Tuesday, April 26, 2011

NDP gallops forward

Two new national polls were added to the model this morning, along with three new riding polls conducted in Quebec. All of these polls are responsible for some shift in the projection, but there have also been a few changes made to my weighting system.

I realized that the poll-weighting tests I had run for the 2006 and 2008 elections were done slightly incorrectly. Rather than only include complete polls like I am doing in this election, my tests included each day of a daily tracking poll. What this did was fill the model with only newer polls, whereas the system I am using actually includes more older polls.

However, I don't tweak my model arbitrarily or because things "don't look right". I need numbers and evidence to back-up any changes to the model. So, I ran some new tests modeling the 2006 election (which has some similarities with the current election because of the sudden swing in support) and found that the best course of action would be to reduce each passing day's weight by 7% rather than 4%. This ages polls more quickly. I also decided to remove the "correction" that was added to the projection to take into account how the polls had been inaccurate in past elections. In the end, the tests ran better without this correction. The correction was very small, and for the most part the growth that the Green Party has been given today is due to that. The other parties were only marginally effected.

So, while some of the changes in this update are artificially influenced, most change can be attributed to the new polls. Let's get to it!
The Conservatives have dropped 1.1 points to 37.5% and have dropped five seats to 146. The Liberals, meanwhile, are also down 1.1 points (to 26.4%) but are unchanged at 75 seats. The New Democrats are up 1.8 points to 20.9% and six seats to 42.

The Bloc Québécois is down 0.3 points nationally to 7.8% and one seat to 44. The Greens are up 0.6 points to 6.1% but remain seatless. André Arthur is the lone independent projected to be elected.

The combination of increased aging, new polls, and the removal of the small correction has led to some major changes at the regional level.

The Conservatives have dropped a point or so in each region, but they are especially hurt by the drop of 1.3 points in Quebec. They are down to 18.7% there.

The Liberals dropped less out west but a lot in Atlantic Canada. Their drops in Quebec and Ontario are especially problematic.

The New Democrats rose everywhere, most importantly in British Columbia, Atlantic Canada, and Quebec. There, they grew their support by 3.5 points in the projection. They are now at 24.2%, only 8.5 points behind the Bloc Québécois, which has dropped 1.4 points to only 32.7%.

In terms of seats, there have been quite a few changes.

In British Columbia, the NDP candidate Kennedy Stewart is now favoured in Burnaby - Douglas, the seat previously held by retired NDP MP Bill Siksay. The Conservatives had been projected to win it earlier.

In Ontario, the Liberals are projected to once again win Ajax - Pickering (Mark Holland) and Brampton - Springdale (Ruby Dhalla), while New Democratic incumbent Malcolm Allen is the projected winner once again in Welland. These are all pick-ups from the Conservatives.

The Liberals move ahead in the two Toronto-area ridings because the incumbent factor has moved from a penalty (incurred when a party is gaining in a province) to a bonus (added when a party is losing in a province compared to the 2008 election).

But it is in Quebec where the most changes have occurred. The New Democrats have picked up four seats, one each from Bloc and Conservatives and two from the Liberals.

The NDP has picked-up two ridings in the Outaouais: Hull-Aylmer from the Liberals and Pontiac from the Conservatives. Nycole Turmel and Mathieu Ravignat are now the respective projected winners.

On the island of Montreal, the NDP has moved ahead in Jeanne-Le Ber (Tyrone Benskin) and Westmout - Ville-Marie (Joanne Corbeil).

Now, it may be surprising to see the NDP knocking off a star Liberal MP and a Conservative cabinet minister. With the NDP soaring into new territory in Quebec, it is very difficult to predict which ridings will be swept away in the orange tide.

But Lawrence Cannon has never exactly been a consensus choice in Pontiac, and the NDP is leading in Montreal, according to several polls. A minister that has never sought headlines and an astronaut opposition MP are not immune from sweeping change. Recall that even the Prime Minister was defeated in her riding in the 1993 election. And the six seats now projected to vote NDP in Quebec are divided into two blocks of three adjacent seats. It seems relatively sensible that support would spill over from one riding to the other. But there is no doubt that the NDP's rise in Quebec will make things difficult for seat projectors on May 2nd.


  1. What would the seat counts have been under the old method?

  2. Ha Eric, you didn't have to go out of your way to explain why you made the changes you did - I think most people who read this site trust you to have made the right decision!

