providing breakdowns of smaller areas. Resulting adjustments to the model are indicated in bold. You can view earlier rounds of polling-related adjustments here (round 1) and here (round 2).

Update Sept. 23: EKOS MB/SK results added. No adjustment yet.

Update Sept. 24: Insights West polls of BC. Adjustments added.

Update Sept. 25: Léger's QC breakdown. Adjustments updated.

Update Sept. 25: Angus Reid's ON breakdown. Adjustments updated.

Update Sept. 26: Mainstreet's Rimouski-Neigette--Témiscouata--Les Basques poll. Adjustments added.

Update Sept. 26: Nanos' ON breakdown. Adjustments updated.

Update Sept. 27: Mainstreet's Laurier--Sainte-Marie poll. Baseline adjustments updated.

Update Sept. 28: Mainstreet's GTA numbers and Forum's Toronto poll. Adjustments updated.

Update Oct. 1: Probe's MB poll and EKOS' MB/SK breakdown. Adjustments added. Also added: Mainstreet's Beauport--Limoilou poll, which did not result in an adjustment, and a Mainstreet Jonquière poll that I had missed, which did result in an adjustment.

Update Oct. 2: Mainstreet's Terrebonne poll. Adjustments added.

Update Oct. 3: Mainstreet's Beloeil--Chambly poll. Adjustments added.

Update Oct. 4: Mainstreet's Berthier--Maskinongé poll. Adjustments added.

Update Oct. 6: Mainstreet's Longueuil--Saint-Hubert poll. Adjustments added.



I. Regional Polls

Québec
Léger's 9/20-24 poll
This poll shows the Conservatives doing better in Quebec City, which shrinks their negative adjustment there; instead, the negative adjustment in the Montreal area grows. Changes from the previous round are in bold:
Quebec City CMA: CON -7, NDP -0.5, LIB +2.5 [BQ adjustment removed]
Montreal CMA: CON -2 (-0.5 in far suburbs), NDP +1.5 (except far suburbs), LIB +2 (Island of Montreal), +1.5 (near suburbs), +1 (far suburbs), GRN +0.5 (Island of Montreal only), BQ -1 (-0.5 in far suburbs)
Rest of Quebec North: CON +3.5, NDP -1.5, LIB -2.5, BQ +1
Rest of Quebec South: CON +4, NDP -1, LIB -3 (-2 in Eastern Townships), GRN -0.5, BQ +1
(Salaberry--Suroît is mostly outside of Montreal CMA but shares some characteristics with it. Since the Montreal CMA and Rest of QC S adjustments go in opposite directions, I have left this riding unadjusted. Moreover, due to the large difference between Quebec City and Rest of QC North adjustments, I am computing the adjustment in Beauport--Côte-de-Beaupré--Île-d'Orléans--Charlevoix and Portneuf--Jacques-Cartier according to their rough population distribution: 55% and 58% respectively in Quebec City CMA, 45% and 42% outside.)

Ontario
Campaign Research's 9/18-19 poll
The main changes in the regional adjustments are as follows:
- I have separated York/Simcoe/Durham and Halton/Peel as Campaign Research concurs with earlier Corbett breakdowns that show the latter shifting more strongly away from Liberals than the former.
- The Liberals did well relative to what one would expect based on their provincial number in Hamilton/Niagara and Southwestern ON.
- The Tories did well relative to what one would expect based on their provincial number in Ottawa and Northern ON.
The updated adjustments are:
York/Simcoe/Durham: CON -1, NDP +2.5, LIB -0.5, GRN -1
Halton/Peel: CON +0.5, NDP +2, LIB -1.5, GRN -1
Hamilton/Niagara: CON -1.5, LIB +1.5
Southwest: NDP -1.5, LIB +1, GRN +0.5
East outside Ottawa: CON -2.5, NDP +0.5, GRN +2
City of Ottawa: CON +1.5, NDP +2, LIB -3, GRN -0.5
North: NDP -1.5, GRN +1.5
I have not changed the adjustment for the 7 central Toronto ridings (CON +2, NDP -5, LIB +2.5, GRN +0.5).
These adjustments, in effect from the 9/20 evening projection, are still fairly cautious - very roughly half the size suggested by the polls - to account for the possibility that some of the apparent effects might be due to random variation.