  3. Yeah, I'm doubting Pontiac. Burnaby Douglas was probably always NDP.

    I think you should include polling inaccuracy from the last election, as some of the opinion polls have a consistent bias.

  4. Thanks for the modification on the model. Although I am still a little suspicious of such a strong historical basis this change seem to capture the momentum in Quebec.

    However, I am doubtful of the prediction for Westmont. This is a pretty hardcore liberal riding, granted the NDP are projected at 42% in Montreal (according to EKOS's latest). I wonder where this support is located. In the french side of the city?


  5. So basically, you've made the projection slightly more volatile - but not significantly so. Probably a good thing in a fast-paced campaign.

  6. There is no way the CPC win Avalon in NL by that margin. If they do win it, it will only be by a couple of hundred votes. But most likely, LPC win that one.

  7. Eric, thanks for the update. If I'm reading your chart correctly, there appears to be only one seat change predicted for the "vote-reach" Ontario and that seat is from independent Conservative to Conservative in Simcoe-Grey. Basically, you're model is predicting the exact same outcome for Ontario that took place in 2008 - no riding will change party. Assuming this is accurate, this really calls into question every pundit who claims Ontario is the key to this election.

  8. The EKOS poll is definetely an outlier, and I'll wait for Wednesday and Thursday for the other polls to return some sanity.

  9. Good morning,

    I follow your blog everyday and am doing a seat projection myself. Obviously, this sudden surge of the NDP can create a huge shift in so many ridings. I for one have identified 36 ridings too close to call and the others are safe bets but this was right before the NDP surge. I'll do it again this week as I have an election pool with a fellow colleague from work. This morning swing in your calculations is very interesting. Especially in Quebec with the 4 seats for the NDP. I agree that Hull-Aylmer is doable. Same for Jeanne-Le Ber. In Pontiac, though, I believe that these folks really want the guy around the decision table and will remain Conservative but with a slimmer margin than in 2008 which was already slim, I agree. I just believe that the CPC machine will work really hard in their 11 ridings to have their people vote and will therefore have the edge there and same for the Bloc. The NDP turnout will be the key in Quebec. As for Westmount, as you had yourself mentionned earlier, in '08, they had a prominent journalist against Garneau, not now. I believe Garneau is safe.

    As for the two switches in Ontario, I tend to disagree that the incumbent has a momentum. On the opposite, I believe that the CPC machine and the undecided will make it really hard for the liberals and that they'll lose both.

    But it was just my 2 cents!

    Keep up good work!

  10. Heartening as it may be to see the Conservatives moving away from a majority again... I don't believe it all. With more votes and a more favorable split of the opposition, I have trouble imagining the CPC only picking up 4 or 5 seats. Moreover, I think EKO's frequent outlier polling, (Even keeping in mind your accounting for house effect) is dragging down the CPC popular vote projection.

  11. This site seems to be betting the farm on EKOS.

    Those large sample polls with outlier low CPC numbers seem to be swamping everything else in the projection.

  12. New Ekos poll out (2nd day in a row). Tories up to 34. If you go to http://en.wikipedia.org/wiki/Opinion_polling_in_the_Canadian_federal_election,_2011 and sort the polls by Conservative vote, you'll see the lowest FOUR (out of about 70) are all from Ekos!

    What do you guys think?

  13. Nice work Eric. Seems like this is closer to the actual result we will see on May 2nd. I do however want to put some comments forward with a few ridings which don't quite add up:

    Esquimalt-Juan de Fuca - With the Liberal decline and rise of NDP in BC, as well as Keith Martin not being there, the Liberal share seems outlandishly high. I would say this riding is TCTC but it will probably be between the CPC and NDP. Why does the model put the Liberals so high?

    Edmonton East - With the NDP doing so well, are there any AB polls showing that same rise there? If so, this riding could actually be closer than in your model.

    Eglinton Lawrence - the gap has increased in favour of the Liberals at a time when logic would dictate it decreases?

    Mississauga South - same comment as above

    Richmond Hill - same comment as above

    Simcoe Grey - What happened here? Did a local poll come out? How did Guergis go from first to third?

    Westmount Ville Marie - While this is hard to believe, the Outremont situation was also difficult to believe so I can buy this. I would still say this is TCTC.

  14. Ekos polled another 1000 last night and will drop the first night of yesterday's poll. Don't know if this will continue until e day as a new rolling poll or not. The CBC has a good two words of caution about the NDP surge. Nick Clegg.