Angus Reid's 9/16-18 poll
Although conducted before the Campaign Research poll, this survey was published after it. The Liberals got strong numbers in the 905, the NDP in Hamilton/Niagara, and the Conservatives in Eastern and Northern ON. The updated adjustments, in effect from the 9/25 night projection, are (changes from the above in bold):
7 central Toronto ridings: NDP -4, LIB +3, GRN +1
York/Simcoe/Durham: CON -1, NDP +2.5, LIB -0.5, GRN -1 (CON -1.5, NDP +2.5, GRN -1 in Toronto CMA)
Halton/Peel: CON -0.5, NDP +2, LIB -0.5, GRN -1
Hamilton/Niagara: CON -1.5, NDP +0.5, LIB +1.5, GRN -0.5
Southwest: CON +0.25, NDP -2, LIB +0.75, GRN +1
East outside Ottawa: CON -1.5, NDP +1, LIB -1, GRN +1.5
City of Ottawa: CON +2, NDP +1.5, LIB -3.5
North: CON +4.5, NDP -3.5, LIB -0.5, GRN -0.5
Again, these remain cautious since nothing is very consistent across pollsters. For the purposes of retroactive computations when producing trends, these adjustments supersede those above, and are deemed to have been made as of the Campaign Research poll.

Nanos' 9/11-24 ON breakdown
These numbers caused a minor change in some adjustments (others remain as above):
York/Simcoe/Durham: CON -1, NDP +2, LIB 0, GRN -1 (CON -1.5, NDP +2, LIB +0.5, GRN -1 in Toronto CMA)
Southwest: CON +0.25, NDP -1.75, LIB +0.5, GRN +1
Again, for simplicity, for the purposes of retroactive computations when producing trends, these adjustments are deemed to have been made at the same time as those above and supersede them.

Mainstreet's 9/23-26 GTA breakdown
These numbers are favourable for the Liberals in the GTA overall, particularly in Peel. This is the opposite of summer Corbett numbers, which suggested that they're doing better in York/Durham. Therefore, I have decided to make the adjustments in those areas very similar. Mainstreet also shows great numbers for the Greens in the Simcoe region, but given that the subsample for that area is likely well under 100 respondents, the adjustment is limited. The changed adjustments are below (remainder as above), including balancing adjustments elsewhere in ON:
Halton/Peel and Durham outside Toronto CMA: CON -1, NDP +2, LIB 0, GRN -1
York and Durham inside Toronto CMA: CON -1.5, NDP +2.5, LIB 0, GRN -1
Simcoe: CON -2, NDP +1.5, LIB -0.5, GRN +1
Southwest: CON +0.5, NDP -1.75, LIB +0.5, GRN +0.75
North: CON +5, NDP -3.5, LIB -0.5, GRN -1

Forum's 9/25-26 Toronto poll
The Forum poll's results are very similar to the summer Forum poll of Toronto. The Central Toronto adjustment is maintained, and no further adjustment is made.

Manitoba/Saskatchewan
EKOS' 9/18-22 poll
Probe's 9/17-26 MB poll
EKOS' 9/26-29 poll
MB/SK breakdowns from national polls are tiny, and few pollsters provide them. Still, they usually suggest that the Tories are gaining more in SK than MB, and this one is no exception. I am close to making an adjustment. With the addition of the latest EKOS and Probe polls, I'm ready to make the following adjustments: CON -1.5, NDP +1.5 in Winnipeg, CON +1, NDP -1 in SK.

The Probe poll also suggests that the Liberals are not doing as well as expected in MB. However, combining the breakdowns from 7 EKOS polls and 2 Campaign Research polls conducted since July shows no signs that the Liberals are underperforming in MB relative to SK. I will keep an eye on this situation...

British Columbia
Insights West's 9/6-10 and 9/19-23 polls
Note: These polls do not factor into the BC provincial average because decided+leaning numbers are not available, and the distribution of undecided voters isn't close to uniform geographically or demographically. 
The regional breakdowns from these polls suggest that the Tories are rising more than expected in the rest of BC, mainly at the expense of the NDP, while the Greens are rising less than expected in Vancouver Island, with the Liberals holding up better there. For all four parties, countervailing effects were observed in Metro Vancouver. For now, I've made some cautious adjustments that are 20-50% of the observed effects:
Metro Vancouver: CON -1.5, NDP +1.5, LIB -0.5, GRN +0.5
Vancouver Island: LIB +1.5, GRN -1.5
Rest of BC: CON +2.5, NDP -2.5
The Metro Vancouver adjustment does not apply to Vancouver Granville since it has already been adjusted following riding polls. No adjustment applies to West Vancouver--Sunshine Coast--Sea to Sky Country, which is partly in Metro Vancouver, and partly in the "rest of BC."