  15. Seems as someone writes we can really use a good strong majority CPC government to take care of the Nation's Business.
    I can't see how any of the projections that do not expect a huge swing in the GTA region to go Conservative can in fact be anything to take notice of.
    Demographics can be a considered factor - that and it is getting more difficult to even find a Liberal nowadays.
    Oh and Harper eats babies.

  16. No way or in no shape or fashion do the Conservatives capture Random Burin St.George's. I've seen this go back and forth on this site's projections all election and it won't happen. The only way the Liberals lose this seat is if the national party implodes (And that may happen) and even then I think the Liberals would win here, it's my seat, I know. John Ottenheimer makes it really interesting but Judy Foote still wins it. Also don't know why the CPC will lose Pontiac or are projected to lose the seat. Yes the NDP are up in Quebec but really Pontiac is the seat of a sitting minister, one of the more powerful 1's in the government, can't see him losing. Great work on the site, going to be an interesting election night next Monday.

  17. People are obsessed with the idea that there's a "left" vote. But most voters aren't really ideological. Especially when you get out of the university educated poly-sci crowd and into the working class areas of Canada. They don't particularly care who is left or right.

    They don't think "hmm, I wish the country were more left". They think "I wish someone would so something about NAFTA and immigration". They don't wish the country would go more right. They think "I want them to cut the GST and double my pension".

    If you don't believe that the NDP is competing directly with the Conservatives... then you can't ever explain how traditional CCF territory went blue. Once you get outside Southern Ontario, it's a fight between Conservatives and New Democrats for working class votes.

  18. Sorry I see at least 5 more seats for the Conservatives at these polling numbers.

    Pontiac-Cabinet minister riding likes being in gov
    Vancouver North- guarenteed Conservative
    Esquimault Juan de fuca- never really Liberal
    Guelph-split Catholic vote goes Con swing riding
    Kingston- military vote

  19. And now we see Frank Graves begin to hedge his numbers and release "absolutely certain to vote" standings. Hmm So on a rolling poll, the Conservatives have gone up quite a bit in one day haven't they?

    "The EKOS-iPolitics survey of Canadians, importantly tracked by likelihood to vote, shows the NDP the choice of 28.1 per cent of those who are “absolutely certain” to cast a ballot May 2. The Conservatives have maintained the support of 35.3 per cent of firmly committed voters, while the Liberals have dropped to an unprecedented third spot with 23.7 per cent support of voters who are certain to exercise their democratic right.

    “If we wanted to make a prediction about what’s going to happen on May 2, this is probably our best bet,” said pollster Frank Graves."

  20. EKOs having the Conservatives at 34% is not outside the margins of error at all. Nanos has never had them above 39%, and EKOs numbers fall within both his and Harris Decima. The only outliers thus far have been the absurdly high numbers that IPSOS-REID has been claiming for the Conservatives.

  21. I hope Saanich-Gulf Island progressives who'd otherwise vote Liberal or NDP vote instead for GPC leader Elizabeth May. She's best positioned to oust CPC incumbent Gary Lunn in this riding AND would make Canadian history with the first Green Party seat in Parliament.

  22. Eric: I love your work, but I fear that Frank Graves has just cheapened it a great deal.

  23. @Anonymous (10:25)

    Don't look at just the final outcome, look at how close some of the races are. The projection today has 7 ridings in Ontario that are within 5% of going Conservative. That's 2 short of a majority, and 5% is definitely close enough that it could shift based solely on election day X factors. There are a similar number of Ontario ridings that are within 5% of swinging away from the Conservatives too.

    The pundits being right or wrong about Ontario being key won't be based on changes (or lack of) in the seat count, but on the number of ridings that were close. If the Conservatives fall short by 2-3% of the vote in enough Ontario ridings to not get a majority (or carry enough Ontario ridings by a 2-3% margin to get a majority), you can bet that they'll be wishing they put just that little bit more effort (or thankful that they did) into Ontario.

  24. Ekos is not an outlier.

    In fact, if we take a look at their results from October 7, 2008 (one week before the last election) we can see how close they've actually been in the past.

    Ekos has a good track record.

    Their '08 seat prjections were also pretty close.

  25. 2 points...

    although it might be over-estimating the influence of your site it almost seems that EKOS is playing with you (and other poll amalgamation sites). 2 huge polls on back to back days that dominate the polling releases after only 2 polls per week???