II. Riding Polls (by Mainstreet unless otherwise noted)

Note: A riding poll will result in a model adjustment ONLY IF it shows a substantial deviation from regional swing and/or it is in a riding with unusual factors. Most riding polls will therefore be ignored: they generally did not help much in 2015.

Debate Commission Polls by EKOS
Charleswood--St. James--Assiniboia--Headingley, 8/26-9/8
Nipissing--Timiskaming, 8/26-9/8
Etobicoke North, 8/26-9/9
Pickering--Uxbridge, 8/26-9/9
As written in the 9/17 projection update: "For now, I will not be making any adjustment based on the debate commission's polls because, while 15-20% (certain and likely) are strong results for the PPC in those four ridings (Etobicoke North, Nipissing--Timiskaming, Pickering--Uxbridge and Charleswood--St. James--Assiniboia--Headingley), that still puts them at least 10% away from actually winning a seat. This does mean my PPC expected seat count (which is just Beauce) is probably a few tenths too low. I may reverse this decision if the PPC takes off in the coming weeks."

Lac-Saint-Jean, 9/10
This poll caused me to take another look at the baseline adjustment for Lac-Saint-Jean, which resulted in a revision. I elected not to make a further adjustment even though the Bloc's poll result is much lower than expected: the PQ did relatively well in the region in the 2018 provincial election.

Bécancour--Nicolet--Saurel, 9/12
The Liberals are a bit closer to the BQ than expected, but not enough to warrant an adjustment.

Portneuf--Jacques-Cartier, 9/13
The results closely matched the model's projection.

Jonquière, 9/14
This poll's results were very close to the projection for the CONs, LIBs and GRNs, but the NDP was a bit higher and the BQ lower than expected. Adjustment: NDP +2.5, BQ -2.5.

Ottawa Centre, 9/15
As one might have expected, even though the Paul Dewar personal effect wasn't easily visible in past election results (which resulted in only a modest baseline adjustment), the NDP appears to be struggling badly in this riding. Some of it might be a Catherine McKenna effect for the Liberals, though the Tories are also higher in this poll than expected. This leads to a riding-level adjustment of CON +5, NDP -10, LIB +5. This adjustment is included in projections starting 9/23.

Shefford, 9/? (published 9/23)
Like Mainstreet's earlier Sherbrooke poll, this one shows the Liberals higher and the Tories lower than expected. Adjustment: CON -4, LIB +4.

Compton--Stanstead, 9/18
Once again, the Liberals are doing better than expected in the Eastern Townships. Here, they may be benefiting from a well-known Bloc ex-MP not running again. Adjustment: LIB +5, BQ -5.

Rimouski-Neigette--Témiscouata--Les Basques, 9/23
The model saw a three-way race so tight that the leader had less than 25% support. The poll is a bit higher then the model for the Bloc, lower for the NDP, and bang on for the other parties. Adjustments: NDP -3, BQ +3.

Laurier--Sainte-Marie, 9/24
See post on baseline adjustments.

Beauport--Limoilou, 9/?
This is very close to where the model had the race. No adjustment.

Terrebonne, 9/25
The Bloc is a few points higher than expected, with the Liberals, Tories and NDP all marginally lower. This is a marginal case for whether an adjustment is necessary. Adjustments: CON -1, NDP -1, LIB -1, BQ +3.

Berthier--Maskinongé, 9/26
The NDP is a dozen points higher than expected, mostly at the expense of the Greens and Tories. As usual, the adjustment based on a single poll is about half of the gap between the projection and the poll, which makes this race essentially a dead heat. Adjustments: CON -2, NDP +6, GRN -3, BQ -1.

Beloeil--Chambly, 9/29
In contrast to Mainstreet's early September poll, which shows the Liberal candidate ahead, this latest poll shows the Bloc leader with a sizable lead. This is not surprising: campaigns matter, and Blanchet has been effective so far. Adjustments: CON -2.5, LIB -1.5, BQ +4.