    If Compas were to put out 3 polls of 5000 this week your numbers would show a huge CPC majority.... I wouldn't trust that prediction either.

    2) "A Léger poll of the riding for Smith's campaign has the Liberals at 36 per cent, the Conservatives at 30 per cent and the NDP at 17 per cent. The telephone poll of 500 respondents was conducted after the leaders' debates.

    Read more: http://www.montrealgazette.com/news/decision-canada/game+Saint+Louis/4673660/story.html#ixzz1Ke5SeS6Z

    A 6 point lead in a riding survey of 500 makes your 20 pt lead somewhat suspect.

  26. BT yes one flaw in Eric's methodology is calling all ridings. Seat projections with a probability of seats as well as low and high values with TCTC status are much more realistic. Honestly ridings projected within 2% are just guesses, some within 5% are no better than guess, perhaps worse.

  27. Eric has pointed out that his benchmark for success is 90%. That means 31 ridings can be incorrectly projected. That's a lot of ridings and is the difference between majority and disaster for the Cons as well as Loyal opposition or 4th place for the NDP. So, as much fun as all this is, given that 2/3 of ridings will stay the same no matter what, getting 69 out of the 100 left correct could probably be done by most of us.

    Sorry Eric.

  28. Ekos uses a different methodology (IVR) to collect data than most other polling companies. Not sure what biases this might imply, but it may be a factor in the numbers.

  29. If I lived in Saanich-Gulf Islands and I had a preferential ballot - i would vote NDP first, Liberal second, then Gary Lunn and then put Elizabeth May dead last. That toxic woman has to be kept out of parliament at all cost. The only way I want her to win is if the Tories are elected in 154 seats and SGI would be number 155.

  30. Here's the problem with everyone jumping on the Ekos poll. And that's that Ipsos Reid has had the Tories a lot higher than everyone else. For all of those who want to see Harper out you should take joy in this. IR has had the Tories consistantly in the 40's since before the election. The two polls are essentially canceling each other out. Where the prediction is right now is actually quite accurate when considering where polls are putting the parties right now. The only thing that still seems a bit off is that NDP support is stil a bit low and Liberal support is still a bit high. Tory support is pretty much right where it should be

  31. Could you show the usual bar chart showing the polls you added?

  32. Anonymous 12:15,

    That's not what I said, I said 90% was the benchmark for failure.

  33. It says a lot about the right wing machine that Frank Graves and EKOS are being so dismissed yet Ipsos is never questioned.

  34. Eric

    Given the obvious volatility of the current situation my feeling is that no pollster is going to get it anywhere near right.

    I think we all need to sit back and take a few deep breaths and relax.

    If the current situation lasts through next Monday there is no way anybody can accurately predict what the out come will be.

  35. Giselle: Not quite true. Blair Wilson was a Green Party MP in the 2006-2008 Parliament. However, he was originally elected as a Liberal, so it would be the first time someone was elected running for the Green Party.

  36. Um is there something in between success and failure Eric? I consider those words to be a duality. Benchmark for failure 90%. Benchmark for success 90%. potaeto potahto

    This is an interesting distinction for someone who calls all ridings and has none TCTC. There's no middle ground in your projection.

  37. I agree with Pete B that without Keith Martin as the star Liberal candidate and with Garrison as a star NDP candidate (who outpolled Conservatives in his two prior runs for this seat), I believe Esquimalt-Juan de Fuca is going NDP this time. As for Anonymous, Troy de Souza has not been able to break through, and maybe the Liberal-NDP split will help him come up through the middle (hopefully not), but I do see this as a contest primarily between Garrison and de Souza.
    -- Patrick from Victoria (E-JDF riding)

  38. Eric, any idea what is going on with Nanos today? Thanks in advance.


  39. The final EKOS poll in 2008 underpolled CPC in virtually every province. I don't believe that CPC will get 33% in BC this election any more than I believe the Bloc will get 4 seats in QC (as projected by EKOS).

    Certainly giving EKOS polls a pro rata share of the projection is skewing these numbers on this site.

  40. This last week will be fascinating. Two possible tweaks I would like to see explored after the election are:
    1. Throwing out the high and low results for each party in the various polls at the national level. It seems to me that Ekos is skewing the results as they have the CPC much lower than the other pollsters. While I suspect that the Ekos numbers will come into line as election day nears they appear to be suspect at the current time.
    2. It may be necessary to weight the various age groups based on the likelihood of voting. The last election the CPC numbers were higher than projected and this may in part be due to the high percentage of support amongst seniors. The commitment level of voters may also play a role.