Longueuil--Saint-Hubert, 9/29
This poll's results are similar to the projection, with the NDP a few points lower and the Liberals a few points higher. I have decided to make small adjustments even though I would usually ignore deviations this small: this is the riding where Pierre Nantel, the NDP MP, defected to the Greens, so it makes sense that some NDP voters may defect as well. Adjustments: NDP -2, LIB +2.

Sunday, September 22, 2019

Projection Update: Mainstreet 9/19-21

The following poll has been added to the model:
Mainstreet's 9/19-21 national poll (national results only; current weight among national polls: 27%)
For a full list of included polls, see previous Projection posts.

Mainstreet has the Liberals dropping 2 points, with all other parties progressing. A 2-point change in a 3-day rolling sample means that the 9/21 sample was 6 points lower than the 9/18 sample for the Liberals. Given a daily decided+leaning sample size of ~650, that is a statistically significant change.

This is (barely) enough to cause another lead change in the projection.

Projection as of the latest national poll (midpoint: September 20)
CON - 151.4 (36.8%)
LIB - 150.3 (33.8%)
NDP - 15.9 (11.9%)
BQ - 15.7 (4.9%)
GRN - 3.7 (9.3%)

IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seats changing hands since the last projection:
- In PE, CONs regain Charlottetown from LIBs.
- In QC, CONs take Saint-Hyacinthe--Acton from the BQ.
- In ON, CONs regain Kanata--Carleton back from LIBs.

Projection Update: Forum 9/19-21

The following poll has been added to the model:
Forum's 9/19-21 national poll (current weight among national polls: 28%)
For a full list of included polls, see previous Projection posts.

This poll is very good news for the Liberals: not only does a tie compare well to the previous Forum result of a 3-point Tory lead, but also the Liberals have a 6-point lead in ON. In fact, the Liberals getting just 12% in AB means that this poll is better news for them than its headline suggests.

The bad news for the Liberals is that the Bloc is up in QC, like in the Nanos and Mainstreet polls (as inferred from the national numbers). The Conservatives are weak in QC.

The NDP and Greens have the same top line number, but the Greens get much better news in terms of support distribution. For the first time in a while, the Greens are projected ahead in 5 seats.

All this has a meaningful impact on the projection, where the Liberals retake the lead, and the BQ is challenging the NDP for third place (and in fact have more seats projected ahead than the NDP).

Projection as of the latest national poll (midpoint: September 20)
LIB - 153.6 (34.2%)
CON - 149.1 (36.7%)
NDP - 15.6 (11.8%)
BQ - 15.1 (4.9%)
GRN - 3.6 (9.1%)

PPC - 0.5
IND - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seats changing hands since the last projection:
- In NL, LIBs get Avalon back from CONs.
- In PE, LIBs get Charlottetown back from CONs.
- In QC, the PPC takes Beauce from CONs.
- In QC, LIBs take Jonquière and Trois-Rivières from CONs.
- In QC, the BQ takes Saint-Hyacinthe--Acton from CONs.
- In ON, LIBs get Mississauga--Erin Mills, St. Catharines, Eglinton--Lawrence and Kanata--Carleton back from CONs.
- In AB, CONs regain Edmonton Strathcona from the NDP.
- In BC, GRNs retake Cowichan--Malahat--Langford from CONs.
- In BC, GRNs retake Victoria from LIBs.

Projection Update: Nanos 9/19-21

The following poll has been added to the model:
Nanos' 9/19-21 national poll (national results only; current weight among national polls: 22%)
For a full list of included polls, see previous Projection posts.

The Tory lead shrinks a bit in this Nanos poll. This almost perfectly offsets the replacement of older Nanos samples (remember that this is a rolling poll) with those that are one day more recent. The projection is almost exactly unchanged.

Projection as of the latest national poll (midpoint: September 20)
CON - 154.4 (37.2%)
LIB - 148.8 (33.9%)
NDP - 15.8 (11.7%)
BQ - 14.8 (4.8%)
GRN - 3.2 (9.2%)

IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seats changing hands since the last projection: none.

Saturday, September 21, 2019

Mapped Projection Update: Mainstreet 9/18-20

The following poll has been added to the model:
Mainstreet's 9/18-20 national poll (national results only; current weight among national polls: 36%)
For a full list of included polls, see previous Projection posts.

Mainstreet, like Nanos did this morning, also sees a drop in Liberal support. However, because the drop is smaller and adding this poll reduces the weight on recent Nanos polls slightly, the projection is only marginally affected.

Projection as of the latest national poll (midpoint: September 19)
CON - 154.5 (37.2%)
LIB - 148.6 (33.9%)
NDP - 16.0 (11.8%)
BQ - 14.7 (4.8%)
GRN - 3.2 (9.1%)

IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seat changing hands since the last projection:
- In QC, the BQ gains Salaberry--Suroît from LIBs.

Given all the changes in the past 24 hours, it's time for a new map:


And here's the potential House of Commons. Note that if I used the rounded projection rather than seats ahead, the Conservatives and Bloc would combine for 169 seats, while the Liberals, NDP and Greens would combine for 168 seats...

Projection Update: Nanos 9/18-20

The following poll has been added to the model:
Nanos' 9/18-20 national poll (national results only; current weight among national polls: 21%)
For a full list of included polls, see previous Projection posts.

This poll gives the Tories their largest national lead in any public poll since mid-August. The Liberals have dropped 3.5 points in 3 days.

Interesting, the Bloc has noticeably jumped, and was up as well in yesterday's Mainstreet poll. While Quebecers, at least judging by the media reaction, may not be as upset by Trudeau's "blackface" as other Canadians, they may be quite annoyed at Trudeau's apparent hypocrisy given that he seems to be planning federal intervention in Bill 21 court cases. That might have been understandable to Quebecers had Trudeau himself been exemplary, but now that Trudeau is revealed to be no better than others, Quebecers may have a lot less patience for his lessons.

Update: It looks like this hasn't happened - the BQ gains appear to come from elsewhere - but it is a risk for Trudeau going forward. Remember that Legault liked repeating during the 2018 campaign that Quebecers were fed up with Couillard's lessons. For now, the Bloc may be gaining among Quebecers that find the reaction in the rest of Canada borderline hysterical.

Or maybe the BQ jump is just a blip. We'll find out in the coming days.

The model sees this poll's results as quite significant, given that Nanos does not usually have a Conservative lean. It is very possible that there will be a pull back later today when the Mainstreet numbers come out. However, for now, the Tories gain the projection lead.

Projection as of the latest national poll (midpoint: September 19)
CON - 154.9 (37.4%)
LIB - 149.7 (34.1%)
NDP - 15.3 (11.6%)
BQ - 14.1 (4.8%)
GRN - 3.1 (8.9%)

IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seats changing hands since the last projection:
- In NL, CONs gain Avalon from LIBs.
- In QC, the BQ gains Laurentides--Labelle and Laurier--Saint-Marie from LIBs.
- In ON, CONs gain Eglinton--Lawrence, St. Catharines and Mississauga--Erin Mills from LIBs.

Friday, September 20, 2019

Projection Update: Campaign Research 9/18-19 ON Poll

The following poll has been added to the model:
Campaign Research's 9/18-19 ON poll (current weight among ON polls: 28%)
For a full list of included polls, see previous Projection posts.

This large sample Campaign Research poll of ON (2,035 respondents total) has very encouraging news for the Tories: they are ahead by 4%. This is the Conservatives' best result in ON in any poll published in the past 2 months. The Liberal ON lead in the projection halves from 4.3% to 2.1%, which has a significant impact on the national numbers.

In terms of the comparison between regions of Ontario and the province as a whole, compared to the last election, this poll has good news for the Tories in Ottawa and Northern ON, and good news for the Liberals in Southwestern ON and Hamilton/Niagara. The regional adjustments have been updated, and will be in the next post on polling-related adjustments.

Projection as of the latest national poll (midpoint: September 18)
LIB - 153.9 (34.5%)
CON - 152.3 (37.2%)
NDP - 15.1 (11.5%)
BQ - 12.6 (4.6%)
GRN - 3.2 (9.0%)

IND - 0.5
PPC - 0.5
If you're new to my 2019 projections, view key interpretive information here.

Seats changing hands since the last projection:
- In ON, CONs gain Kenora, Markham--Stouffville, Mississauga--Lakeshore, Oakville and Mississauga--Streetsville from LIBs.

Additionally, the following changes occurred due to changes in the regional adjustments:
- CONs gain Kanata--Carleton from LIBs. (Addendum: This, of course, does not account for the scandal affecting the CON candidate.)
- LIBs get St. Catharines back from CONs.