  41. Could someone in this community explain several results ?
    When I go on the ProjectDemocracy website and I click on certain ridings in and around Montreal, the NPD is not second but first. If the turnout is high on the 2nd of May, the NPD will win more than 6 seats in Quebec. There not dividing the vote,they are taking votes not only from the Bloc, They are winning by a small margins, but winning. Moreover, when you compare past election results, with the present bar charts, the NPD will win more than 6 seats.
    This election in Quebec makes me think of provincial elections in which we saw a landslide.

  42. [In fact, if we take a look at their results from October 7, 2008 (one week before the last election) we can see how close they've actually been in the past.]

    1. that's not the last poll from that election. why are you focusing on that one?
    2. Like everyone else in 2008, they underestimated the Conservative vote.

  43. Anonymous 12:47,

    Obviously I'm aiming for 100%. Accuracy between 90% and 100% would be success of varying degrees. DemocraticSpace, I believe, was 91% or 92% in 2008. It isn't a black or white situation. 98% is obviously a lot better than, say, 93%.

    But once you get below 90% you are in incumbency rate territory, and a projection needs to be at least better than that.

    Either way, I'm sure plenty of people will gloat if I'm not 100% accurate.


    No Nanos today because they didn't poll yesterday (Easter).

  44. Given the existence of house effects, is the model affected by the rate at which polling companies release polls? I.e. earlier was the model at the mercy of Nanos's house effect and given Ekos' more recent heavy polling does this pull your model into it's house effect? (Nanos and Ekos being far enough apart consistently that at least one of them must be a little wonky)

  45. @Winston

    1. The reason I chose October 7/08 poll is as a point of comparison for Ekos accuracy. The polls were discussing in this thread, and the October 7/08 poll both fall one week before E-Day. I would have chosen to compare the "last polls" from '08 and '11...but I can't go into the future. A comparison of polls from 7 days before each E-Day will have to suffice.

    2. Yes Ekos did put the Conservative support (in the October 7 08 poll) 3.5 points behind their E-night numbers, but they also put the Liberals 1.5 points behind theirs. The NDP was placed 2 points ahead of its.

    Even if these deviations are taken into account, we still have a similar picture in the Ekos polls released yesterday and today.

  46. A couple quick notes on EKOS:

    Ekos uses recorded voices and a "touch phone" system to do surveys, as opposed to live operators - They are more or less Canada's equivalent to Rasmussen in the United States for you US political junkies.

    If the truth be told, all polling, not just robo-polling like EKOS, faces staggering challenges - declining participation rates, caller ID, cell phones... the list goes on and on....

    Because of this, polling is a lot more "art" than it is "science" - Pollsters make a fairly large number of emperical adjustments in their methodologies, weights, calling strategies, sampling frames etc to get around all these issues. - This is an opinion, but "true" probability sampling that folks like Gallup used to do is, well, basically dead - it just simply is not possible any more.

    EKOS has always shown the Tories +/- 5% lower than the average of the other polls, abd clearly the EKOS IVR polling is catching a different pool of fish than the other pollsters.

    In 2000 Rasmussen in the US looked like he had the world by the tail and his great huge 1000 person a day tracking polls gained a lot of attention.... Except his final result in 2000 (Bush +10.5 versus the reality of Gore +0.5%) was a total disaster... being out by 11% in a sample size of 7000... well... something was just plain wrong structurally....

    To be fair in 2004 Rasmussen adjusted things and was damn near perfect, and was "pretty good" in 2008 as well... (a bit shakey again in 2010) so we know IVR can work, we also know it can blow up really badly....

    I guess we will know on May 2nd if the EKOS IVR system works in Canada, in 2011, in this election..

    The Vorlon

  47. "It says a lot about the right wing machine that Frank Graves and EKOS are being so dismissed yet Ipsos is never questioned."

    It won't matter one way or the other if it's substantiated by results. The thing is that IPSOS does have a significant pro-Tory tilt in its results just as EKOs has an anti-tilt.

    But, (and this is key) IPSOS isn't the one making headlines with their results.

    Maybe that's because people do believe IPSOS's top-line numbers or maybe that's because the media, (like me) have internalized the house effect we expect from them. Ekos however is predicting things like the NDP winning 50 seats in Quebec, which doesn't even seem in line with their own numbers. It's hard to believe...

  48. @ Rob ....

    Actually, Ekos over stated Liberal Support by .2 pts, and NDP support by 1.2 pts. They slightly underpolled the Bloc by .2 pts and over polled the Greens by 2.8 pts. The best pollster of 2008 was Angus Reid - 4.4 total pts across 5 parties

  49. http://greenparty.ca/media-release/2011-04-26/elizabeth-may-leading-saanich-gulf-islands

    An interesting internal poll showing Elizabeth May in the lead in Saanich Gulf Islands

  50. Interesting, but keep in mind that it is an internal poll.

  51. @Rob, the thing is that I'm not seeing a reason there to think Ekos is more accurate then the other pollsters. This election Ekos seems to be the odd man out which renders me suspicious of their data. (I'm also suspicious of results like 43% con in some other polls)

  52. Hate to say this Eric but you have to do something about Anonymous!!

    One person ? OK

    Two people ? Iffy

    Three or more?? Not on !!

    I think I've counted so far FIVE people using the Anonymous name. Please fix. Thanks.

  53. Could someone please enlighten me as to why Frank Graves is so "bad" as some posters note. Has his reputation suffered in the past? Has he been pushing a certain party? Was he misleading in his polls? Or are these just personal opinions?

    BTW, I agree that EKOS projecting 53 QC seats for the NDP seems like SF. :)

  54. Where do people get the idea that Ipsos favours conservatives?

    Final poll in 2008 election: Conservatives 34% (actual 37.65%), Liberals 29% (actual 26.26%), NDP 18% (actual 18.18%), Green 8% (actual 6.78%), and Bloc 9% (actual 9.98%).

  55. Your analysis seems to be at odds with your chart of Seat Projections. eg. You mention Welland as having been 'given' to the NDP, but still show:
    Conservative 33.2% NDP 33.0% Lib 28.0%

    Similarly: Ajax Pickering & others.

    Please explain.

  56. You must be looking at older numbers. Those are not the current projections.

  57. (Where do people get the idea that Ipsos favours conservatives? )

    Look at the polling for 2004 and 2006. In both of those cases Ipsos favored the conservatives. 2008 is a bad election to look at for accuracy information because pretty much everyone was wrong. (I think there was an election day swing due to good news about the economy)

  58. Re: Graves

    To me, its a matter of opinion.

    Like Rasmussen in the US, EKOS polls seem to appear frequently, and are frequently orientated towards recent events; leading or following media interest in subjects. Mr. Graves also then speaks at length in articles about the public and what their responses imply, and what politicians should be trying to do... something about it seems more 24hour cable news than others in terms of trying to fill space with information and opinion. That does not mean the polls are wrong but it doesn't increase confidence. Maybe its just that he gets a lot of attention.

  59. Earlier today (April 26) your Ontario graph listed CPC at 48% on April 24, in agreement with Nanos number. Now your figures have changed retroactively, so that the CPC April 24 number has dropped 5% to 43%, while NDP is up 3% and LPC and Green both higher as well. Similarly in Atlantic Canada, the April 24 results look better for NDP and worse for CPC than they did earlier today. How do polling results change retroactively?

  60. Anonymous 18:45,

    What do you mean, for the Federal Opinion Polling Trends chart?

    That's because for a poll like EKOS, it was added to the daily averages for April 22, April 23, and April 24, the days in which it was in the field.

  61. Eric

    According to a squib on CTV's Power Play there have been already something on the order of 2 million voters at advanced polls !!

    Does this compare with 2008 for example or is this something new ??

  62. Advanced polling has been reported at 35% higher than 2008. However advanced polls have never been on a holiday weekend before especially when two of the polling days were holidays for many people. The weather was also very nice on Sat in most of the country. I wouldn't read any particular party support into these numbers. Liberals can say it's the 800,000 phantom Liberals showing up. The NDP can say its their wave. Conservatives can say their voters are more motivated and more likely to vote early. We don't know what this means, even for voter turnout on e day.


COMMENT MODERATION POLICY - Please be respectful when commenting. If choosing to remain anonymous, please sign your comment with some sort of pseudonym to avoid confusion. Please do not use any derogatory terms for fellow commenters, parties, or politicians. Inflammatory and overly partisan comments will not be posted. PLEASE KEEP DISCUSSION ON TOPIC.